diff --git a/cts/cts-scheduler.in b/cts/cts-scheduler.in index fc9790b671..7ba2415804 100644 --- a/cts/cts-scheduler.in +++ b/cts/cts-scheduler.in @@ -1,1587 +1,1588 @@ #!@PYTHON@ """ Regression tests for Pacemaker's scheduler """ __copyright__ = "Copyright 2004-2021 the Pacemaker project contributors" __license__ = "GNU General Public License version 2 or later (GPLv2+) WITHOUT ANY WARRANTY" import io import os import re import sys import stat import shlex import shutil import argparse import subprocess import platform DESC = """Regression tests for Pacemaker's scheduler""" # Each entry in TESTS is a group of tests, where each test consists of a # test base name, test description, and additional test arguments. # Test groups will be separated by newlines in output. TESTS = [ [ [ "simple1", "Offline" ], [ "simple2", "Start" ], [ "simple3", "Start 2" ], [ "simple4", "Start Failed" ], [ "simple6", "Stop Start" ], [ "simple7", "Shutdown" ], #[ "simple8", "Stonith" ], #[ "simple9", "Lower version" ], #[ "simple10", "Higher version" ], [ "simple11", "Priority (ne)" ], [ "simple12", "Priority (eq)" ], [ "simple8", "Stickiness" ], ], [ [ "group1", "Group" ], [ "group2", "Group + Native" ], [ "group3", "Group + Group" ], [ "group4", "Group + Native (nothing)" ], [ "group5", "Group + Native (move)" ], [ "group6", "Group + Group (move)" ], [ "group7", "Group colocation" ], [ "group13", "Group colocation (cant run)" ], [ "group8", "Group anti-colocation" ], [ "group9", "Group recovery" ], [ "group10", "Group partial recovery" ], [ "group11", "Group target_role" ], [ "group14", "Group stop (graph terminated)" ], [ "group15", "Negative group colocation" ], [ "bug-1573", "Partial stop of a group with two children" ], [ "bug-1718", "Mandatory group ordering - Stop group_FUN" ], [ "bug-lf-2613", "Move group on failure" ], [ "bug-lf-2619", "Move group on clone failure" ], [ "group-fail", "Ensure stop order is preserved for partially active groups" ], [ "group-unmanaged", "No need to restart r115 because r114 is unmanaged" ], [ "group-unmanaged-stopped", "Make sure r115 is stopped when r114 fails" ], [ "group-dependents", "Account for the location preferences of things colocated with a group" ], [ "group-stop-ordering", "Ensure blocked group member stop does not force other member stops" ], ], [ [ "rsc_dep1", "Must not" ], [ "rsc_dep3", "Must" ], [ "rsc_dep5", "Must not 3" ], [ "rsc_dep7", "Must 3" ], [ "rsc_dep10", "Must (but cant)" ], [ "rsc_dep2", "Must (running)" ], [ "rsc_dep8", "Must (running : alt)" ], [ "rsc_dep4", "Must (running + move)" ], [ "asymmetric", "Asymmetric - require explicit location constraints" ], ], [ [ "orphan-0", "Orphan ignore" ], [ "orphan-1", "Orphan stop" ], [ "orphan-2", "Orphan stop, remove failcount" ], ], [ [ "params-0", "Params: No change" ], [ "params-1", "Params: Changed" ], [ "params-2", "Params: Resource definition" ], [ "params-3", "Params: Restart instead of reload if start pending" ], [ "params-4", "Params: Reload" ], [ "params-5", "Params: Restart based on probe digest" ], [ "novell-251689", "Resource definition change + target_role=stopped" ], [ "bug-lf-2106", "Restart all anonymous clone instances after config change" ], [ "params-6", "Params: Detect reload in previously migrated resource" ], [ "nvpair-id-ref", "Support id-ref in nvpair with optional name" ], [ "not-reschedule-unneeded-monitor", "Do not reschedule unneeded monitors while resource definitions have changed" ], [ "reload-becomes-restart", "Cancel reload if restart becomes required" ], ], [ [ "target-0", "Target Role : baseline" ], [ "target-1", "Target Role : promoted" ], [ "target-2", "Target Role : invalid" ], ], [ [ "base-score", "Set a node's default score for all nodes" ], ], [ [ "date-1", "Dates", [ "-t", "2005-020" ] ], [ "date-2", "Date Spec - Pass", [ "-t", "2005-020T12:30" ] ], [ "date-3", "Date Spec - Fail", [ "-t", "2005-020T11:30" ] ], [ "origin", "Timing of recurring operations", [ "-t", "2014-05-07 00:28:00" ] ], [ "probe-0", "Probe (anon clone)" ], [ "probe-1", "Pending Probe" ], [ "probe-2", "Correctly re-probe cloned groups" ], [ "probe-3", "Probe (pending node)" ], [ "probe-4", "Probe (pending node + stopped resource)" ], + [ "probe-pending-node", "Probe (pending node + unmanaged resource)" ], [ "standby", "Standby" ], [ "comments", "Comments" ], ], [ [ "one-or-more-0", "Everything starts" ], [ "one-or-more-1", "Nothing starts because of A" ], [ "one-or-more-2", "D can start because of C" ], [ "one-or-more-3", "D cannot start because of B and C" ], [ "one-or-more-4", "D cannot start because of target-role" ], [ "one-or-more-5", "Start A and F even though C and D are stopped" ], [ "one-or-more-6", "Leave A running even though B is stopped" ], [ "one-or-more-7", "Leave A running even though C is stopped" ], [ "bug-5140-require-all-false", "Allow basegrp:0 to stop" ], [ "clone-require-all-1", "clone B starts node 3 and 4" ], [ "clone-require-all-2", "clone B remains stopped everywhere" ], [ "clone-require-all-3", "clone B stops everywhere because A stops everywhere" ], [ "clone-require-all-4", "clone B remains on node 3 and 4 with only one instance of A remaining" ], [ "clone-require-all-5", "clone B starts on node 1 3 and 4" ], [ "clone-require-all-6", "clone B remains active after shutting down instances of A" ], [ "clone-require-all-7", "clone A and B both start at the same time. all instances of A start before B" ], [ "clone-require-all-no-interleave-1", "C starts everywhere after A and B" ], [ "clone-require-all-no-interleave-2", "C starts on nodes 1, 2, and 4 with only one active instance of B" ], [ "clone-require-all-no-interleave-3", "C remains active when instance of B is stopped on one node and started on another" ], [ "one-or-more-unrunnable-instances", "Avoid dependencies on instances that won't ever be started" ], ], [ [ "location-date-rules-1", "Use location constraints with ineffective date-based rules" ], [ "location-date-rules-2", "Use location constraints with effective date-based rules" ], [ "nvpair-date-rules-1", "Use nvpair blocks with a variety of date-based rules" ], [ "value-source", "Use location constraints with node attribute expressions using value-source" ], [ "rule-dbl-as-auto-number-match", "Floating-point rule values default to number comparison: match" ], [ "rule-dbl-as-auto-number-no-match", "Floating-point rule values default to number comparison: no " "match" ], [ "rule-dbl-as-integer-match", "Floating-point rule values set to integer comparison: match" ], [ "rule-dbl-as-integer-no-match", "Floating-point rule values set to integer comparison: no match" ], [ "rule-dbl-as-number-match", "Floating-point rule values set to number comparison: match" ], [ "rule-dbl-as-number-no-match", "Floating-point rule values set to number comparison: no match" ], [ "rule-dbl-parse-fail-default-str-match", "Floating-point rule values fail to parse, default to string " "comparison: match" ], [ "rule-dbl-parse-fail-default-str-no-match", "Floating-point rule values fail to parse, default to string " "comparison: no match" ], [ "rule-int-as-auto-integer-match", "Integer rule values default to integer comparison: match" ], [ "rule-int-as-auto-integer-no-match", "Integer rule values default to integer comparison: no match" ], [ "rule-int-as-integer-match", "Integer rule values set to integer comparison: match" ], [ "rule-int-as-integer-no-match", "Integer rule values set to integer comparison: no match" ], [ "rule-int-as-number-match", "Integer rule values set to number comparison: match" ], [ "rule-int-as-number-no-match", "Integer rule values set to number comparison: no match" ], [ "rule-int-parse-fail-default-str-match", "Integer rule values fail to parse, default to string " "comparison: match" ], [ "rule-int-parse-fail-default-str-no-match", "Integer rule values fail to parse, default to string " "comparison: no match" ], ], [ [ "order1", "Order start 1" ], [ "order2", "Order start 2" ], [ "order3", "Order stop" ], [ "order4", "Order (multiple)" ], [ "order5", "Order (move)" ], [ "order6", "Order (move w/ restart)" ], [ "order7", "Order (mandatory)" ], [ "order-optional", "Order (score=0)" ], [ "order-required", "Order (score=INFINITY)" ], [ "bug-lf-2171", "Prevent group start when clone is stopped" ], [ "order-clone", "Clone ordering should be able to prevent startup of dependent clones" ], [ "order-sets", "Ordering for resource sets" ], [ "order-serialize", "Serialize resources without inhibiting migration" ], [ "order-serialize-set", "Serialize a set of resources without inhibiting migration" ], [ "clone-order-primitive", "Order clone start after a primitive" ], [ "clone-order-16instances", "Verify ordering of 16 cloned resources" ], [ "order-optional-keyword", "Order (optional keyword)" ], [ "order-mandatory", "Order (mandatory keyword)" ], [ "bug-lf-2493", "Don't imply colocation requirements when applying ordering constraints with clones" ], [ "ordered-set-basic-startup", "Constraint set with default order settings" ], [ "ordered-set-natural", "Allow natural set ordering" ], [ "order-wrong-kind", "Order (error)" ], ], [ [ "coloc-loop", "Colocation - loop" ], [ "coloc-many-one", "Colocation - many-to-one" ], [ "coloc-list", "Colocation - many-to-one with list" ], [ "coloc-group", "Colocation - groups" ], [ "coloc-unpromoted-anti", "Anti-colocation with unpromoted shouldn't prevent promoted colocation" ], [ "coloc-attr", "Colocation based on node attributes" ], [ "coloc-negative-group", "Negative colocation with a group" ], [ "coloc-intra-set", "Intra-set colocation" ], [ "bug-lf-2435", "Colocation sets with a negative score" ], [ "coloc-clone-stays-active", "Ensure clones don't get stopped/demoted because a dependent must stop" ], [ "coloc_fp_logic", "Verify floating point calculations in colocation are working" ], [ "colo_promoted_w_native", "cl#5070 - Verify promotion order is affected when colocating promoted with primitive" ], [ "colo_unpromoted_w_native", "cl#5070 - Verify promotion order is affected when colocating unpromoted with primitive" ], [ "anti-colocation-order", "cl#5187 - Prevent resources in an anti-colocation from even temporarily running on a same node" ], [ "anti-colocation-promoted", "Organize order of actions for promoted resources in anti-colocations" ], [ "anti-colocation-unpromoted", "Organize order of actions for unpromoted resources in anti-colocations" ], [ "enforce-colo1", "Always enforce B with A INFINITY" ], [ "complex_enforce_colo", "Always enforce B with A INFINITY. (make sure heat-engine stops)" ], [ "coloc-dependee-should-stay", "Stickiness outweighs group colocation" ], [ "coloc-dependee-should-move", "Group colocation outweighs stickiness" ], [ "colocation-influence", "Respect colocation influence" ], ], [ [ "rsc-sets-seq-true", "Resource Sets - sequential=false" ], [ "rsc-sets-seq-false", "Resource Sets - sequential=true" ], [ "rsc-sets-clone", "Resource Sets - Clone" ], [ "rsc-sets-promoted", "Resource Sets - Promoted" ], [ "rsc-sets-clone-1", "Resource Sets - Clone (lf#2404)" ], ], [ [ "attrs1", "string: eq (and)" ], [ "attrs2", "string: lt / gt (and)" ], [ "attrs3", "string: ne (or)" ], [ "attrs4", "string: exists" ], [ "attrs5", "string: not_exists" ], [ "attrs6", "is_dc: true" ], [ "attrs7", "is_dc: false" ], [ "attrs8", "score_attribute" ], [ "per-node-attrs", "Per node resource parameters" ], ], [ [ "mon-rsc-1", "Schedule Monitor - start" ], [ "mon-rsc-2", "Schedule Monitor - move" ], [ "mon-rsc-3", "Schedule Monitor - pending start" ], [ "mon-rsc-4", "Schedule Monitor - move/pending start" ], ], [ [ "rec-rsc-0", "Resource Recover - no start" ], [ "rec-rsc-1", "Resource Recover - start" ], [ "rec-rsc-2", "Resource Recover - monitor" ], [ "rec-rsc-3", "Resource Recover - stop - ignore" ], [ "rec-rsc-4", "Resource Recover - stop - block" ], [ "rec-rsc-5", "Resource Recover - stop - fence" ], [ "rec-rsc-6", "Resource Recover - multiple - restart" ], [ "rec-rsc-7", "Resource Recover - multiple - stop" ], [ "rec-rsc-8", "Resource Recover - multiple - block" ], [ "rec-rsc-9", "Resource Recover - group/group" ], [ "monitor-recovery", "on-fail=block + resource recovery detected by recurring monitor" ], [ "stop-failure-no-quorum", "Stop failure without quorum" ], [ "stop-failure-no-fencing", "Stop failure without fencing available" ], [ "stop-failure-with-fencing", "Stop failure with fencing available" ], [ "multiple-active-block-group", "Support of multiple-active=block for resource groups" ], [ "multiple-monitor-one-failed", "Consider resource failed if any of the configured monitor operations failed" ], ], [ [ "quorum-1", "No quorum - ignore" ], [ "quorum-2", "No quorum - freeze" ], [ "quorum-3", "No quorum - stop" ], [ "quorum-4", "No quorum - start anyway" ], [ "quorum-5", "No quorum - start anyway (group)" ], [ "quorum-6", "No quorum - start anyway (clone)" ], [ "bug-cl-5212", "No promotion with no-quorum-policy=freeze" ], [ "suicide-needed-inquorate", "no-quorum-policy=suicide: suicide necessary" ], [ "suicide-not-needed-initial-quorum", "no-quorum-policy=suicide: suicide not necessary at initial quorum" ], [ "suicide-not-needed-never-quorate", "no-quorum-policy=suicide: suicide not necessary if never quorate" ], [ "suicide-not-needed-quorate", "no-quorum-policy=suicide: suicide necessary if quorate" ], ], [ [ "rec-node-1", "Node Recover - Startup - no fence" ], [ "rec-node-2", "Node Recover - Startup - fence" ], [ "rec-node-3", "Node Recover - HA down - no fence" ], [ "rec-node-4", "Node Recover - HA down - fence" ], [ "rec-node-5", "Node Recover - CRM down - no fence" ], [ "rec-node-6", "Node Recover - CRM down - fence" ], [ "rec-node-7", "Node Recover - no quorum - ignore" ], [ "rec-node-8", "Node Recover - no quorum - freeze" ], [ "rec-node-9", "Node Recover - no quorum - stop" ], [ "rec-node-10", "Node Recover - no quorum - stop w/fence" ], [ "rec-node-11", "Node Recover - CRM down w/ group - fence" ], [ "rec-node-12", "Node Recover - nothing active - fence" ], [ "rec-node-13", "Node Recover - failed resource + shutdown - fence" ], [ "rec-node-15", "Node Recover - unknown lrm section" ], [ "rec-node-14", "Serialize all stonith's" ], ], [ [ "multi1", "Multiple Active (stop/start)" ], ], [ [ "migrate-begin", "Normal migration" ], [ "migrate-success", "Completed migration" ], [ "migrate-partial-1", "Completed migration, missing stop on source" ], [ "migrate-partial-2", "Successful migrate_to only" ], [ "migrate-partial-3", "Successful migrate_to only, target down" ], [ "migrate-partial-4", "Migrate from the correct host after migrate_to+migrate_from" ], [ "bug-5186-partial-migrate", "Handle partial migration when src node loses membership" ], [ "migrate-fail-2", "Failed migrate_from" ], [ "migrate-fail-3", "Failed migrate_from + stop on source" ], [ "migrate-fail-4", "Failed migrate_from + stop on target - ideally we wouldn't need to re-stop on target" ], [ "migrate-fail-5", "Failed migrate_from + stop on source and target" ], [ "migrate-fail-6", "Failed migrate_to" ], [ "migrate-fail-7", "Failed migrate_to + stop on source" ], [ "migrate-fail-8", "Failed migrate_to + stop on target - ideally we wouldn't need to re-stop on target" ], [ "migrate-fail-9", "Failed migrate_to + stop on source and target" ], [ "migration-ping-pong", "Old migrate_to failure + successful migrate_from on same node" ], [ "migrate-stop", "Migration in a stopping stack" ], [ "migrate-start", "Migration in a starting stack" ], [ "migrate-stop_start", "Migration in a restarting stack" ], [ "migrate-stop-complex", "Migration in a complex stopping stack" ], [ "migrate-start-complex", "Migration in a complex starting stack" ], [ "migrate-stop-start-complex", "Migration in a complex moving stack" ], [ "migrate-shutdown", "Order the post-migration 'stop' before node shutdown" ], [ "migrate-1", "Migrate (migrate)" ], [ "migrate-2", "Migrate (stable)" ], [ "migrate-3", "Migrate (failed migrate_to)" ], [ "migrate-4", "Migrate (failed migrate_from)" ], [ "novell-252693", "Migration in a stopping stack" ], [ "novell-252693-2", "Migration in a starting stack" ], [ "novell-252693-3", "Non-Migration in a starting and stopping stack" ], [ "bug-1820", "Migration in a group" ], [ "bug-1820-1", "Non-migration in a group" ], [ "migrate-5", "Primitive migration with a clone" ], [ "migrate-fencing", "Migration after Fencing" ], [ "migrate-both-vms", "Migrate two VMs that have no colocation" ], [ "migration-behind-migrating-remote", "Migrate resource behind migrating remote connection" ], [ "1-a-then-bm-move-b", "Advanced migrate logic. A then B. migrate B" ], [ "2-am-then-b-move-a", "Advanced migrate logic, A then B, migrate A without stopping B" ], [ "3-am-then-bm-both-migrate", "Advanced migrate logic. A then B. migrate both" ], [ "4-am-then-bm-b-not-migratable", "Advanced migrate logic, A then B, B not migratable" ], [ "5-am-then-bm-a-not-migratable", "Advanced migrate logic. A then B. move both, a not migratable" ], [ "6-migrate-group", "Advanced migrate logic, migrate a group" ], [ "7-migrate-group-one-unmigratable", "Advanced migrate logic, migrate group mixed with allow-migrate true/false" ], [ "8-am-then-bm-a-migrating-b-stopping", "Advanced migrate logic, A then B, A migrating, B stopping" ], [ "9-am-then-bm-b-migrating-a-stopping", "Advanced migrate logic, A then B, B migrate, A stopping" ], [ "10-a-then-bm-b-move-a-clone", "Advanced migrate logic, A clone then B, migrate B while stopping A" ], [ "11-a-then-bm-b-move-a-clone-starting", "Advanced migrate logic, A clone then B, B moving while A is start/stopping" ], [ "a-promote-then-b-migrate", "A promote then B start. migrate B" ], [ "a-demote-then-b-migrate", "A demote then B stop. migrate B" ], # @TODO: If pacemaker implements versioned attributes, uncomment this test #[ "migrate-versioned", "Disable migration for versioned resources" ], [ "bug-lf-2422", "Dependency on partially active group - stop ocfs:*" ], ], [ [ "clone-anon-probe-1", "Probe the correct (anonymous) clone instance for each node" ], [ "clone-anon-probe-2", "Avoid needless re-probing of anonymous clones" ], [ "clone-anon-failcount", "Merge failcounts for anonymous clones" ], [ "force-anon-clone-max", "Update clone-max properly when forcing a clone to be anonymous" ], [ "anon-instance-pending", "Assign anonymous clone instance numbers properly when action pending" ], [ "inc0", "Incarnation start" ], [ "inc1", "Incarnation start order" ], [ "inc2", "Incarnation silent restart, stop, move" ], [ "inc3", "Inter-incarnation ordering, silent restart, stop, move" ], [ "inc4", "Inter-incarnation ordering, silent restart, stop, move (ordered)" ], [ "inc5", "Inter-incarnation ordering, silent restart, stop, move (restart 1)" ], [ "inc6", "Inter-incarnation ordering, silent restart, stop, move (restart 2)" ], [ "inc7", "Clone colocation" ], [ "inc8", "Clone anti-colocation" ], [ "inc9", "Non-unique clone" ], [ "inc10", "Non-unique clone (stop)" ], [ "inc11", "Primitive colocation with clones" ], [ "inc12", "Clone shutdown" ], [ "cloned-group", "Make sure only the correct number of cloned groups are started" ], [ "cloned-group-stop", "Ensure stopping qpidd also stops glance and cinder" ], [ "clone-no-shuffle", "Don't prioritize allocation of instances that must be moved" ], [ "clone-max-zero", "Orphan processing with clone-max=0" ], [ "clone-anon-dup", "Bug LF#2087 - Correctly parse the state of anonymous clones that are active more than once per node" ], [ "bug-lf-2160", "Don't shuffle clones due to colocation" ], [ "bug-lf-2213", "clone-node-max enforcement for cloned groups" ], [ "bug-lf-2153", "Clone ordering constraints" ], [ "bug-lf-2361", "Ensure clones observe mandatory ordering constraints if the LHS is unrunnable" ], [ "bug-lf-2317", "Avoid needless restart of primitive depending on a clone" ], [ "clone-colocate-instance-1", "Colocation with a specific clone instance (negative example)" ], [ "clone-colocate-instance-2", "Colocation with a specific clone instance" ], [ "clone-order-instance", "Ordering with specific clone instances" ], [ "bug-lf-2453", "Enforce mandatory clone ordering without colocation" ], [ "bug-lf-2508", "Correctly reconstruct the status of anonymous cloned groups" ], [ "bug-lf-2544", "Balanced clone placement" ], [ "bug-lf-2445", "Redistribute clones with node-max > 1 and stickiness = 0" ], [ "bug-lf-2574", "Avoid clone shuffle" ], [ "bug-lf-2581", "Avoid group restart due to unrelated clone (re)start" ], [ "bug-cl-5168", "Don't shuffle clones" ], [ "bug-cl-5170", "Prevent clone from starting with on-fail=block" ], [ "clone-fail-block-colocation", "Move colocated group when failed clone has on-fail=block" ], [ "clone-interleave-1", "Clone-3 cannot start on pcmk-1 due to interleaved ordering (no colocation)" ], [ "clone-interleave-2", "Clone-3 must stop on pcmk-1 due to interleaved ordering (no colocation)" ], [ "clone-interleave-3", "Clone-3 must be recovered on pcmk-1 due to interleaved ordering (no colocation)" ], [ "rebalance-unique-clones", "Rebalance unique clone instances with no stickiness" ], [ "clone-requires-quorum-recovery", "Clone with requires=quorum on failed node needing recovery" ], [ "clone-requires-quorum", "Clone with requires=quorum with presumed-inactive instance on failed node" ], ], [ [ "cloned_start_one", "order first clone then clone... first clone_min=2" ], [ "cloned_start_two", "order first clone then clone... first clone_min=2" ], [ "cloned_stop_one", "order first clone then clone... first clone_min=2" ], [ "cloned_stop_two", "order first clone then clone... first clone_min=2" ], [ "clone_min_interleave_start_one", "order first clone then clone... first clone_min=2 and then has interleave=true" ], [ "clone_min_interleave_start_two", "order first clone then clone... first clone_min=2 and then has interleave=true" ], [ "clone_min_interleave_stop_one", "order first clone then clone... first clone_min=2 and then has interleave=true" ], [ "clone_min_interleave_stop_two", "order first clone then clone... first clone_min=2 and then has interleave=true" ], [ "clone_min_start_one", "order first clone then primitive... first clone_min=2" ], [ "clone_min_start_two", "order first clone then primitive... first clone_min=2" ], [ "clone_min_stop_all", "order first clone then primitive... first clone_min=2" ], [ "clone_min_stop_one", "order first clone then primitive... first clone_min=2" ], [ "clone_min_stop_two", "order first clone then primitive... first clone_min=2" ], ], [ [ "unfence-startup", "Clean unfencing" ], [ "unfence-definition", "Unfencing when the agent changes" ], [ "unfence-parameters", "Unfencing when the agent parameters changes" ], [ "unfence-device", "Unfencing when a cluster has only fence devices" ], ], [ [ "promoted-0", "Stopped -> Unpromoted" ], [ "promoted-1", "Stopped -> Promote" ], [ "promoted-2", "Stopped -> Promote : notify" ], [ "promoted-3", "Stopped -> Promote : promoted location" ], [ "promoted-4", "Started -> Promote : promoted location" ], [ "promoted-5", "Promoted -> Promoted" ], [ "promoted-6", "Promoted -> Promoted (2)" ], [ "promoted-7", "Promoted -> Fenced" ], [ "promoted-8", "Promoted -> Fenced -> Moved" ], [ "promoted-9", "Stopped + Promotable + No quorum" ], [ "promoted-10", "Stopped -> Promotable : notify with monitor" ], [ "promoted-11", "Stopped -> Promote : colocation" ], [ "novell-239082", "Demote/Promote ordering" ], [ "novell-239087", "Stable promoted placement" ], [ "promoted-12", "Promotion based solely on rsc_location constraints" ], [ "promoted-13", "Include preferences of colocated resources when placing promoted" ], [ "promoted-demote", "Ordering when actions depends on demoting an unpromoted resource" ], [ "promoted-ordering", "Prevent resources from starting that need a promoted" ], [ "bug-1765", "Verify promoted-with-promoted colocation does not stop unpromoted instances" ], [ "promoted-group", "Promotion of cloned groups" ], [ "bug-lf-1852", "Don't shuffle promotable instances unnecessarily" ], [ "promoted-failed-demote", "Don't retry failed demote actions" ], [ "promoted-failed-demote-2", "Don't retry failed demote actions (notify=false)" ], [ "promoted-depend", "Ensure resources that depend on promoted instance don't get allocated until that does" ], [ "promoted-reattach", "Re-attach to a running promoted" ], [ "promoted-allow-start", "Don't include promoted score if it would prevent allocation" ], [ "promoted-colocation", "Allow promoted instances placemaker to be influenced by colocation constraints" ], [ "promoted-pseudo", "Make sure promote/demote pseudo actions are created correctly" ], [ "promoted-role", "Prevent target-role from promoting more than promoted-max instances" ], [ "bug-lf-2358", "Anti-colocation of promoted instances" ], [ "promoted-promotion-constraint", "Mandatory promoted colocation constraints" ], [ "unmanaged-promoted", "Ensure role is preserved for unmanaged resources" ], [ "promoted-unmanaged-monitor", "Start correct monitor for unmanaged promoted instances" ], [ "promoted-demote-2", "Demote does not clear past failure" ], [ "promoted-move", "Move promoted based on failure of colocated group" ], [ "promoted-probed-score", "Observe the promotion score of probed resources" ], [ "colocation_constraint_stops_promoted", "cl#5054 - Ensure promoted is demoted when stopped by colocation constraint" ], [ "colocation_constraint_stops_unpromoted", "cl#5054 - Ensure unpromoted is not demoted when stopped by colocation constraint" ], [ "order_constraint_stops_promoted", "cl#5054 - Ensure promoted is demoted when stopped by order constraint" ], [ "order_constraint_stops_unpromoted", "cl#5054 - Ensure unpromoted is not demoted when stopped by order constraint" ], [ "promoted_monitor_restart", "cl#5072 - Ensure promoted monitor operation will start after promotion" ], [ "bug-rh-880249", "Handle replacement of an m/s resource with a primitive" ], [ "bug-5143-ms-shuffle", "Prevent promoted instance shuffling due to promotion score" ], [ "promoted-demote-block", "Block promotion if demote fails with on-fail=block" ], [ "promoted-dependent-ban", "Don't stop instances from being active because a dependent is banned from that host" ], [ "promoted-stop", "Stop instances due to location constraint with role=Started" ], [ "promoted-partially-demoted-group", "Allow partially demoted group to finish demoting" ], [ "bug-cl-5213", "Ensure role colocation with -INFINITY is enforced" ], [ "bug-cl-5219", "Allow unrelated resources with a common colocation target to remain promoted" ], [ "promoted-asymmetrical-order", "Fix the behaviors of multi-state resources with asymmetrical ordering" ], [ "promoted-notify", "Promotion with notifications" ], [ "promoted-score-startup", "Use permanent promoted scores without LRM history" ], [ "failed-demote-recovery", "Recover resource in unpromoted role after demote fails" ], [ "failed-demote-recovery-promoted", "Recover resource in promoted role after demote fails" ], [ "on_fail_demote1", "Recovery with on-fail=\"demote\" on healthy cluster, remote, guest, and bundle nodes" ], [ "on_fail_demote2", "Recovery with on-fail=\"demote\" with promotion on different node" ], [ "on_fail_demote3", "Recovery with on-fail=\"demote\" with no promotion" ], [ "on_fail_demote4", "Recovery with on-fail=\"demote\" on failed cluster, remote, guest, and bundle nodes" ], [ "no_quorum_demote", "Promotable demotion and primitive stop with no-quorum-policy=\"demote\"" ], [ "no-promote-on-unrunnable-guest", "Don't select bundle instance for promotion when container can't run" ], ], [ [ "history-1", "Correctly parse stateful-1 resource state" ], ], [ [ "managed-0", "Managed (reference)" ], [ "managed-1", "Not managed - down" ], [ "managed-2", "Not managed - up" ], [ "bug-5028", "Shutdown should block if anything depends on an unmanaged resource" ], [ "bug-5028-detach", "Ensure detach still works" ], [ "bug-5028-bottom", "Ensure shutdown still blocks if the blocked resource is at the bottom of the stack" ], [ "unmanaged-stop-1", "cl#5155 - Block the stop of resources if any depending resource is unmanaged" ], [ "unmanaged-stop-2", "cl#5155 - Block the stop of resources if the first resource in a mandatory stop order is unmanaged" ], [ "unmanaged-stop-3", "cl#5155 - Block the stop of resources if any depending resource in a group is unmanaged" ], [ "unmanaged-stop-4", "cl#5155 - Block the stop of resources if any depending resource in the middle of a group is unmanaged" ], [ "unmanaged-block-restart", "Block restart of resources if any dependent resource in a group is unmanaged" ], ], [ [ "interleave-0", "Interleave (reference)" ], [ "interleave-1", "coloc - not interleaved" ], [ "interleave-2", "coloc - interleaved" ], [ "interleave-3", "coloc - interleaved (2)" ], [ "interleave-pseudo-stop", "Interleaved clone during stonith" ], [ "interleave-stop", "Interleaved clone during stop" ], [ "interleave-restart", "Interleaved clone during dependency restart" ], ], [ [ "notify-0", "Notify reference" ], [ "notify-1", "Notify simple" ], [ "notify-2", "Notify simple, confirm" ], [ "notify-3", "Notify move, confirm" ], [ "novell-239079", "Notification priority" ], #[ "notify-2", "Notify - 764" ], [ "notifs-for-unrunnable", "Don't schedule notifications for an unrunnable action" ], [ "route-remote-notify", "Route remote notify actions through correct cluster node" ], [ "notify-behind-stopping-remote", "Don't schedule notifications behind stopped remote" ], ], [ [ "594", "OSDL #594 - Unrunnable actions scheduled in transition" ], [ "662", "OSDL #662 - Two resources start on one node when incarnation_node_max = 1" ], [ "696", "OSDL #696 - CRM starts stonith RA without monitor" ], [ "726", "OSDL #726 - Attempting to schedule rsc_posic041_monitor_5000 _after_ a stop" ], [ "735", "OSDL #735 - Correctly detect that rsc_hadev1 is stopped on hadev3" ], [ "764", "OSDL #764 - Missing monitor op for DoFencing:child_DoFencing:1" ], [ "797", "OSDL #797 - Assert triggered: task_id_i > max_call_id" ], [ "829", "OSDL #829" ], [ "994", "OSDL #994 - Stopping the last resource in a resource group causes the entire group to be restarted" ], [ "994-2", "OSDL #994 - with a dependent resource" ], [ "1360", "OSDL #1360 - Clone stickiness" ], [ "1484", "OSDL #1484 - on_fail=stop" ], [ "1494", "OSDL #1494 - Clone stability" ], [ "unrunnable-1", "Unrunnable" ], [ "unrunnable-2", "Unrunnable 2" ], [ "stonith-0", "Stonith loop - 1" ], [ "stonith-1", "Stonith loop - 2" ], [ "stonith-2", "Stonith loop - 3" ], [ "stonith-3", "Stonith startup" ], [ "stonith-4", "Stonith node state" ], [ "dc-fence-ordering", "DC needs fencing while other nodes are shutting down" ], [ "bug-1572-1", "Recovery of groups depending on promotable role" ], [ "bug-1572-2", "Recovery of groups depending on promotable role when promoted is not re-promoted" ], [ "bug-1685", "Depends-on-promoted ordering" ], [ "bug-1822", "Don't promote partially active groups" ], [ "bug-pm-11", "New resource added to a m/s group" ], [ "bug-pm-12", "Recover only the failed portion of a cloned group" ], [ "bug-n-387749", "Don't shuffle clone instances" ], [ "bug-n-385265", "Don't ignore the failure stickiness of group children - resource_idvscommon should stay stopped" ], [ "bug-n-385265-2", "Ensure groups are migrated instead of remaining partially active on the current node" ], [ "bug-lf-1920", "Correctly handle probes that find active resources" ], [ "bnc-515172", "Location constraint with multiple expressions" ], [ "colocate-primitive-with-clone", "Optional colocation with a clone" ], [ "use-after-free-merge", "Use-after-free in native_merge_weights" ], [ "bug-lf-2551", "STONITH ordering for stop" ], [ "bug-lf-2606", "Stonith implies demote" ], [ "bug-lf-2474", "Ensure resource op timeout takes precedence over op_defaults" ], [ "bug-suse-707150", "Prevent vm-01 from starting due to colocation/ordering" ], [ "bug-5014-A-start-B-start", "Verify when A starts B starts using symmetrical=false" ], [ "bug-5014-A-stop-B-started", "Verify when A stops B does not stop if it has already started using symmetric=false" ], [ "bug-5014-A-stopped-B-stopped", "Verify when A is stopped and B has not started, B does not start before A using symmetric=false" ], [ "bug-5014-CthenAthenB-C-stopped", "Verify when C then A is symmetrical=true, A then B is symmetric=false, and C is stopped that nothing starts" ], [ "bug-5014-CLONE-A-start-B-start", "Verify when A starts B starts using clone resources with symmetric=false" ], [ "bug-5014-CLONE-A-stop-B-started", "Verify when A stops B does not stop if it has already started using clone resources with symmetric=false" ], [ "bug-5014-GROUP-A-start-B-start", "Verify when A starts B starts when using group resources with symmetric=false" ], [ "bug-5014-GROUP-A-stopped-B-started", "Verify when A stops B does not stop if it has already started using group resources with symmetric=false" ], [ "bug-5014-GROUP-A-stopped-B-stopped", "Verify when A is stopped and B has not started, B does not start before A using group resources with symmetric=false" ], [ "bug-5014-ordered-set-symmetrical-false", "Verify ordered sets work with symmetrical=false" ], [ "bug-5014-ordered-set-symmetrical-true", "Verify ordered sets work with symmetrical=true" ], [ "clbz5007-promotable-colocation", "Verify use of colocation scores other than INFINITY and -INFINITY work on multi-state resources" ], [ "bug-5038", "Prevent restart of anonymous clones when clone-max decreases" ], [ "bug-5025-1", "Automatically clean up failcount after resource config change with reload" ], [ "bug-5025-2", "Make sure clear failcount action isn't set when config does not change" ], [ "bug-5025-3", "Automatically clean up failcount after resource config change with restart" ], [ "bug-5025-4", "Clear failcount when last failure is a start op and rsc attributes changed" ], [ "failcount", "Ensure failcounts are correctly expired" ], [ "failcount-block", "Ensure failcounts are not expired when on-fail=block is present" ], [ "per-op-failcount", "Ensure per-operation failcount is handled and not passed to fence agent" ], [ "on-fail-ignore", "Ensure on-fail=ignore works even beyond migration-threshold" ], [ "monitor-onfail-restart", "bug-5058 - Monitor failure with on-fail set to restart" ], [ "monitor-onfail-stop", "bug-5058 - Monitor failure wiht on-fail set to stop" ], [ "bug-5059", "No need to restart p_stateful1:*" ], [ "bug-5069-op-enabled", "Test on-fail=ignore with failure when monitor is enabled" ], [ "bug-5069-op-disabled", "Test on-fail-ignore with failure when monitor is disabled" ], [ "obsolete-lrm-resource", "cl#5115 - Do not use obsolete lrm_resource sections" ], [ "expire-non-blocked-failure", "Ignore failure-timeout only if the failed operation has on-fail=block" ], [ "asymmetrical-order-move", "Respect asymmetrical ordering when trying to move resources" ], [ "asymmetrical-order-restart", "Respect asymmetrical ordering when restarting dependent resource" ], [ "start-then-stop-with-unfence", "Avoid graph loop with start-then-stop constraint plus unfencing" ], [ "order-expired-failure", "Order failcount cleanup after remote fencing" ], [ "ignore_stonith_rsc_order1", "cl#5056- Ignore order constraint between stonith and non-stonith rsc" ], [ "ignore_stonith_rsc_order2", "cl#5056- Ignore order constraint with group rsc containing mixed stonith and non-stonith" ], [ "ignore_stonith_rsc_order3", "cl#5056- Ignore order constraint, stonith clone and mixed group" ], [ "ignore_stonith_rsc_order4", "cl#5056- Ignore order constraint, stonith clone and clone with nested mixed group" ], [ "honor_stonith_rsc_order1", "cl#5056- Honor order constraint, stonith clone and pure stonith group(single rsc)" ], [ "honor_stonith_rsc_order2", "cl#5056- Honor order constraint, stonith clone and pure stonith group(multiple rsc)" ], [ "honor_stonith_rsc_order3", "cl#5056- Honor order constraint, stonith clones with nested pure stonith group" ], [ "honor_stonith_rsc_order4", "cl#5056- Honor order constraint, between two native stonith rscs" ], [ "multiply-active-stonith", "Multiply active stonith" ], [ "probe-timeout", "cl#5099 - Default probe timeout" ], [ "order-first-probes", "cl#5301 - respect order constraints when relevant resources are being probed" ], [ "concurrent-fencing", "Allow performing fencing operations in parallel" ], [ "priority-fencing-delay", "Delay fencing targeting the more significant node" ], ], [ [ "systemhealth1", "System Health () #1" ], [ "systemhealth2", "System Health () #2" ], [ "systemhealth3", "System Health () #3" ], [ "systemhealthn1", "System Health (None) #1" ], [ "systemhealthn2", "System Health (None) #2" ], [ "systemhealthn3", "System Health (None) #3" ], [ "systemhealthm1", "System Health (Migrate On Red) #1" ], [ "systemhealthm2", "System Health (Migrate On Red) #2" ], [ "systemhealthm3", "System Health (Migrate On Red) #3" ], [ "systemhealtho1", "System Health (Only Green) #1" ], [ "systemhealtho2", "System Health (Only Green) #2" ], [ "systemhealtho3", "System Health (Only Green) #3" ], [ "systemhealthp1", "System Health (Progessive) #1" ], [ "systemhealthp2", "System Health (Progessive) #2" ], [ "systemhealthp3", "System Health (Progessive) #3" ], ], [ [ "utilization", "Placement Strategy - utilization" ], [ "minimal", "Placement Strategy - minimal" ], [ "balanced", "Placement Strategy - balanced" ], ], [ [ "placement-stickiness", "Optimized Placement Strategy - stickiness" ], [ "placement-priority", "Optimized Placement Strategy - priority" ], [ "placement-location", "Optimized Placement Strategy - location" ], [ "placement-capacity", "Optimized Placement Strategy - capacity" ], ], [ [ "utilization-order1", "Utilization Order - Simple" ], [ "utilization-order2", "Utilization Order - Complex" ], [ "utilization-order3", "Utilization Order - Migrate" ], [ "utilization-order4", "Utilization Order - Live Migration (bnc#695440)" ], [ "utilization-shuffle", "Don't displace prmExPostgreSQLDB2 on act2, Start prmExPostgreSQLDB1 on act3" ], [ "load-stopped-loop", "Avoid transition loop due to load_stopped (cl#5044)" ], [ "load-stopped-loop-2", "cl#5235 - Prevent graph loops that can be introduced by load_stopped -> migrate_to ordering" ], ], [ [ "colocated-utilization-primitive-1", "Colocated Utilization - Primitive" ], [ "colocated-utilization-primitive-2", "Colocated Utilization - Choose the most capable node" ], [ "colocated-utilization-group", "Colocated Utilization - Group" ], [ "colocated-utilization-clone", "Colocated Utilization - Clone" ], [ "utilization-check-allowed-nodes", "Only check the capacities of the nodes that can run the resource" ], ], [ [ "reprobe-target_rc", "Ensure correct target_rc for reprobe of inactive resources" ], [ "node-maintenance-1", "cl#5128 - Node maintenance" ], [ "node-maintenance-2", "cl#5128 - Node maintenance (coming out of maintenance mode)" ], [ "shutdown-maintenance-node", "Do not fence a maintenance node if it shuts down cleanly" ], [ "rsc-maintenance", "Per-resource maintenance" ], ], [ [ "not-installed-agent", "The resource agent is missing" ], [ "not-installed-tools", "Something the resource agent needs is missing" ], ], [ [ "stopped-monitor-00", "Stopped Monitor - initial start" ], [ "stopped-monitor-01", "Stopped Monitor - failed started" ], [ "stopped-monitor-02", "Stopped Monitor - started multi-up" ], [ "stopped-monitor-03", "Stopped Monitor - stop started" ], [ "stopped-monitor-04", "Stopped Monitor - failed stop" ], [ "stopped-monitor-05", "Stopped Monitor - start unmanaged" ], [ "stopped-monitor-06", "Stopped Monitor - unmanaged multi-up" ], [ "stopped-monitor-07", "Stopped Monitor - start unmanaged multi-up" ], [ "stopped-monitor-08", "Stopped Monitor - migrate" ], [ "stopped-monitor-09", "Stopped Monitor - unmanage started" ], [ "stopped-monitor-10", "Stopped Monitor - unmanaged started multi-up" ], [ "stopped-monitor-11", "Stopped Monitor - stop unmanaged started" ], [ "stopped-monitor-12", "Stopped Monitor - unmanaged started multi-up (target-role=Stopped)" ], [ "stopped-monitor-20", "Stopped Monitor - initial stop" ], [ "stopped-monitor-21", "Stopped Monitor - stopped single-up" ], [ "stopped-monitor-22", "Stopped Monitor - stopped multi-up" ], [ "stopped-monitor-23", "Stopped Monitor - start stopped" ], [ "stopped-monitor-24", "Stopped Monitor - unmanage stopped" ], [ "stopped-monitor-25", "Stopped Monitor - unmanaged stopped multi-up" ], [ "stopped-monitor-26", "Stopped Monitor - start unmanaged stopped" ], [ "stopped-monitor-27", "Stopped Monitor - unmanaged stopped multi-up (target-role=Started)" ], [ "stopped-monitor-30", "Stopped Monitor - new node started" ], [ "stopped-monitor-31", "Stopped Monitor - new node stopped" ], ], [ # This is a combo test to check: # - probe timeout defaults to the minimum-interval monitor's # - duplicate recurring operations are ignored # - if timeout spec is bad, the default timeout is used # - failure is blocked with on-fail=block even if ISO8601 interval is specified # - started/stopped role monitors are started/stopped on right nodes [ "intervals", "Recurring monitor interval handling" ], ], [ [ "ticket-primitive-1", "Ticket - Primitive (loss-policy=stop, initial)" ], [ "ticket-primitive-2", "Ticket - Primitive (loss-policy=stop, granted)" ], [ "ticket-primitive-3", "Ticket - Primitive (loss-policy-stop, revoked)" ], [ "ticket-primitive-4", "Ticket - Primitive (loss-policy=demote, initial)" ], [ "ticket-primitive-5", "Ticket - Primitive (loss-policy=demote, granted)" ], [ "ticket-primitive-6", "Ticket - Primitive (loss-policy=demote, revoked)" ], [ "ticket-primitive-7", "Ticket - Primitive (loss-policy=fence, initial)" ], [ "ticket-primitive-8", "Ticket - Primitive (loss-policy=fence, granted)" ], [ "ticket-primitive-9", "Ticket - Primitive (loss-policy=fence, revoked)" ], [ "ticket-primitive-10", "Ticket - Primitive (loss-policy=freeze, initial)" ], [ "ticket-primitive-11", "Ticket - Primitive (loss-policy=freeze, granted)" ], [ "ticket-primitive-12", "Ticket - Primitive (loss-policy=freeze, revoked)" ], [ "ticket-primitive-13", "Ticket - Primitive (loss-policy=stop, standby, granted)" ], [ "ticket-primitive-14", "Ticket - Primitive (loss-policy=stop, granted, standby)" ], [ "ticket-primitive-15", "Ticket - Primitive (loss-policy=stop, standby, revoked)" ], [ "ticket-primitive-16", "Ticket - Primitive (loss-policy=demote, standby, granted)" ], [ "ticket-primitive-17", "Ticket - Primitive (loss-policy=demote, granted, standby)" ], [ "ticket-primitive-18", "Ticket - Primitive (loss-policy=demote, standby, revoked)" ], [ "ticket-primitive-19", "Ticket - Primitive (loss-policy=fence, standby, granted)" ], [ "ticket-primitive-20", "Ticket - Primitive (loss-policy=fence, granted, standby)" ], [ "ticket-primitive-21", "Ticket - Primitive (loss-policy=fence, standby, revoked)" ], [ "ticket-primitive-22", "Ticket - Primitive (loss-policy=freeze, standby, granted)" ], [ "ticket-primitive-23", "Ticket - Primitive (loss-policy=freeze, granted, standby)" ], [ "ticket-primitive-24", "Ticket - Primitive (loss-policy=freeze, standby, revoked)" ], ], [ [ "ticket-group-1", "Ticket - Group (loss-policy=stop, initial)" ], [ "ticket-group-2", "Ticket - Group (loss-policy=stop, granted)" ], [ "ticket-group-3", "Ticket - Group (loss-policy-stop, revoked)" ], [ "ticket-group-4", "Ticket - Group (loss-policy=demote, initial)" ], [ "ticket-group-5", "Ticket - Group (loss-policy=demote, granted)" ], [ "ticket-group-6", "Ticket - Group (loss-policy=demote, revoked)" ], [ "ticket-group-7", "Ticket - Group (loss-policy=fence, initial)" ], [ "ticket-group-8", "Ticket - Group (loss-policy=fence, granted)" ], [ "ticket-group-9", "Ticket - Group (loss-policy=fence, revoked)" ], [ "ticket-group-10", "Ticket - Group (loss-policy=freeze, initial)" ], [ "ticket-group-11", "Ticket - Group (loss-policy=freeze, granted)" ], [ "ticket-group-12", "Ticket - Group (loss-policy=freeze, revoked)" ], [ "ticket-group-13", "Ticket - Group (loss-policy=stop, standby, granted)" ], [ "ticket-group-14", "Ticket - Group (loss-policy=stop, granted, standby)" ], [ "ticket-group-15", "Ticket - Group (loss-policy=stop, standby, revoked)" ], [ "ticket-group-16", "Ticket - Group (loss-policy=demote, standby, granted)" ], [ "ticket-group-17", "Ticket - Group (loss-policy=demote, granted, standby)" ], [ "ticket-group-18", "Ticket - Group (loss-policy=demote, standby, revoked)" ], [ "ticket-group-19", "Ticket - Group (loss-policy=fence, standby, granted)" ], [ "ticket-group-20", "Ticket - Group (loss-policy=fence, granted, standby)" ], [ "ticket-group-21", "Ticket - Group (loss-policy=fence, standby, revoked)" ], [ "ticket-group-22", "Ticket - Group (loss-policy=freeze, standby, granted)" ], [ "ticket-group-23", "Ticket - Group (loss-policy=freeze, granted, standby)" ], [ "ticket-group-24", "Ticket - Group (loss-policy=freeze, standby, revoked)" ], ], [ [ "ticket-clone-1", "Ticket - Clone (loss-policy=stop, initial)" ], [ "ticket-clone-2", "Ticket - Clone (loss-policy=stop, granted)" ], [ "ticket-clone-3", "Ticket - Clone (loss-policy-stop, revoked)" ], [ "ticket-clone-4", "Ticket - Clone (loss-policy=demote, initial)" ], [ "ticket-clone-5", "Ticket - Clone (loss-policy=demote, granted)" ], [ "ticket-clone-6", "Ticket - Clone (loss-policy=demote, revoked)" ], [ "ticket-clone-7", "Ticket - Clone (loss-policy=fence, initial)" ], [ "ticket-clone-8", "Ticket - Clone (loss-policy=fence, granted)" ], [ "ticket-clone-9", "Ticket - Clone (loss-policy=fence, revoked)" ], [ "ticket-clone-10", "Ticket - Clone (loss-policy=freeze, initial)" ], [ "ticket-clone-11", "Ticket - Clone (loss-policy=freeze, granted)" ], [ "ticket-clone-12", "Ticket - Clone (loss-policy=freeze, revoked)" ], [ "ticket-clone-13", "Ticket - Clone (loss-policy=stop, standby, granted)" ], [ "ticket-clone-14", "Ticket - Clone (loss-policy=stop, granted, standby)" ], [ "ticket-clone-15", "Ticket - Clone (loss-policy=stop, standby, revoked)" ], [ "ticket-clone-16", "Ticket - Clone (loss-policy=demote, standby, granted)" ], [ "ticket-clone-17", "Ticket - Clone (loss-policy=demote, granted, standby)" ], [ "ticket-clone-18", "Ticket - Clone (loss-policy=demote, standby, revoked)" ], [ "ticket-clone-19", "Ticket - Clone (loss-policy=fence, standby, granted)" ], [ "ticket-clone-20", "Ticket - Clone (loss-policy=fence, granted, standby)" ], [ "ticket-clone-21", "Ticket - Clone (loss-policy=fence, standby, revoked)" ], [ "ticket-clone-22", "Ticket - Clone (loss-policy=freeze, standby, granted)" ], [ "ticket-clone-23", "Ticket - Clone (loss-policy=freeze, granted, standby)" ], [ "ticket-clone-24", "Ticket - Clone (loss-policy=freeze, standby, revoked)" ], ], [ [ "ticket-promoted-1", "Ticket - Promoted (loss-policy=stop, initial)" ], [ "ticket-promoted-2", "Ticket - Promoted (loss-policy=stop, granted)" ], [ "ticket-promoted-3", "Ticket - Promoted (loss-policy-stop, revoked)" ], [ "ticket-promoted-4", "Ticket - Promoted (loss-policy=demote, initial)" ], [ "ticket-promoted-5", "Ticket - Promoted (loss-policy=demote, granted)" ], [ "ticket-promoted-6", "Ticket - Promoted (loss-policy=demote, revoked)" ], [ "ticket-promoted-7", "Ticket - Promoted (loss-policy=fence, initial)" ], [ "ticket-promoted-8", "Ticket - Promoted (loss-policy=fence, granted)" ], [ "ticket-promoted-9", "Ticket - Promoted (loss-policy=fence, revoked)" ], [ "ticket-promoted-10", "Ticket - Promoted (loss-policy=freeze, initial)" ], [ "ticket-promoted-11", "Ticket - Promoted (loss-policy=freeze, granted)" ], [ "ticket-promoted-12", "Ticket - Promoted (loss-policy=freeze, revoked)" ], [ "ticket-promoted-13", "Ticket - Promoted (loss-policy=stop, standby, granted)" ], [ "ticket-promoted-14", "Ticket - Promoted (loss-policy=stop, granted, standby)" ], [ "ticket-promoted-15", "Ticket - Promoted (loss-policy=stop, standby, revoked)" ], [ "ticket-promoted-16", "Ticket - Promoted (loss-policy=demote, standby, granted)" ], [ "ticket-promoted-17", "Ticket - Promoted (loss-policy=demote, granted, standby)" ], [ "ticket-promoted-18", "Ticket - Promoted (loss-policy=demote, standby, revoked)" ], [ "ticket-promoted-19", "Ticket - Promoted (loss-policy=fence, standby, granted)" ], [ "ticket-promoted-20", "Ticket - Promoted (loss-policy=fence, granted, standby)" ], [ "ticket-promoted-21", "Ticket - Promoted (loss-policy=fence, standby, revoked)" ], [ "ticket-promoted-22", "Ticket - Promoted (loss-policy=freeze, standby, granted)" ], [ "ticket-promoted-23", "Ticket - Promoted (loss-policy=freeze, granted, standby)" ], [ "ticket-promoted-24", "Ticket - Promoted (loss-policy=freeze, standby, revoked)" ], ], [ [ "ticket-rsc-sets-1", "Ticket - Resource sets (1 ticket, initial)" ], [ "ticket-rsc-sets-2", "Ticket - Resource sets (1 ticket, granted)" ], [ "ticket-rsc-sets-3", "Ticket - Resource sets (1 ticket, revoked)" ], [ "ticket-rsc-sets-4", "Ticket - Resource sets (2 tickets, initial)" ], [ "ticket-rsc-sets-5", "Ticket - Resource sets (2 tickets, granted)" ], [ "ticket-rsc-sets-6", "Ticket - Resource sets (2 tickets, granted)" ], [ "ticket-rsc-sets-7", "Ticket - Resource sets (2 tickets, revoked)" ], [ "ticket-rsc-sets-8", "Ticket - Resource sets (1 ticket, standby, granted)" ], [ "ticket-rsc-sets-9", "Ticket - Resource sets (1 ticket, granted, standby)" ], [ "ticket-rsc-sets-10", "Ticket - Resource sets (1 ticket, standby, revoked)" ], [ "ticket-rsc-sets-11", "Ticket - Resource sets (2 tickets, standby, granted)" ], [ "ticket-rsc-sets-12", "Ticket - Resource sets (2 tickets, standby, granted)" ], [ "ticket-rsc-sets-13", "Ticket - Resource sets (2 tickets, granted, standby)" ], [ "ticket-rsc-sets-14", "Ticket - Resource sets (2 tickets, standby, revoked)" ], [ "cluster-specific-params", "Cluster-specific instance attributes based on rules" ], [ "site-specific-params", "Site-specific instance attributes based on rules" ], ], [ [ "template-1", "Template - 1" ], [ "template-2", "Template - 2" ], [ "template-3", "Template - 3 (merge operations)" ], [ "template-coloc-1", "Template - Colocation 1" ], [ "template-coloc-2", "Template - Colocation 2" ], [ "template-coloc-3", "Template - Colocation 3" ], [ "template-order-1", "Template - Order 1" ], [ "template-order-2", "Template - Order 2" ], [ "template-order-3", "Template - Order 3" ], [ "template-ticket", "Template - Ticket" ], [ "template-rsc-sets-1", "Template - Resource Sets 1" ], [ "template-rsc-sets-2", "Template - Resource Sets 2" ], [ "template-rsc-sets-3", "Template - Resource Sets 3" ], [ "template-rsc-sets-4", "Template - Resource Sets 4" ], [ "template-clone-primitive", "Cloned primitive from template" ], [ "template-clone-group", "Cloned group from template" ], [ "location-sets-templates", "Resource sets and templates - Location" ], [ "tags-coloc-order-1", "Tags - Colocation and Order (Simple)" ], [ "tags-coloc-order-2", "Tags - Colocation and Order (Resource Sets with Templates)" ], [ "tags-location", "Tags - Location" ], [ "tags-ticket", "Tags - Ticket" ], ], [ [ "container-1", "Container - initial" ], [ "container-2", "Container - monitor failed" ], [ "container-3", "Container - stop failed" ], [ "container-4", "Container - reached migration-threshold" ], [ "container-group-1", "Container in group - initial" ], [ "container-group-2", "Container in group - monitor failed" ], [ "container-group-3", "Container in group - stop failed" ], [ "container-group-4", "Container in group - reached migration-threshold" ], [ "container-is-remote-node", "Place resource within container when container is remote-node" ], [ "bug-rh-1097457", "Kill user defined container/contents ordering" ], [ "bug-cl-5247", "Graph loop when recovering m/s resource in a container" ], [ "bundle-order-startup", "Bundle startup ordering" ], [ "bundle-order-partial-start", "Bundle startup ordering when some dependencies are already running" ], [ "bundle-order-partial-start-2", "Bundle startup ordering when some dependencies and the container are already running" ], [ "bundle-order-stop", "Bundle stop ordering" ], [ "bundle-order-partial-stop", "Bundle startup ordering when some dependencies are already stopped" ], [ "bundle-order-stop-on-remote", "Stop nested resource after bringing up the connection" ], [ "bundle-order-startup-clone", "Prevent startup because bundle isn't promoted" ], [ "bundle-order-startup-clone-2", "Bundle startup with clones" ], [ "bundle-order-stop-clone", "Stop bundle because clone is stopping" ], [ "bundle-nested-colocation", "Colocation of nested connection resources" ], [ "bundle-order-fencing", "Order pseudo bundle fencing after parent node fencing if both are happening" ], [ "bundle-probe-order-1", "order 1" ], [ "bundle-probe-order-2", "order 2" ], [ "bundle-probe-order-3", "order 3" ], [ "bundle-probe-remotes", "Ensure remotes get probed too" ], [ "bundle-replicas-change", "Change bundle from 1 replica to multiple" ], [ "nested-remote-recovery", "Recover bundle's container hosted on remote node" ], ], [ [ "whitebox-fail1", "Fail whitebox container rsc" ], [ "whitebox-fail2", "Fail cluster connection to guest node" ], [ "whitebox-fail3", "Failed containers should not run nested on remote nodes" ], [ "whitebox-start", "Start whitebox container with resources assigned to it" ], [ "whitebox-stop", "Stop whitebox container with resources assigned to it" ], [ "whitebox-move", "Move whitebox container with resources assigned to it" ], [ "whitebox-asymmetric", "Verify connection rsc opts-in based on container resource" ], [ "whitebox-ms-ordering", "Verify promote/demote can not occur before connection is established" ], [ "whitebox-ms-ordering-move", "Stop/Start cycle within a moving container" ], [ "whitebox-orphaned", "Properly shutdown orphaned whitebox container" ], [ "whitebox-orphan-ms", "Properly tear down orphan ms resources on remote-nodes" ], [ "whitebox-unexpectedly-running", "Recover container nodes the cluster did not start" ], [ "whitebox-migrate1", "Migrate both container and connection resource" ], [ "whitebox-imply-stop-on-fence", "imply stop action on container node rsc when host node is fenced" ], [ "whitebox-nested-group", "Verify guest remote-node works nested in a group" ], [ "guest-node-host-dies", "Verify guest node is recovered if host goes away" ], [ "guest-node-cleanup", "Order guest node connection recovery after container probe" ], [ "guest-host-not-fenceable", "Actions on guest node are unrunnable if host is unclean and cannot be fenced" ], ], [ [ "remote-startup-probes", "Baremetal remote-node startup probes" ], [ "remote-startup", "Startup a newly discovered remote-nodes with no status" ], [ "remote-fence-unclean", "Fence unclean baremetal remote-node" ], [ "remote-fence-unclean2", "Fence baremetal remote-node after cluster node fails and connection can not be recovered" ], [ "remote-fence-unclean-3", "Probe failed remote nodes (triggers fencing)" ], [ "remote-move", "Move remote-node connection resource" ], [ "remote-disable", "Disable a baremetal remote-node" ], [ "remote-probe-disable", "Probe then stop a baremetal remote-node" ], [ "remote-orphaned", "Properly shutdown orphaned connection resource" ], [ "remote-orphaned2", "verify we can handle orphaned remote connections with active resources on the remote" ], [ "remote-recover", "Recover connection resource after cluster-node fails" ], [ "remote-stale-node-entry", "Make sure we properly handle leftover remote-node entries in the node section" ], [ "remote-partial-migrate", "Make sure partial migrations are handled before ops on the remote node" ], [ "remote-partial-migrate2", "Make sure partial migration target is prefered for remote connection" ], [ "remote-recover-fail", "Make sure start failure causes fencing if rsc are active on remote" ], [ "remote-start-fail", "Make sure a start failure does not result in fencing if no active resources are on remote" ], [ "remote-unclean2", "Make monitor failure always results in fencing, even if no rsc are active on remote" ], [ "remote-fence-before-reconnect", "Fence before clearing recurring monitor failure" ], [ "remote-recovery", "Recover remote connections before attempting demotion" ], [ "remote-recover-connection", "Optimistically recovery of only the connection" ], [ "remote-recover-all", "Fencing when the connection has no home" ], [ "remote-recover-no-resources", "Fencing when the connection has no home and no active resources" ], [ "remote-recover-unknown", "Fencing when the connection has no home and the remote has no operation history" ], [ "remote-reconnect-delay", "Waiting for remote reconnect interval to expire" ], [ "remote-connection-unrecoverable", "Remote connection host must be fenced, with connection unrecoverable" ], [ "cancel-behind-moving-remote", "Route recurring monitor cancellations through original node of a moving remote connection" ], ], [ [ "resource-discovery", "Exercises resource-discovery location constraint option" ], [ "rsc-discovery-per-node", "Disable resource discovery per node" ], [ "shutdown-lock", "Ensure shutdown lock works properly" ], [ "shutdown-lock-expiration", "Ensure shutdown lock expiration works properly" ], ], [ [ "op-defaults", "Test op_defaults conditional expressions" ], [ "op-defaults-2", "Test op_defaults AND'ed conditional expressions" ], [ "op-defaults-3", "Test op_defaults precedence" ], [ "rsc-defaults", "Test rsc_defaults conditional expressions" ], [ "rsc-defaults-2", "Test rsc_defaults conditional expressions without type" ], ], [ [ "stop-all-resources", "Test stop-all-resources=true "], ], [ [ "ocf_degraded-remap-ocf_ok", "Test degraded remapped to OK" ], [ "ocf_degraded_promoted-remap-ocf_ok", "Test degraded promoted remapped to OK"], ], # @TODO: If pacemaker implements versioned attributes, uncomment these tests #[ # [ "versioned-resources", "Start resources with #ra-version rules" ], # [ "restart-versioned", "Restart resources on #ra-version change" ], # [ "reload-versioned", "Reload resources on #ra-version change" ], #], #[ # [ "versioned-operations-1", "Use #ra-version to configure operations of native resources" ], # [ "versioned-operations-2", "Use #ra-version to configure operations of stonith resources" ], # [ "versioned-operations-3", "Use #ra-version to configure operations of promotable resources" ], # [ "versioned-operations-4", "Use #ra-version to configure operations of groups of the resources" ], #], ] TESTS_64BIT = [ [ [ "year-2038", "Check handling of timestamps beyond 2038-01-19 03:14:08 UTC" ], ], ] # Constants substituted in the build process class BuildVars(object): SBINDIR = "@sbindir@" BUILDDIR = "@abs_top_builddir@" CRM_SCHEMA_DIRECTORY = "@CRM_SCHEMA_DIRECTORY@" # These values must be kept in sync with crm_exit_t class CrmExit(object): OK = 0 ERROR = 1 NOT_INSTALLED = 5 NOINPUT = 66 CANTCREAT = 73 def is_executable(path): """ Check whether a file at a given path is executable. """ try: return os.stat(path)[stat.ST_MODE] & stat.S_IXUSR except OSError: return False def diff(file1, file2, **kwargs): """ Call diff on two files """ return subprocess.call([ "diff", "-u", "-N", "--ignore-all-space", "--ignore-blank-lines", file1, file2 ], **kwargs) def sort_file(filename): """ Sort a file alphabetically """ with io.open(filename, "rt") as f: lines = sorted(f) with io.open(filename, "wt") as f: f.writelines(lines) def remove_files(filenames): """ Remove a list of files """ for filename in filenames: try: os.remove(filename) except OSError: pass def normalize(filename): """ Remove text from a file that isn't important for comparison """ if not hasattr(normalize, "patterns"): normalize.patterns = [ re.compile(r'crm_feature_set="[^"]*"'), re.compile(r'batch-limit="[0-9]*"') ] if os.path.isfile(filename): with io.open(filename, "rt") as f: lines = f.readlines() with io.open(filename, "wt") as f: for line in lines: for pattern in normalize.patterns: line = pattern.sub("", line) f.write(line) def cat(filename, dest=sys.stdout): """ Copy a file to a destination file descriptor """ with io.open(filename, "rt") as f: shutil.copyfileobj(f, dest) class CtsScheduler(object): """ Regression tests for Pacemaker's scheduler """ def _parse_args(self, argv): """ Parse command-line arguments """ parser = argparse.ArgumentParser(description=DESC) parser.add_argument('-V', '--verbose', action='count', help='Display any differences from expected output') parser.add_argument('--run', metavar='TEST', help=('Run only single specified test (any further ' 'arguments will be passed to crm_simulate)')) parser.add_argument('--update', action='store_true', help='Update expected results with actual results') parser.add_argument('-b', '--binary', metavar='PATH', help='Specify path to crm_simulate') parser.add_argument('-i', '--io-dir', metavar='PATH', help='Specify path to regression test data directory') parser.add_argument('-o', '--out-dir', metavar='PATH', help='Specify where intermediate and output files should go') parser.add_argument('-v', '--valgrind', action='store_true', help='Run all commands under valgrind') parser.add_argument('--valgrind-dhat', action='store_true', help='Run all commands under valgrind with heap analyzer') parser.add_argument('--valgrind-skip-output', action='store_true', help='If running under valgrind, do not display output') parser.add_argument('--testcmd-options', metavar='OPTIONS', default='', help='Additional options for command under test') # argparse can't handle "everything after --run TEST", so grab that self.single_test_args = [] narg = 0 for arg in argv: narg = narg + 1 if arg == '--run': (argv, self.single_test_args) = (argv[:narg+1], argv[narg+1:]) break self.args = parser.parse_args(argv[1:]) def _error(self, s): print(" * ERROR: %s" % s) def _failed(self, s): print(" * FAILED: %s" % s) def _get_valgrind_cmd(self): """ Return command arguments needed (or not) to run valgrind """ if self.args.valgrind: os.environ['G_SLICE'] = "always-malloc" return [ "valgrind", "-q", "--gen-suppressions=all", "--time-stamp=yes", "--trace-children=no", "--show-reachable=no", "--leak-check=full", "--num-callers=20", "--suppressions=%s/valgrind-pcmk.suppressions" % (self.test_home) ] if self.args.valgrind_dhat: os.environ['G_SLICE'] = "always-malloc" return [ "valgrind", "--tool=exp-dhat", "--time-stamp=yes", "--trace-children=no", "--show-top-n=100", "--num-callers=4" ] return [] def _get_simulator_cmd(self): """ Locate the simulation binary """ if self.args.binary is None: self.args.binary = BuildVars.BUILDDIR + "/tools/crm_simulate" if not is_executable(self.args.binary): self.args.binary = BuildVars.SBINDIR + "/crm_simulate" if not is_executable(self.args.binary): # @TODO it would be more pythonic to raise an exception self._error("Test binary " + self.args.binary + " not found") sys.exit(CrmExit.NOT_INSTALLED) return [ self.args.binary ] + shlex.split(self.args.testcmd_options) def set_schema_env(self): """ Ensure schema directory environment variable is set, if possible """ try: return os.environ['PCMK_schema_directory'] except KeyError: for d in [ os.path.join(BuildVars.BUILDDIR, "xml"), BuildVars.CRM_SCHEMA_DIRECTORY ]: if os.path.isdir(d): os.environ['PCMK_schema_directory'] = d return d return None def __init__(self, argv=sys.argv): self._parse_args(argv) # Where this executable lives self.test_home = os.path.dirname(os.path.realpath(argv[0])) # Where test data resides if self.args.io_dir is None: self.args.io_dir = os.path.join(self.test_home, "scheduler") self.xml_input_dir = os.path.join(self.args.io_dir, "xml") self.expected_dir = os.path.join(self.args.io_dir, "exp") self.dot_expected_dir = os.path.join(self.args.io_dir, "dot") self.scores_dir = os.path.join(self.args.io_dir, "scores") self.summary_dir = os.path.join(self.args.io_dir, "summary") self.stderr_expected_dir = os.path.join(self.args.io_dir, "stderr") # Where to store generated files if self.args.out_dir is None: self.args.out_dir = self.args.io_dir self.failed_filename = os.path.join(self.test_home, ".regression.failed.diff") else: self.failed_filename = os.path.join(self.args.out_dir, ".regression.failed.diff") os.environ['CIB_shadow_dir'] = self.args.out_dir self.failed_file = None self.outfile_out_dir = os.path.join(self.args.out_dir, "out") self.dot_out_dir = os.path.join(self.args.out_dir, "dot") self.scores_out_dir = os.path.join(self.args.out_dir, "scores") self.summary_out_dir = os.path.join(self.args.out_dir, "summary") self.stderr_out_dir = os.path.join(self.args.out_dir, "stderr") self.valgrind_out_dir = os.path.join(self.args.out_dir, "valgrind") # Single test mode (if requested) try: # User can give test base name or file name of a test input self.args.run = os.path.splitext(os.path.basename(self.args.run))[0] except (AttributeError, TypeError): pass # --run was not specified self.set_schema_env() # Arguments needed (or not) to run commands self.valgrind_args = self._get_valgrind_cmd() self.simulate_args = self._get_simulator_cmd() # Test counters self.num_failed = 0 self.num_tests = 0 # Ensure that the main output directory exists # We don't want to create it with os.makedirs below if not os.path.isdir(self.args.out_dir): self._error("Output directory missing; can't create output files") sys.exit(CrmExit.CANTCREAT) # Create output subdirectories if they don't exist try: os.makedirs(self.outfile_out_dir, 0o755, True) os.makedirs(self.dot_out_dir, 0o755, True) os.makedirs(self.scores_out_dir, 0o755, True) os.makedirs(self.summary_out_dir, 0o755, True) os.makedirs(self.stderr_out_dir, 0o755, True) if self.valgrind_args: os.makedirs(self.valgrind_out_dir, 0o755, True) except OSError as ex: self._error("Unable to create output subdirectory: %s" % ex) remove_files([ self.outfile_out_dir, self.dot_out_dir, self.scores_out_dir, self.summary_out_dir, self.stderr_out_dir, ]) sys.exit(CrmExit.CANTCREAT) def _compare_files(self, filename1, filename2): """ Add any file differences to failed results """ if diff(filename1, filename2, stdout=subprocess.DEVNULL) != 0: diff(filename1, filename2, stdout=self.failed_file, stderr=subprocess.DEVNULL) self.failed_file.write("\n"); return True return False def run_one(self, test_name, test_desc, test_args=[]): """ Run one scheduler test """ print(" Test %-25s %s" % ((test_name + ":"), test_desc)) did_fail = False self.num_tests = self.num_tests + 1 # Test inputs input_filename = os.path.join( self.xml_input_dir, "%s.xml" % test_name) expected_filename = os.path.join( self.expected_dir, "%s.exp" % test_name) dot_expected_filename = os.path.join( self.dot_expected_dir, "%s.dot" % test_name) scores_filename = os.path.join( self.scores_dir, "%s.scores" % test_name) summary_filename = os.path.join( self.summary_dir, "%s.summary" % test_name) stderr_expected_filename = os.path.join( self.stderr_expected_dir, "%s.stderr" % test_name) # (Intermediate) test outputs output_filename = os.path.join( self.outfile_out_dir, "%s.out" % test_name) dot_output_filename = os.path.join( self.dot_out_dir, "%s.dot.pe" % test_name) score_output_filename = os.path.join( self.scores_out_dir, "%s.scores.pe" % test_name) summary_output_filename = os.path.join( self.summary_out_dir, "%s.summary.pe" % test_name) stderr_output_filename = os.path.join( self.stderr_out_dir, "%s.stderr.pe" % test_name) valgrind_output_filename = os.path.join( self.valgrind_out_dir, "%s.valgrind" % test_name) # Common arguments for running test test_cmd = [] if self.valgrind_args: test_cmd = self.valgrind_args + [ "--log-file=%s" % valgrind_output_filename ] test_cmd = test_cmd + self.simulate_args # @TODO It would be more pythonic to raise exceptions for errors, # then perhaps it would be nice to make a single-test class # Ensure necessary test inputs exist if not os.path.isfile(input_filename): self._error("No input") self.num_failed = self.num_failed + 1 return CrmExit.NOINPUT if not self.args.update and not os.path.isfile(expected_filename): self._error("no stored output") return CrmExit.NOINPUT # Run simulation to generate summary output if self.args.run: # Single test mode test_cmd_full = test_cmd + [ '-x', input_filename, '-S' ] + test_args print(" ".join(test_cmd_full)) else: # @TODO Why isn't test_args added here? test_cmd_full = test_cmd + [ '-x', input_filename, '-S' ] with io.open(summary_output_filename, "wt") as f: simulation = subprocess.Popen(test_cmd_full, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, env=os.environ) # This makes diff happy regardless of --enable-compat-2.0. # Use sed -E to make Linux and BSD special characters more compatible. sed = subprocess.Popen(["sed", "-E", "-e", "s/ocf::/ocf:/g", "-e", r"s/Masters:/Promoted:/", "-e", r"s/Slaves:/Unpromoted:/", "-e", r"s/ Master( |\[|$)/ Promoted\1/", "-e", r"s/ Slave / Unpromoted /", ], stdin=simulation.stdout, stdout=f, stderr=subprocess.STDOUT) simulation.stdout.close() sed.communicate() if self.args.run: cat(summary_output_filename) # Re-run simulation to generate dot, graph, and scores test_cmd_full = test_cmd + [ '-x', input_filename, '-D', dot_output_filename, '-G', output_filename, '-sSQ' ] + test_args with io.open(stderr_output_filename, "wt") as f_stderr, \ io.open(score_output_filename, "wt") as f_score: rc = subprocess.call(test_cmd_full, stdout=f_score, stderr=f_stderr, env=os.environ) # Check for test command failure if rc != CrmExit.OK: self._failed("Test returned: %d" % rc) did_fail = True print(" ".join(test_cmd_full)) # Check for valgrind errors if self.valgrind_args and not self.args.valgrind_skip_output: if os.stat(valgrind_output_filename).st_size > 0: self._failed("Valgrind reported errors") did_fail = True cat(valgrind_output_filename) remove_files([ valgrind_output_filename ]) # Check for core dump if os.path.isfile("core"): self._failed("Core-file detected: core." + test_name) did_fail = True os.rename("core", "%s/core.%s" % (self.test_home, test_name)) # Check any stderr output if os.path.isfile(stderr_expected_filename): if self._compare_files(stderr_expected_filename, stderr_output_filename): self._failed("stderr changed") did_fail = True elif os.stat(stderr_output_filename).st_size > 0: self._failed("Output was written to stderr") did_fail = True cat(stderr_output_filename) remove_files([ stderr_output_filename ]) # Check whether output graph exists, and normalize it if (not os.path.isfile(output_filename) or os.stat(output_filename).st_size == 0): self._error("No graph produced") did_fail = True self.num_failed = self.num_failed + 1 remove_files([ output_filename ]) return CrmExit.ERROR normalize(output_filename) # Check whether dot output exists, and sort it if (not os.path.isfile(dot_output_filename) or os.stat(dot_output_filename).st_size == 0): self._error("No dot-file summary produced") did_fail = True self.num_failed = self.num_failed + 1 remove_files([ dot_output_filename, output_filename ]) return CrmExit.ERROR with io.open(dot_output_filename, "rt") as f: first_line = f.readline() # "digraph" line with opening brace lines = f.readlines() last_line = lines[-1] # closing brace del lines[-1] lines = sorted(set(lines)) # unique sort with io.open(dot_output_filename, "wt") as f: f.write(first_line) f.writelines(lines) f.write(last_line) # Check whether score output exists, and sort it if (not os.path.isfile(score_output_filename) or os.stat(score_output_filename).st_size == 0): self._error("No allocation scores produced") did_fail = True self.num_failed = self.num_failed + 1 remove_files([ score_output_filename, output_filename ]) return CrmExit.ERROR else: sort_file(score_output_filename) if self.args.update: shutil.copyfile(output_filename, expected_filename) shutil.copyfile(dot_output_filename, dot_expected_filename) shutil.copyfile(score_output_filename, scores_filename) shutil.copyfile(summary_output_filename, summary_filename) print(" Updated expected outputs") if self._compare_files(summary_filename, summary_output_filename): self._failed("summary changed") did_fail = True if self._compare_files(dot_expected_filename, dot_output_filename): self._failed("dot-file summary changed") did_fail = True else: remove_files([ dot_output_filename ]) if self._compare_files(expected_filename, output_filename): self._failed("xml-file changed") did_fail = True if self._compare_files(scores_filename, score_output_filename): self._failed("scores-file changed") did_fail = True remove_files([ output_filename, score_output_filename, summary_output_filename]) if did_fail: self.num_failed = self.num_failed + 1 return CrmExit.ERROR return CrmExit.OK def run_all(self): """ Run all defined tests """ if platform.architecture()[0] == "64bit": TESTS.extend(TESTS_64BIT) for group in TESTS: for test in group: try: args = test[2] except IndexError: args = [] self.run_one(test[0], test[1], args) print() def _print_summary(self): """ Print a summary of parameters for this test run """ print("Test home is:\t" + self.test_home) print("Test binary is:\t" + self.args.binary) if 'PCMK_schema_directory' in os.environ: print("Schema home is:\t" + os.environ['PCMK_schema_directory']) if self.valgrind_args != []: print("Activating memory testing with valgrind") print() def _test_results(self): if self.num_failed == 0: return CrmExit.OK if os.path.isfile(self.failed_filename) and os.stat(self.failed_filename).st_size != 0: if self.args.verbose: self._error("Results of %d failed tests (out of %d):" % (self.num_failed, self.num_tests)) cat(self.failed_filename) else: self._error("Results of %d failed tests (out of %d) are in %s" % (self.num_failed, self.num_tests, self.failed_filename)) self._error("Use -V to display them after running the tests") else: self._error("%d (of %d) tests failed (no diff results)" % (self.num_failed, self.num_tests)) if os.path.isfile(self.failed_filename): os.remove(self.failed_filename) return CrmExit.ERROR def run(self): """ Run test(s) as specified """ self._print_summary() # Zero out the error log self.failed_file = io.open(self.failed_filename, "wt") if self.args.run is None: print("Performing the following tests from " + self.args.io_dir) print() self.run_all() print() self.failed_file.close() rc = self._test_results() else: rc = self.run_one(self.args.run, "Single shot", self.single_test_args) self.failed_file.close() cat(self.failed_filename) return rc if __name__ == "__main__": sys.exit(CtsScheduler().run()) # vim: set filetype=python expandtab tabstop=4 softtabstop=4 shiftwidth=4 textwidth=120: diff --git a/cts/scheduler/dot/bundle-probe-order-2.dot b/cts/scheduler/dot/bundle-probe-order-2.dot index 0cce3fd989..770619546a 100644 --- a/cts/scheduler/dot/bundle-probe-order-2.dot +++ b/cts/scheduler/dot/bundle-probe-order-2.dot @@ -1,6 +1,9 @@ digraph "g" { +"galera-bundle-0_monitor_30000 centos2" [ style=bold color="green" fontcolor="black"] +"galera-bundle-docker-0_monitor_60000 centos2" [ style=bold color="green" fontcolor="black"] "galera-bundle-docker-1_monitor_0 centos2" [ style=bold color="green" fontcolor="black"] "galera-bundle-docker-2_monitor_0 centos1" [ style=bold color="green" fontcolor="black"] "galera-bundle-docker-2_monitor_0 centos2" [ style=bold color="green" fontcolor="black"] "galera-bundle-docker-2_monitor_0 centos3" [ style=bold color="green" fontcolor="black"] +"galera:0_monitor_0 galera-bundle-0" [ style=bold color="green" fontcolor="black"] } diff --git a/cts/scheduler/dot/bundle-probe-order-3.dot b/cts/scheduler/dot/bundle-probe-order-3.dot index a4b109fa2c..53a384b71c 100644 --- a/cts/scheduler/dot/bundle-probe-order-3.dot +++ b/cts/scheduler/dot/bundle-probe-order-3.dot @@ -1,9 +1,10 @@ digraph "g" { "galera-bundle-0_monitor_0 centos1" [ style=bold color="green" fontcolor="black"] "galera-bundle-0_monitor_0 centos2" [ style=bold color="green" fontcolor="black"] "galera-bundle-0_monitor_0 centos3" [ style=bold color="green" fontcolor="black"] +"galera-bundle-docker-0_monitor_60000 centos2" [ style=bold color="green" fontcolor="black"] "galera-bundle-docker-1_monitor_0 centos2" [ style=bold color="green" fontcolor="black"] "galera-bundle-docker-2_monitor_0 centos1" [ style=bold color="green" fontcolor="black"] "galera-bundle-docker-2_monitor_0 centos2" [ style=bold color="green" fontcolor="black"] "galera-bundle-docker-2_monitor_0 centos3" [ style=bold color="green" fontcolor="black"] } diff --git a/cts/scheduler/dot/probe-pending-node.dot b/cts/scheduler/dot/probe-pending-node.dot new file mode 100644 index 0000000000..d8f1c9f22b --- /dev/null +++ b/cts/scheduler/dot/probe-pending-node.dot @@ -0,0 +1,2 @@ + digraph "g" { +} diff --git a/cts/scheduler/exp/bundle-probe-order-2.exp b/cts/scheduler/exp/bundle-probe-order-2.exp index d6174e74df..5b280508da 100644 --- a/cts/scheduler/exp/bundle-probe-order-2.exp +++ b/cts/scheduler/exp/bundle-probe-order-2.exp @@ -1,38 +1,65 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/cts/scheduler/exp/bundle-probe-order-3.exp b/cts/scheduler/exp/bundle-probe-order-3.exp index e1f60e7f5b..69140a4184 100644 --- a/cts/scheduler/exp/bundle-probe-order-3.exp +++ b/cts/scheduler/exp/bundle-probe-order-3.exp @@ -1,65 +1,74 @@ + + + + + + + + + - + - + - + - + - + - + diff --git a/cts/scheduler/exp/probe-pending-node.exp b/cts/scheduler/exp/probe-pending-node.exp new file mode 100644 index 0000000000..56e315ff01 --- /dev/null +++ b/cts/scheduler/exp/probe-pending-node.exp @@ -0,0 +1 @@ + diff --git a/cts/scheduler/scores/probe-pending-node.scores b/cts/scheduler/scores/probe-pending-node.scores new file mode 100644 index 0000000000..020a1a0371 --- /dev/null +++ b/cts/scheduler/scores/probe-pending-node.scores @@ -0,0 +1,61 @@ + +pcmk__clone_allocate: fs_UC5_SAPMNT-clone allocation score on gcdoubwap01: 0 +pcmk__clone_allocate: fs_UC5_SAPMNT-clone allocation score on gcdoubwap02: 0 +pcmk__clone_allocate: fs_UC5_SAPMNT:0 allocation score on gcdoubwap01: 0 +pcmk__clone_allocate: fs_UC5_SAPMNT:0 allocation score on gcdoubwap02: 0 +pcmk__clone_allocate: fs_UC5_SAPMNT:1 allocation score on gcdoubwap01: 0 +pcmk__clone_allocate: fs_UC5_SAPMNT:1 allocation score on gcdoubwap02: 0 +pcmk__clone_allocate: fs_UC5_SYS-clone allocation score on gcdoubwap01: 0 +pcmk__clone_allocate: fs_UC5_SYS-clone allocation score on gcdoubwap02: 0 +pcmk__clone_allocate: fs_UC5_SYS:0 allocation score on gcdoubwap01: 0 +pcmk__clone_allocate: fs_UC5_SYS:0 allocation score on gcdoubwap02: 0 +pcmk__clone_allocate: fs_UC5_SYS:1 allocation score on gcdoubwap01: 0 +pcmk__clone_allocate: fs_UC5_SYS:1 allocation score on gcdoubwap02: 0 +pcmk__group_allocate: fs_UC5_ascs allocation score on gcdoubwap01: 0 +pcmk__group_allocate: fs_UC5_ascs allocation score on gcdoubwap02: 0 +pcmk__group_allocate: fs_UC5_ers allocation score on gcdoubwap01: 0 +pcmk__group_allocate: fs_UC5_ers allocation score on gcdoubwap02: 0 +pcmk__group_allocate: grp_UC5_ascs allocation score on gcdoubwap01: 0 +pcmk__group_allocate: grp_UC5_ascs allocation score on gcdoubwap02: 0 +pcmk__group_allocate: grp_UC5_ers allocation score on gcdoubwap01: 0 +pcmk__group_allocate: grp_UC5_ers allocation score on gcdoubwap02: 0 +pcmk__group_allocate: rsc_sap_UC5_ASCS11 allocation score on gcdoubwap01: 0 +pcmk__group_allocate: rsc_sap_UC5_ASCS11 allocation score on gcdoubwap02: 0 +pcmk__group_allocate: rsc_sap_UC5_ERS12 allocation score on gcdoubwap01: 0 +pcmk__group_allocate: rsc_sap_UC5_ERS12 allocation score on gcdoubwap02: 0 +pcmk__group_allocate: rsc_vip_gcp_ascs allocation score on gcdoubwap01: INFINITY +pcmk__group_allocate: rsc_vip_gcp_ascs allocation score on gcdoubwap02: 0 +pcmk__group_allocate: rsc_vip_gcp_ers allocation score on gcdoubwap01: 0 +pcmk__group_allocate: rsc_vip_gcp_ers allocation score on gcdoubwap02: 0 +pcmk__group_allocate: rsc_vip_init_ers allocation score on gcdoubwap01: 0 +pcmk__group_allocate: rsc_vip_init_ers allocation score on gcdoubwap02: 0 +pcmk__group_allocate: rsc_vip_int_ascs allocation score on gcdoubwap01: 0 +pcmk__group_allocate: rsc_vip_int_ascs allocation score on gcdoubwap02: 0 +pcmk__native_allocate: fs_UC5_SAPMNT:0 allocation score on gcdoubwap01: 0 +pcmk__native_allocate: fs_UC5_SAPMNT:0 allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: fs_UC5_SAPMNT:1 allocation score on gcdoubwap01: 0 +pcmk__native_allocate: fs_UC5_SAPMNT:1 allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: fs_UC5_SYS:0 allocation score on gcdoubwap01: 0 +pcmk__native_allocate: fs_UC5_SYS:0 allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: fs_UC5_SYS:1 allocation score on gcdoubwap01: 0 +pcmk__native_allocate: fs_UC5_SYS:1 allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: fs_UC5_ascs allocation score on gcdoubwap01: 0 +pcmk__native_allocate: fs_UC5_ascs allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: fs_UC5_ers allocation score on gcdoubwap01: -INFINITY +pcmk__native_allocate: fs_UC5_ers allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: rsc_sap_UC5_ASCS11 allocation score on gcdoubwap01: -INFINITY +pcmk__native_allocate: rsc_sap_UC5_ASCS11 allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: rsc_sap_UC5_ERS12 allocation score on gcdoubwap01: -INFINITY +pcmk__native_allocate: rsc_sap_UC5_ERS12 allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: rsc_vip_gcp_ascs allocation score on gcdoubwap01: -INFINITY +pcmk__native_allocate: rsc_vip_gcp_ascs allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: rsc_vip_gcp_ers allocation score on gcdoubwap01: -INFINITY +pcmk__native_allocate: rsc_vip_gcp_ers allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: rsc_vip_init_ers allocation score on gcdoubwap01: 0 +pcmk__native_allocate: rsc_vip_init_ers allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: rsc_vip_int_ascs allocation score on gcdoubwap01: INFINITY +pcmk__native_allocate: rsc_vip_int_ascs allocation score on gcdoubwap02: -INFINITY +pcmk__native_allocate: stonith_gcdoubwap01 allocation score on gcdoubwap01: -INFINITY +pcmk__native_allocate: stonith_gcdoubwap01 allocation score on gcdoubwap02: 0 +pcmk__native_allocate: stonith_gcdoubwap02 allocation score on gcdoubwap01: 0 +pcmk__native_allocate: stonith_gcdoubwap02 allocation score on gcdoubwap02: -INFINITY diff --git a/cts/scheduler/summary/bundle-probe-order-2.summary b/cts/scheduler/summary/bundle-probe-order-2.summary index 681d607669..024c472ff9 100644 --- a/cts/scheduler/summary/bundle-probe-order-2.summary +++ b/cts/scheduler/summary/bundle-probe-order-2.summary @@ -1,31 +1,34 @@ Using the original execution date of: 2017-10-12 07:31:57Z Current cluster status: * Node List: * GuestNode galera-bundle-0@centos2: maintenance * Online: [ centos1 centos2 centos3 ] * Full List of Resources: * Container bundle set: galera-bundle [docker.io/tripleoupstream/centos-binary-mariadb:latest] (unmanaged): * galera-bundle-0 (ocf:heartbeat:galera): Stopped centos2 (unmanaged) * galera-bundle-1 (ocf:heartbeat:galera): Stopped (unmanaged) * galera-bundle-2 (ocf:heartbeat:galera): Stopped (unmanaged) Transition Summary: Executing Cluster Transition: + * Resource action: galera:0 monitor on galera-bundle-0 + * Resource action: galera-bundle-docker-0 monitor=60000 on centos2 + * Resource action: galera-bundle-0 monitor=30000 on centos2 * Resource action: galera-bundle-docker-1 monitor on centos2 * Resource action: galera-bundle-docker-2 monitor on centos3 * Resource action: galera-bundle-docker-2 monitor on centos2 * Resource action: galera-bundle-docker-2 monitor on centos1 Using the original execution date of: 2017-10-12 07:31:57Z Revised Cluster Status: * Node List: * GuestNode galera-bundle-0@centos2: maintenance * Online: [ centos1 centos2 centos3 ] * Full List of Resources: * Container bundle set: galera-bundle [docker.io/tripleoupstream/centos-binary-mariadb:latest] (unmanaged): * galera-bundle-0 (ocf:heartbeat:galera): Stopped centos2 (unmanaged) * galera-bundle-1 (ocf:heartbeat:galera): Stopped (unmanaged) * galera-bundle-2 (ocf:heartbeat:galera): Stopped (unmanaged) diff --git a/cts/scheduler/summary/bundle-probe-order-3.summary b/cts/scheduler/summary/bundle-probe-order-3.summary index f089618a1c..331bd87f62 100644 --- a/cts/scheduler/summary/bundle-probe-order-3.summary +++ b/cts/scheduler/summary/bundle-probe-order-3.summary @@ -1,32 +1,33 @@ Using the original execution date of: 2017-10-12 07:31:57Z Current cluster status: * Node List: * Online: [ centos1 centos2 centos3 ] * Full List of Resources: * Container bundle set: galera-bundle [docker.io/tripleoupstream/centos-binary-mariadb:latest] (unmanaged): * galera-bundle-0 (ocf:heartbeat:galera): Stopped centos2 (unmanaged) * galera-bundle-1 (ocf:heartbeat:galera): Stopped (unmanaged) * galera-bundle-2 (ocf:heartbeat:galera): Stopped (unmanaged) Transition Summary: Executing Cluster Transition: + * Resource action: galera-bundle-docker-0 monitor=60000 on centos2 * Resource action: galera-bundle-0 monitor on centos3 * Resource action: galera-bundle-0 monitor on centos2 * Resource action: galera-bundle-0 monitor on centos1 * Resource action: galera-bundle-docker-1 monitor on centos2 * Resource action: galera-bundle-docker-2 monitor on centos3 * Resource action: galera-bundle-docker-2 monitor on centos2 * Resource action: galera-bundle-docker-2 monitor on centos1 Using the original execution date of: 2017-10-12 07:31:57Z Revised Cluster Status: * Node List: * Online: [ centos1 centos2 centos3 ] * Full List of Resources: * Container bundle set: galera-bundle [docker.io/tripleoupstream/centos-binary-mariadb:latest] (unmanaged): * galera-bundle-0 (ocf:heartbeat:galera): Stopped centos2 (unmanaged) * galera-bundle-1 (ocf:heartbeat:galera): Stopped (unmanaged) * galera-bundle-2 (ocf:heartbeat:galera): Stopped (unmanaged) diff --git a/cts/scheduler/summary/probe-pending-node.summary b/cts/scheduler/summary/probe-pending-node.summary new file mode 100644 index 0000000000..208186b834 --- /dev/null +++ b/cts/scheduler/summary/probe-pending-node.summary @@ -0,0 +1,55 @@ +Using the original execution date of: 2021-06-11 13:55:24Z + + *** Resource management is DISABLED *** + The cluster will not attempt to start, stop or recover services + +Current cluster status: + * Node List: + * Node gcdoubwap02: pending + * Online: [ gcdoubwap01 ] + + * Full List of Resources: + * stonith_gcdoubwap01 (stonith:fence_gce): Stopped (unmanaged) + * stonith_gcdoubwap02 (stonith:fence_gce): Stopped (unmanaged) + * Clone Set: fs_UC5_SAPMNT-clone [fs_UC5_SAPMNT] (unmanaged): + * Stopped: [ gcdoubwap01 gcdoubwap02 ] + * Clone Set: fs_UC5_SYS-clone [fs_UC5_SYS] (unmanaged): + * Stopped: [ gcdoubwap01 gcdoubwap02 ] + * Resource Group: grp_UC5_ascs (unmanaged): + * rsc_vip_int_ascs (ocf:heartbeat:IPaddr2): Stopped (unmanaged) + * rsc_vip_gcp_ascs (ocf:heartbeat:gcp-vpc-move-vip): Started gcdoubwap01 (unmanaged) + * fs_UC5_ascs (ocf:heartbeat:Filesystem): Stopped (unmanaged) + * rsc_sap_UC5_ASCS11 (ocf:heartbeat:SAPInstance): Stopped (unmanaged) + * Resource Group: grp_UC5_ers (unmanaged): + * rsc_vip_init_ers (ocf:heartbeat:IPaddr2): Stopped (unmanaged) + * rsc_vip_gcp_ers (ocf:heartbeat:gcp-vpc-move-vip): Stopped (unmanaged) + * fs_UC5_ers (ocf:heartbeat:Filesystem): Stopped (unmanaged) + * rsc_sap_UC5_ERS12 (ocf:heartbeat:SAPInstance): Stopped (unmanaged) + +Transition Summary: + +Executing Cluster Transition: +Using the original execution date of: 2021-06-11 13:55:24Z + +Revised Cluster Status: + * Node List: + * Node gcdoubwap02: pending + * Online: [ gcdoubwap01 ] + + * Full List of Resources: + * stonith_gcdoubwap01 (stonith:fence_gce): Stopped (unmanaged) + * stonith_gcdoubwap02 (stonith:fence_gce): Stopped (unmanaged) + * Clone Set: fs_UC5_SAPMNT-clone [fs_UC5_SAPMNT] (unmanaged): + * Stopped: [ gcdoubwap01 gcdoubwap02 ] + * Clone Set: fs_UC5_SYS-clone [fs_UC5_SYS] (unmanaged): + * Stopped: [ gcdoubwap01 gcdoubwap02 ] + * Resource Group: grp_UC5_ascs (unmanaged): + * rsc_vip_int_ascs (ocf:heartbeat:IPaddr2): Stopped (unmanaged) + * rsc_vip_gcp_ascs (ocf:heartbeat:gcp-vpc-move-vip): Started gcdoubwap01 (unmanaged) + * fs_UC5_ascs (ocf:heartbeat:Filesystem): Stopped (unmanaged) + * rsc_sap_UC5_ASCS11 (ocf:heartbeat:SAPInstance): Stopped (unmanaged) + * Resource Group: grp_UC5_ers (unmanaged): + * rsc_vip_init_ers (ocf:heartbeat:IPaddr2): Stopped (unmanaged) + * rsc_vip_gcp_ers (ocf:heartbeat:gcp-vpc-move-vip): Stopped (unmanaged) + * fs_UC5_ers (ocf:heartbeat:Filesystem): Stopped (unmanaged) + * rsc_sap_UC5_ERS12 (ocf:heartbeat:SAPInstance): Stopped (unmanaged) diff --git a/cts/scheduler/xml/probe-pending-node.xml b/cts/scheduler/xml/probe-pending-node.xml new file mode 100644 index 0000000000..9f55c9261c --- /dev/null +++ b/cts/scheduler/xml/probe-pending-node.xml @@ -0,0 +1,247 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/daemons/pacemakerd/pcmkd_corosync.c b/daemons/pacemakerd/pcmkd_corosync.c index c9f95018f5..4aec8706aa 100644 --- a/daemons/pacemakerd/pcmkd_corosync.c +++ b/daemons/pacemakerd/pcmkd_corosync.c @@ -1,352 +1,352 @@ /* * Copyright 2010-2021 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include "pacemakerd.h" #include #include /* for calls to stat() */ #include /* For basename() and dirname() */ #include #include /* For getpwname() */ #include #include #include #include #include #include /* for crm_ipc_is_authentic_process */ #include #include /* PCMK__SPECIAL_PID* */ static corosync_cfg_handle_t cfg_handle = 0; static mainloop_timer_t *reconnect_timer = NULL; /* =::=::=::= CFG - Shutdown stuff =::=::=::= */ static void cfg_shutdown_callback(corosync_cfg_handle_t h, corosync_cfg_shutdown_flags_t flags) { crm_info("Corosync wants to shut down: %s", (flags == COROSYNC_CFG_SHUTDOWN_FLAG_IMMEDIATE) ? "immediate" : (flags == COROSYNC_CFG_SHUTDOWN_FLAG_REGARDLESS) ? "forced" : "optional"); /* Never allow corosync to shut down while we're running */ corosync_cfg_replyto_shutdown(h, COROSYNC_CFG_SHUTDOWN_FLAG_NO); } static corosync_cfg_callbacks_t cfg_callbacks = { .corosync_cfg_shutdown_callback = cfg_shutdown_callback, }; static int pcmk_cfg_dispatch(gpointer user_data) { corosync_cfg_handle_t *handle = (corosync_cfg_handle_t *) user_data; cs_error_t rc = corosync_cfg_dispatch(*handle, CS_DISPATCH_ALL); if (rc != CS_OK) { return -1; } return 0; } static void close_cfg(void) { if (cfg_handle != 0) { #ifdef HAVE_COROSYNC_CFG_TRACKSTART /* Ideally, we would call corosync_cfg_trackstop(cfg_handle) here, but a * bug in corosync 3.1.1 and 3.1.2 makes it hang forever. Thankfully, * it's not necessary since we exit immediately after this. */ #endif corosync_cfg_finalize(cfg_handle); cfg_handle = 0; } } static gboolean cluster_reconnect_cb(gpointer data) { if (cluster_connect_cfg()) { mainloop_timer_del(reconnect_timer); reconnect_timer = NULL; crm_notice("Cluster reconnect succeeded"); } else { crm_info("Cluster reconnect failed" "(connection will be reattempted once per second)"); } /* * In theory this will continue forever. In practice the CIB connection from * attrd will timeout and shut down Pacemaker when it gets bored. */ return TRUE; } static void cfg_connection_destroy(gpointer user_data) { crm_warn("Lost connection to cluster layer " "(connection will be reattempted once per second)"); corosync_cfg_finalize(cfg_handle); cfg_handle = 0; reconnect_timer = mainloop_timer_add("corosync reconnect", 1000, TRUE, cluster_reconnect_cb, NULL); mainloop_timer_start(reconnect_timer); } void cluster_disconnect_cfg(void) { close_cfg(); if (reconnect_timer != NULL) { /* The mainloop should be gone by this point, so this isn't necessary, * but cleaning up memory should make valgrind happier. */ mainloop_timer_del(reconnect_timer); reconnect_timer = NULL; } } #define cs_repeat(counter, max, code) do { \ code; \ if(rc == CS_ERR_TRY_AGAIN || rc == CS_ERR_QUEUE_FULL) { \ counter++; \ crm_debug("Retrying Corosync operation after %ds", counter); \ sleep(counter); \ } else { \ break; \ } \ } while(counter < max) gboolean cluster_connect_cfg(void) { cs_error_t rc; int fd = -1, retries = 0, rv; uid_t found_uid = 0; gid_t found_gid = 0; pid_t found_pid = 0; uint32_t nodeid; static struct mainloop_fd_callbacks cfg_fd_callbacks = { .dispatch = pcmk_cfg_dispatch, .destroy = cfg_connection_destroy, }; cs_repeat(retries, 30, rc = corosync_cfg_initialize(&cfg_handle, &cfg_callbacks)); if (rc != CS_OK) { crm_crit("Could not connect to Corosync CFG: %s " CRM_XS " rc=%d", cs_strerror(rc), rc); return FALSE; } rc = corosync_cfg_fd_get(cfg_handle, &fd); if (rc != CS_OK) { crm_crit("Could not get Corosync CFG descriptor: %s " CRM_XS " rc=%d", cs_strerror(rc), rc); goto bail; } /* CFG provider run as root (in given user namespace, anyway)? */ if (!(rv = crm_ipc_is_authentic_process(fd, (uid_t) 0,(gid_t) 0, &found_pid, &found_uid, &found_gid))) { crm_crit("Rejecting Corosync CFG provider because process %lld " "is running as uid %lld gid %lld, not root", (long long) PCMK__SPECIAL_PID_AS_0(found_pid), (long long) found_uid, (long long) found_gid); goto bail; } else if (rv < 0) { crm_crit("Could not authenticate Corosync CFG provider: %s " CRM_XS " rc=%d", strerror(-rv), -rv); goto bail; } retries = 0; cs_repeat(retries, 30, rc = corosync_cfg_local_get(cfg_handle, &nodeid)); if (rc != CS_OK) { crm_crit("Could not get local node ID from Corosync: %s " CRM_XS " rc=%d", cs_strerror(rc), rc); goto bail; } crm_debug("Corosync reports local node ID is %lu", (unsigned long) nodeid); #ifdef HAVE_COROSYNC_CFG_TRACKSTART rc = corosync_cfg_trackstart(cfg_handle, 0); if (rc != CS_OK) { crm_crit("Could not enable Corosync CFG shutdown tracker: %s " CRM_XS " rc=%d", cs_strerror(rc), rc); goto bail; } #endif mainloop_add_fd("corosync-cfg", G_PRIORITY_DEFAULT, fd, &cfg_handle, &cfg_fd_callbacks); return TRUE; bail: corosync_cfg_finalize(cfg_handle); return FALSE; } void pcmkd_shutdown_corosync(void) { cs_error_t rc; if (cfg_handle == 0) { crm_warn("Unable to shut down Corosync: No connection"); return; } crm_info("Asking Corosync to shut down"); rc = corosync_cfg_try_shutdown(cfg_handle, COROSYNC_CFG_SHUTDOWN_FLAG_IMMEDIATE); if (rc == CS_OK) { close_cfg(); } else { crm_warn("Corosync shutdown failed: %s " CRM_XS " rc=%d", cs_strerror(rc), rc); } } /* =::=::=::= Configuration =::=::=::= */ static int get_config_opt(uint64_t unused, cmap_handle_t object_handle, const char *key, char **value, const char *fallback) { int rc = 0, retries = 0; cs_repeat(retries, 5, rc = cmap_get_string(object_handle, key, value)); if (rc != CS_OK) { crm_trace("Search for %s failed %d, defaulting to %s", key, rc, fallback); if (fallback) { *value = strdup(fallback); } else { *value = NULL; } } crm_trace("%s: %s", key, *value); return rc; } gboolean mcp_read_config(void) { cs_error_t rc = CS_OK; int retries = 0; cmap_handle_t local_handle; uint64_t config = 0; int fd = -1; uid_t found_uid = 0; gid_t found_gid = 0; pid_t found_pid = 0; int rv; enum cluster_type_e stack; // There can be only one possibility do { - rc = cmap_initialize_map(&local_handle, CMAP_MAP_ICMAP); + rc = pcmk__init_cmap(&local_handle); if (rc != CS_OK) { retries++; crm_info("Could not connect to Corosync CMAP: %s (retrying in %ds) " CRM_XS " rc=%d", cs_strerror(rc), retries, rc); sleep(retries); } else { break; } } while (retries < 5); if (rc != CS_OK) { crm_crit("Could not connect to Corosync CMAP: %s " CRM_XS " rc=%d", cs_strerror(rc), rc); return FALSE; } rc = cmap_fd_get(local_handle, &fd); if (rc != CS_OK) { crm_crit("Could not get Corosync CMAP descriptor: %s " CRM_XS " rc=%d", cs_strerror(rc), rc); cmap_finalize(local_handle); return FALSE; } /* CMAP provider run as root (in given user namespace, anyway)? */ if (!(rv = crm_ipc_is_authentic_process(fd, (uid_t) 0,(gid_t) 0, &found_pid, &found_uid, &found_gid))) { crm_crit("Rejecting Corosync CMAP provider because process %lld " "is running as uid %lld gid %lld, not root", (long long) PCMK__SPECIAL_PID_AS_0(found_pid), (long long) found_uid, (long long) found_gid); cmap_finalize(local_handle); return FALSE; } else if (rv < 0) { crm_crit("Could not authenticate Corosync CMAP provider: %s " CRM_XS " rc=%d", strerror(-rv), -rv); cmap_finalize(local_handle); return FALSE; } stack = get_cluster_type(); if (stack != pcmk_cluster_corosync) { crm_crit("Expected corosync stack but detected %s " CRM_XS " stack=%d", name_for_cluster_type(stack), stack); return FALSE; } crm_info("Reading configuration for %s stack", name_for_cluster_type(stack)); pcmk__set_env_option("cluster_type", "corosync"); pcmk__set_env_option("quorum_type", "corosync"); // If debug logging is not configured, check whether corosync has it if (pcmk__env_option("debug") == NULL) { char *debug_enabled = NULL; get_config_opt(config, local_handle, "logging.debug", &debug_enabled, "off"); if (crm_is_true(debug_enabled)) { pcmk__set_env_option("debug", "1"); if (get_crm_log_level() < LOG_DEBUG) { set_crm_log_level(LOG_DEBUG); } } else { pcmk__set_env_option("debug", "0"); } free(debug_enabled); } if(local_handle){ gid_t gid = 0; if (pcmk_daemon_user(NULL, &gid) < 0) { crm_warn("Could not authorize group with Corosync " CRM_XS " No group found for user %s", CRM_DAEMON_USER); } else { char key[PATH_MAX]; snprintf(key, PATH_MAX, "uidgid.gid.%u", gid); rc = cmap_set_uint8(local_handle, key, 1); if (rc != CS_OK) { crm_warn("Could not authorize group with Corosync: %s " CRM_XS " group=%u rc=%d", pcmk__cs_err_str(rc), gid, rc); } } } cmap_finalize(local_handle); return TRUE; } diff --git a/include/crm/cluster/internal.h b/include/crm/cluster/internal.h index bf75b8cc87..9bc57c617c 100644 --- a/include/crm/cluster/internal.h +++ b/include/crm/cluster/internal.h @@ -1,122 +1,133 @@ /* - * Copyright 2004-2020 the Pacemaker project contributors + * Copyright 2004-2021 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef CRM_CLUSTER_INTERNAL__H # define CRM_CLUSTER_INTERNAL__H # include // uint32_t, uint64_t # include /* *INDENT-OFF* */ enum crm_proc_flag { crm_proc_none = 0x00000001, // Cluster layers crm_proc_cpg = 0x04000000, // Daemons crm_proc_execd = 0x00000010, crm_proc_based = 0x00000100, crm_proc_controld = 0x00000200, crm_proc_attrd = 0x00001000, crm_proc_schedulerd = 0x00010000, crm_proc_fenced = 0x00100000, }; /* *INDENT-ON* */ /*! * \internal * \brief Return the process bit corresponding to the current cluster stack * * \return Process flag if detectable, otherwise 0 */ static inline uint32_t crm_get_cluster_proc(void) { switch (get_cluster_type()) { case pcmk_cluster_corosync: return crm_proc_cpg; default: break; } return crm_proc_none; } /*! * \internal * \brief Get log-friendly string description of a Corosync return code * * \param[in] error Corosync return code * * \return Log-friendly string description corresponding to \p error */ static inline const char * pcmk__cs_err_str(int error) { # if SUPPORT_COROSYNC switch (error) { case CS_OK: return "OK"; case CS_ERR_LIBRARY: return "Library error"; case CS_ERR_VERSION: return "Version error"; case CS_ERR_INIT: return "Initialization error"; case CS_ERR_TIMEOUT: return "Timeout"; case CS_ERR_TRY_AGAIN: return "Try again"; case CS_ERR_INVALID_PARAM: return "Invalid parameter"; case CS_ERR_NO_MEMORY: return "No memory"; case CS_ERR_BAD_HANDLE: return "Bad handle"; case CS_ERR_BUSY: return "Busy"; case CS_ERR_ACCESS: return "Access error"; case CS_ERR_NOT_EXIST: return "Doesn't exist"; case CS_ERR_NAME_TOO_LONG: return "Name too long"; case CS_ERR_EXIST: return "Exists"; case CS_ERR_NO_SPACE: return "No space"; case CS_ERR_INTERRUPT: return "Interrupt"; case CS_ERR_NAME_NOT_FOUND: return "Name not found"; case CS_ERR_NO_RESOURCES: return "No resources"; case CS_ERR_NOT_SUPPORTED: return "Not supported"; case CS_ERR_BAD_OPERATION: return "Bad operation"; case CS_ERR_FAILED_OPERATION: return "Failed operation"; case CS_ERR_MESSAGE_ERROR: return "Message error"; case CS_ERR_QUEUE_FULL: return "Queue full"; case CS_ERR_QUEUE_NOT_AVAILABLE: return "Queue not available"; case CS_ERR_BAD_FLAGS: return "Bad flags"; case CS_ERR_TOO_BIG: return "Too big"; case CS_ERR_NO_SECTIONS: return "No sections"; } # endif return "Corosync error"; } # if SUPPORT_COROSYNC + +#if 0 +/* This is the new way to do it, but we still support all Corosync 2 versions, + * and this isn't always available. A better alternative here would be to check + * for support in the configure script and enable this conditionally. + */ +#define pcmk__init_cmap(handle) cmap_initialize_map((handle), CMAP_MAP_ICMAP) +#else +#define pcmk__init_cmap(handle) cmap_initialize(handle) +#endif + char *pcmk__corosync_cluster_name(void); bool pcmk__corosync_add_nodes(xmlNode *xml_parent); # endif crm_node_t *crm_update_peer_proc(const char *source, crm_node_t * peer, uint32_t flag, const char *status); crm_node_t *pcmk__update_peer_state(const char *source, crm_node_t *node, const char *state, uint64_t membership); void pcmk__update_peer_expected(const char *source, crm_node_t *node, const char *expected); void pcmk__reap_unseen_nodes(uint64_t ring_id); void pcmk__corosync_quorum_connect(gboolean (*dispatch)(unsigned long long, gboolean), void (*destroy) (gpointer)); crm_node_t *pcmk__search_node_caches(unsigned int id, const char *uname, uint32_t flags); crm_node_t *pcmk__search_cluster_node_cache(unsigned int id, const char *uname); void pcmk__refresh_node_caches_from_cib(xmlNode *cib); crm_node_t *pcmk__search_known_node_cache(unsigned int id, const char *uname, uint32_t flags); #endif diff --git a/lib/cluster/corosync.c b/lib/cluster/corosync.c index 972fe8250a..d7bd453d9e 100644 --- a/lib/cluster/corosync.c +++ b/lib/cluster/corosync.c @@ -1,799 +1,814 @@ /* * Copyright 2004-2020 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include /* U64T ~ PRIu64 */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* PCMK__SPECIAL_PID* */ #include "crmcluster_private.h" static quorum_handle_t pcmk_quorum_handle = 0; static gboolean (*quorum_app_callback)(unsigned long long seq, gboolean quorate) = NULL; /*! * \internal * \brief Get the Corosync UUID associated with a Pacemaker node * * \param[in] node Pacemaker node * * \return Newly allocated string with node's Corosync UUID, or NULL if unknown * \note It is the caller's responsibility to free the result with free(). */ char * pcmk__corosync_uuid(crm_node_t *node) { if ((node != NULL) && is_corosync_cluster()) { if (node->id > 0) { return crm_strdup_printf("%u", node->id); } else { crm_info("Node %s is not yet known by corosync", node->uname); } } return NULL; } static bool node_name_is_valid(const char *key, const char *name) { int octet; if (name == NULL) { crm_trace("%s is empty", key); return false; } else if (sscanf(name, "%d.%d.%d.%d", &octet, &octet, &octet, &octet) == 4) { crm_trace("%s contains an IPv4 address (%s), ignoring", key, name); return false; } else if (strstr(name, ":") != NULL) { crm_trace("%s contains an IPv6 address (%s), ignoring", key, name); return false; } crm_trace("'%s: %s' is valid", key, name); return true; } /* * \internal * \brief Get Corosync node name corresponding to a node ID * * \param[in] cmap_handle Connection to Corosync CMAP * \param[in] nodeid Node ID to check * * \return Newly allocated string with name or (if no name) IP address * associated with first address assigned to a Corosync node ID (or NULL * if unknown) * \note It is the caller's responsibility to free the result with free(). */ char * pcmk__corosync_name(uint64_t /*cmap_handle_t */ cmap_handle, uint32_t nodeid) { // Originally based on corosync-quorumtool.c:node_name() int lpc = 0; cs_error_t rc = CS_OK; int retries = 0; char *name = NULL; cmap_handle_t local_handle = 0; int fd = -1; uid_t found_uid = 0; gid_t found_gid = 0; pid_t found_pid = 0; int rv; if (nodeid == 0) { nodeid = get_local_nodeid(0); } if (cmap_handle == 0 && local_handle == 0) { retries = 0; crm_trace("Initializing CMAP connection"); do { - rc = cmap_initialize_map(&local_handle, CMAP_MAP_ICMAP); + rc = pcmk__init_cmap(&local_handle); if (rc != CS_OK) { retries++; crm_debug("API connection setup failed: %s. Retrying in %ds", cs_strerror(rc), retries); sleep(retries); } } while (retries < 5 && rc != CS_OK); if (rc != CS_OK) { crm_warn("Could not connect to Cluster Configuration Database API, error %s", cs_strerror(rc)); local_handle = 0; } } if (cmap_handle == 0) { cmap_handle = local_handle; rc = cmap_fd_get(cmap_handle, &fd); if (rc != CS_OK) { crm_err("Could not obtain the CMAP API connection: %s (%d)", cs_strerror(rc), rc); goto bail; } /* CMAP provider run as root (in given user namespace, anyway)? */ if (!(rv = crm_ipc_is_authentic_process(fd, (uid_t) 0,(gid_t) 0, &found_pid, &found_uid, &found_gid))) { crm_err("CMAP provider is not authentic:" " process %lld (uid: %lld, gid: %lld)", (long long) PCMK__SPECIAL_PID_AS_0(found_pid), (long long) found_uid, (long long) found_gid); goto bail; } else if (rv < 0) { crm_err("Could not verify authenticity of CMAP provider: %s (%d)", strerror(-rv), -rv); goto bail; } } while (name == NULL && cmap_handle != 0) { uint32_t id = 0; char *key = NULL; key = crm_strdup_printf("nodelist.node.%d.nodeid", lpc); rc = cmap_get_uint32(cmap_handle, key, &id); crm_trace("Checking %u vs %u from %s", nodeid, id, key); free(key); if (rc != CS_OK) { break; } if (nodeid == id) { crm_trace("Searching for node name for %u in nodelist.node.%d %s", nodeid, lpc, crm_str(name)); if (name == NULL) { key = crm_strdup_printf("nodelist.node.%d.name", lpc); cmap_get_string(cmap_handle, key, &name); crm_trace("%s = %s", key, crm_str(name)); free(key); } if (name == NULL) { key = crm_strdup_printf("nodelist.node.%d.ring0_addr", lpc); cmap_get_string(cmap_handle, key, &name); crm_trace("%s = %s", key, crm_str(name)); if (!node_name_is_valid(key, name)) { free(name); name = NULL; } free(key); } break; } lpc++; } bail: if(local_handle) { cmap_finalize(local_handle); } if (name == NULL) { crm_info("Unable to get node name for nodeid %u", nodeid); } return name; } /*! * \internal * \brief Disconnect from Corosync cluster * * \param[in] cluster Cluster connection to disconnect */ void pcmk__corosync_disconnect(crm_cluster_t *cluster) { cluster_disconnect_cpg(cluster); if (pcmk_quorum_handle) { quorum_finalize(pcmk_quorum_handle); pcmk_quorum_handle = 0; } crm_notice("Disconnected from Corosync"); } /*! * \internal * \brief Dispatch function for quorum connection file descriptor * * \param[in] user_data Ignored * * \return 0 on success, -1 on error (per mainloop_io_t interface) */ static int quorum_dispatch_cb(gpointer user_data) { int rc = quorum_dispatch(pcmk_quorum_handle, CS_DISPATCH_ALL); if (rc < 0) { crm_err("Connection to the Quorum API failed: %d", rc); quorum_finalize(pcmk_quorum_handle); pcmk_quorum_handle = 0; return -1; } return 0; } /*! * \internal * \brief Notification callback for Corosync quorum connection * * \param[in] handle Corosync quorum connection * \param[in] quorate Whether cluster is quorate * \param[in] ring_id Corosync ring ID * \param[in] view_list_entries Number of entries in \p view_list * \param[in] view_list Corosync node IDs in membership */ static void quorum_notification_cb(quorum_handle_t handle, uint32_t quorate, uint64_t ring_id, uint32_t view_list_entries, uint32_t *view_list) { int i; GHashTableIter iter; crm_node_t *node = NULL; static gboolean init_phase = TRUE; if (quorate != crm_have_quorum) { if (quorate) { crm_notice("Quorum acquired " CRM_XS " membership=%" U64T " members=%lu", ring_id, (long unsigned int)view_list_entries); } else { crm_warn("Quorum lost " CRM_XS " membership=%" U64T " members=%lu", ring_id, (long unsigned int)view_list_entries); } crm_have_quorum = quorate; } else { crm_info("Quorum %s " CRM_XS " membership=%" U64T " members=%lu", (quorate? "retained" : "still lost"), ring_id, (long unsigned int)view_list_entries); } if (view_list_entries == 0 && init_phase) { crm_info("Corosync membership is still forming, ignoring"); return; } init_phase = FALSE; /* Reset last_seen for all cached nodes so we can tell which ones aren't * in the view list */ g_hash_table_iter_init(&iter, crm_peer_cache); while (g_hash_table_iter_next(&iter, NULL, (gpointer *) &node)) { node->last_seen = 0; } /* Update the peer cache for each node in view list */ for (i = 0; i < view_list_entries; i++) { uint32_t id = view_list[i]; crm_debug("Member[%d] %u ", i, id); /* Get this node's peer cache entry (adding one if not already there) */ node = crm_get_peer(id, NULL); if (node->uname == NULL) { char *name = pcmk__corosync_name(0, id); crm_info("Obtaining name for new node %u", id); node = crm_get_peer(id, name); free(name); } /* Update the node state (including updating last_seen to ring_id) */ pcmk__update_peer_state(__func__, node, CRM_NODE_MEMBER, ring_id); } /* Remove any peer cache entries we didn't update */ pcmk__reap_unseen_nodes(ring_id); if (quorum_app_callback) { quorum_app_callback(ring_id, quorate); } } /*! * \internal * \brief Connect to Corosync quorum service * * \param[in] dispatch Connection dispatch callback * \param[in] destroy Connection destroy callback */ void pcmk__corosync_quorum_connect(gboolean (*dispatch)(unsigned long long, gboolean), void (*destroy)(gpointer)) { cs_error_t rc; int fd = 0; int quorate = 0; uint32_t quorum_type = 0; struct mainloop_fd_callbacks quorum_fd_callbacks; - quorum_model_v0_data_t quorum_model_data; uid_t found_uid = 0; gid_t found_gid = 0; pid_t found_pid = 0; int rv; quorum_fd_callbacks.dispatch = quorum_dispatch_cb; quorum_fd_callbacks.destroy = destroy; - quorum_model_data.model = QUORUM_MODEL_V0; - quorum_model_data.quorum_notify_fn = quorum_notification_cb; - crm_debug("Configuring Pacemaker to obtain quorum from Corosync"); - rc = quorum_model_initialize(&pcmk_quorum_handle, QUORUM_MODEL_V0, - (quorum_model_data_t *)&quorum_model_data, &quorum_type, NULL); + { +#if 0 + // New way but not supported by all Corosync 2 versions + quorum_model_v0_data_t quorum_model_data = { + .model = QUORUM_MODEL_V0, + .quorum_notify_fn = quorum_notification_cb, + }; + + rc = quorum_model_initialize(&pcmk_quorum_handle, QUORUM_MODEL_V0, + (quorum_model_data_t *) &quorum_model_data, + &quorum_type, NULL); +#else + quorum_callbacks_t quorum_callbacks = { + .quorum_notify_fn = quorum_notification_cb, + }; + + rc = quorum_initialize(&pcmk_quorum_handle, &quorum_callbacks, + &quorum_type); +#endif + } + if (rc != CS_OK) { crm_err("Could not connect to the Quorum API: %s (%d)", cs_strerror(rc), rc); goto bail; } else if (quorum_type != QUORUM_SET) { crm_err("Corosync quorum is not configured"); goto bail; } rc = quorum_fd_get(pcmk_quorum_handle, &fd); if (rc != CS_OK) { crm_err("Could not obtain the Quorum API connection: %s (%d)", strerror(rc), rc); goto bail; } /* Quorum provider run as root (in given user namespace, anyway)? */ if (!(rv = crm_ipc_is_authentic_process(fd, (uid_t) 0,(gid_t) 0, &found_pid, &found_uid, &found_gid))) { crm_err("Quorum provider is not authentic:" " process %lld (uid: %lld, gid: %lld)", (long long) PCMK__SPECIAL_PID_AS_0(found_pid), (long long) found_uid, (long long) found_gid); rc = CS_ERR_ACCESS; goto bail; } else if (rv < 0) { crm_err("Could not verify authenticity of Quorum provider: %s (%d)", strerror(-rv), -rv); rc = CS_ERR_ACCESS; goto bail; } rc = quorum_getquorate(pcmk_quorum_handle, &quorate); if (rc != CS_OK) { crm_err("Could not obtain the current Quorum API state: %d", rc); goto bail; } if (quorate) { crm_notice("Quorum acquired"); } else { crm_warn("Quorum lost"); } quorum_app_callback = dispatch; crm_have_quorum = quorate; rc = quorum_trackstart(pcmk_quorum_handle, CS_TRACK_CHANGES | CS_TRACK_CURRENT); if (rc != CS_OK) { crm_err("Could not setup Quorum API notifications: %d", rc); goto bail; } mainloop_add_fd("quorum", G_PRIORITY_HIGH, fd, dispatch, &quorum_fd_callbacks); pcmk__corosync_add_nodes(NULL); bail: if (rc != CS_OK) { quorum_finalize(pcmk_quorum_handle); } } /*! * \internal * \brief Connect to Corosync cluster layer * * \param[in] cluster Initialized cluster object to connect */ gboolean pcmk__corosync_connect(crm_cluster_t *cluster) { crm_node_t *peer = NULL; enum cluster_type_e stack = get_cluster_type(); crm_peer_init(); if (stack != pcmk_cluster_corosync) { crm_err("Invalid cluster type: %s " CRM_XS " stack=%d", name_for_cluster_type(stack), stack); return FALSE; } if (!cluster_connect_cpg(cluster)) { // Error message was logged by cluster_connect_cpg() return FALSE; } crm_info("Connection to %s established", name_for_cluster_type(stack)); cluster->nodeid = get_local_nodeid(0); if (cluster->nodeid == 0) { crm_err("Could not determine local node ID"); return FALSE; } cluster->uname = get_node_name(0); if (cluster->uname == NULL) { crm_err("Could not determine local node name"); return FALSE; } // Ensure local node always exists in peer cache peer = crm_get_peer(cluster->nodeid, cluster->uname); cluster->uuid = pcmk__corosync_uuid(peer); return TRUE; } /*! * \internal * \brief Check whether a Corosync cluster is active * * \return pcmk_cluster_corosync if Corosync is found, else pcmk_cluster_unknown */ enum cluster_type_e pcmk__corosync_detect(void) { int rc = CS_OK; cmap_handle_t handle; - rc = cmap_initialize_map(&handle, CMAP_MAP_ICMAP); + rc = pcmk__init_cmap(&handle); switch(rc) { case CS_OK: break; case CS_ERR_SECURITY: crm_debug("Failed to initialize the cmap API: Permission denied (%d)", rc); /* It's there, we just can't talk to it. * Good enough for us to identify as 'corosync' */ return pcmk_cluster_corosync; default: crm_info("Failed to initialize the cmap API: %s (%d)", pcmk__cs_err_str(rc), rc); return pcmk_cluster_unknown; } cmap_finalize(handle); return pcmk_cluster_corosync; } /*! * \brief Check whether a Corosync cluster peer is active * * \param[in] node Node to check * * \return TRUE if \p node is an active Corosync peer, otherwise FALSE */ gboolean crm_is_corosync_peer_active(const crm_node_t *node) { if (node == NULL) { crm_trace("Corosync peer inactive: NULL"); return FALSE; } else if (!pcmk__str_eq(node->state, CRM_NODE_MEMBER, pcmk__str_casei)) { crm_trace("Corosync peer %s inactive: state=%s", node->uname, node->state); return FALSE; } else if (!pcmk_is_set(node->processes, crm_proc_cpg)) { crm_trace("Corosync peer %s inactive: processes=%.16x", node->uname, node->processes); return FALSE; } return TRUE; } /*! * \internal * \brief Load Corosync node list (via CMAP) into peer cache and optionally XML * * \param[in] xml_parent If not NULL, add a entry to this for each node * * \return true if any nodes were found, false otherwise */ bool pcmk__corosync_add_nodes(xmlNode *xml_parent) { int lpc = 0; cs_error_t rc = CS_OK; int retries = 0; bool any = false; cmap_handle_t cmap_handle; int fd = -1; uid_t found_uid = 0; gid_t found_gid = 0; pid_t found_pid = 0; int rv; do { - rc = cmap_initialize_map(&cmap_handle, CMAP_MAP_ICMAP); + rc = pcmk__init_cmap(&cmap_handle); if (rc != CS_OK) { retries++; crm_debug("API connection setup failed: %s. Retrying in %ds", cs_strerror(rc), retries); sleep(retries); } } while (retries < 5 && rc != CS_OK); if (rc != CS_OK) { crm_warn("Could not connect to Cluster Configuration Database API, error %d", rc); return false; } rc = cmap_fd_get(cmap_handle, &fd); if (rc != CS_OK) { crm_err("Could not obtain the CMAP API connection: %s (%d)", cs_strerror(rc), rc); goto bail; } /* CMAP provider run as root (in given user namespace, anyway)? */ if (!(rv = crm_ipc_is_authentic_process(fd, (uid_t) 0,(gid_t) 0, &found_pid, &found_uid, &found_gid))) { crm_err("CMAP provider is not authentic:" " process %lld (uid: %lld, gid: %lld)", (long long) PCMK__SPECIAL_PID_AS_0(found_pid), (long long) found_uid, (long long) found_gid); goto bail; } else if (rv < 0) { crm_err("Could not verify authenticity of CMAP provider: %s (%d)", strerror(-rv), -rv); goto bail; } crm_peer_init(); crm_trace("Initializing corosync nodelist"); for (lpc = 0; TRUE; lpc++) { uint32_t nodeid = 0; char *name = NULL; char *key = NULL; key = crm_strdup_printf("nodelist.node.%d.nodeid", lpc); rc = cmap_get_uint32(cmap_handle, key, &nodeid); free(key); if (rc != CS_OK) { break; } name = pcmk__corosync_name(cmap_handle, nodeid); if (name != NULL) { GHashTableIter iter; crm_node_t *node = NULL; g_hash_table_iter_init(&iter, crm_peer_cache); while (g_hash_table_iter_next(&iter, NULL, (gpointer *) &node)) { if(node && node->uname && strcasecmp(node->uname, name) == 0) { if (node->id && node->id != nodeid) { crm_crit("Nodes %u and %u share the same name '%s': shutting down", node->id, nodeid, name); crm_exit(CRM_EX_FATAL); } } } } if (nodeid > 0 || name != NULL) { crm_trace("Initializing node[%d] %u = %s", lpc, nodeid, name); crm_get_peer(nodeid, name); } if (nodeid > 0 && name != NULL) { any = true; if (xml_parent) { xmlNode *node = create_xml_node(xml_parent, XML_CIB_TAG_NODE); crm_xml_set_id(node, "%u", nodeid); crm_xml_add(node, XML_ATTR_UNAME, name); } } free(name); } bail: cmap_finalize(cmap_handle); return any; } /*! * \internal * \brief Get cluster name from Corosync configuration (via CMAP) * * \return Newly allocated string with cluster name if configured, or NULL */ char * pcmk__corosync_cluster_name(void) { cmap_handle_t handle; char *cluster_name = NULL; cs_error_t rc = CS_OK; int fd = -1; uid_t found_uid = 0; gid_t found_gid = 0; pid_t found_pid = 0; int rv; - rc = cmap_initialize_map(&handle, CMAP_MAP_ICMAP); + rc = pcmk__init_cmap(&handle); if (rc != CS_OK) { crm_info("Failed to initialize the cmap API: %s (%d)", cs_strerror(rc), rc); return NULL; } rc = cmap_fd_get(handle, &fd); if (rc != CS_OK) { crm_err("Could not obtain the CMAP API connection: %s (%d)", cs_strerror(rc), rc); goto bail; } /* CMAP provider run as root (in given user namespace, anyway)? */ if (!(rv = crm_ipc_is_authentic_process(fd, (uid_t) 0,(gid_t) 0, &found_pid, &found_uid, &found_gid))) { crm_err("CMAP provider is not authentic:" " process %lld (uid: %lld, gid: %lld)", (long long) PCMK__SPECIAL_PID_AS_0(found_pid), (long long) found_uid, (long long) found_gid); goto bail; } else if (rv < 0) { crm_err("Could not verify authenticity of CMAP provider: %s (%d)", strerror(-rv), -rv); goto bail; } rc = cmap_get_string(handle, "totem.cluster_name", &cluster_name); if (rc != CS_OK) { crm_info("Cannot get totem.cluster_name: %s (%d)", cs_strerror(rc), rc); } else { crm_debug("cmap totem.cluster_name = '%s'", cluster_name); } bail: cmap_finalize(handle); return cluster_name; } /*! * \internal * \brief Check (via CMAP) whether Corosync configuration has a node list * * \return true if Corosync has node list, otherwise false */ bool pcmk__corosync_has_nodelist(void) { cs_error_t cs_rc = CS_OK; int retries = 0; cmap_handle_t cmap_handle; cmap_iter_handle_t iter_handle; char key_name[CMAP_KEYNAME_MAXLEN + 1]; int fd = -1; uid_t found_uid = 0; gid_t found_gid = 0; pid_t found_pid = 0; int rc = pcmk_ok; static bool got_result = false; static bool result = false; if (got_result) { return result; } // Connect to CMAP do { - cs_rc = cmap_initialize_map(&cmap_handle, CMAP_MAP_ICMAP); + cs_rc = pcmk__init_cmap(&cmap_handle); if (cs_rc != CS_OK) { retries++; crm_debug("CMAP connection failed: %s (rc=%d, retrying in %ds)", cs_strerror(cs_rc), cs_rc, retries); sleep(retries); } } while ((retries < 5) && (cs_rc != CS_OK)); if (cs_rc != CS_OK) { crm_warn("Assuming Corosync does not have node list: " "CMAP connection failed (%s) " CRM_XS " rc=%d", cs_strerror(cs_rc), cs_rc); return false; } // Get CMAP connection file descriptor cs_rc = cmap_fd_get(cmap_handle, &fd); if (cs_rc != CS_OK) { crm_warn("Assuming Corosync does not have node list: " "CMAP unusable (%s) " CRM_XS " rc=%d", cs_strerror(cs_rc), cs_rc); goto bail; } // Check whether CMAP connection is authentic (i.e. provided by root) rc = crm_ipc_is_authentic_process(fd, (uid_t) 0, (gid_t) 0, &found_pid, &found_uid, &found_gid); if (rc == 0) { crm_warn("Assuming Corosync does not have node list: " "CMAP provider is inauthentic " CRM_XS " pid=%lld uid=%lld gid=%lld", (long long) PCMK__SPECIAL_PID_AS_0(found_pid), (long long) found_uid, (long long) found_gid); goto bail; } else if (rc < 0) { crm_warn("Assuming Corosync does not have node list: " "Could not verify CMAP authenticity (%s) " CRM_XS " rc=%d", pcmk_strerror(rc), rc); goto bail; } // Check whether nodelist section is presetn cs_rc = cmap_iter_init(cmap_handle, "nodelist", &iter_handle); if (cs_rc != CS_OK) { crm_warn("Assuming Corosync does not have node list: " "CMAP not readable (%s) " CRM_XS " rc=%d", cs_strerror(cs_rc), cs_rc); goto bail; } cs_rc = cmap_iter_next(cmap_handle, iter_handle, key_name, NULL, NULL); if (cs_rc == CS_OK) { result = true; } cmap_iter_finalize(cmap_handle, iter_handle); got_result = true; crm_debug("Corosync %s node list", (result? "has" : "does not have")); bail: cmap_finalize(cmap_handle); return result; } diff --git a/lib/pengine/utils.c b/lib/pengine/utils.c index 5ef742e7c3..965824bd81 100644 --- a/lib/pengine/utils.c +++ b/lib/pengine/utils.c @@ -1,2494 +1,2500 @@ /* * Copyright 2004-2021 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include "pe_status_private.h" extern bool pcmk__is_daemon; extern xmlNode *get_object_root(const char *object_type, xmlNode * the_root); void print_str_str(gpointer key, gpointer value, gpointer user_data); gboolean ghash_free_str_str(gpointer key, gpointer value, gpointer user_data); static void unpack_operation(pe_action_t * action, xmlNode * xml_obj, pe_resource_t * container, pe_working_set_t * data_set, guint interval_ms); static xmlNode *find_rsc_op_entry_helper(pe_resource_t * rsc, const char *key, gboolean include_disabled); #if ENABLE_VERSIONED_ATTRS pe_rsc_action_details_t * pe_rsc_action_details(pe_action_t *action) { pe_rsc_action_details_t *details; CRM_CHECK(action != NULL, return NULL); if (action->action_details == NULL) { action->action_details = calloc(1, sizeof(pe_rsc_action_details_t)); CRM_CHECK(action->action_details != NULL, return NULL); } details = (pe_rsc_action_details_t *) action->action_details; if (details->versioned_parameters == NULL) { details->versioned_parameters = create_xml_node(NULL, XML_TAG_OP_VER_ATTRS); } if (details->versioned_meta == NULL) { details->versioned_meta = create_xml_node(NULL, XML_TAG_OP_VER_META); } return details; } static void pe_free_rsc_action_details(pe_action_t *action) { pe_rsc_action_details_t *details; if ((action == NULL) || (action->action_details == NULL)) { return; } details = (pe_rsc_action_details_t *) action->action_details; if (details->versioned_parameters) { free_xml(details->versioned_parameters); } if (details->versioned_meta) { free_xml(details->versioned_meta); } action->action_details = NULL; } #endif /*! * \internal * \brief Check whether we can fence a particular node * * \param[in] data_set Working set for cluster * \param[in] node Name of node to check * * \return true if node can be fenced, false otherwise */ bool pe_can_fence(pe_working_set_t *data_set, pe_node_t *node) { if (pe__is_guest_node(node)) { /* Guest nodes are fenced by stopping their container resource. We can * do that if the container's host is either online or fenceable. */ pe_resource_t *rsc = node->details->remote_rsc->container; for (GList *n = rsc->running_on; n != NULL; n = n->next) { pe_node_t *container_node = n->data; if (!container_node->details->online && !pe_can_fence(data_set, container_node)) { return false; } } return true; } else if (!pcmk_is_set(data_set->flags, pe_flag_stonith_enabled)) { return false; /* Turned off */ } else if (!pcmk_is_set(data_set->flags, pe_flag_have_stonith_resource)) { return false; /* No devices */ } else if (pcmk_is_set(data_set->flags, pe_flag_have_quorum)) { return true; } else if (data_set->no_quorum_policy == no_quorum_ignore) { return true; } else if(node == NULL) { return false; } else if(node->details->online) { crm_notice("We can fence %s without quorum because they're in our membership", node->details->uname); return true; } crm_trace("Cannot fence %s", node->details->uname); return false; } /*! * \internal * \brief Copy a node object * * \param[in] this_node Node object to copy * * \return Newly allocated shallow copy of this_node * \note This function asserts on errors and is guaranteed to return non-NULL. */ pe_node_t * pe__copy_node(const pe_node_t *this_node) { pe_node_t *new_node = NULL; CRM_ASSERT(this_node != NULL); new_node = calloc(1, sizeof(pe_node_t)); CRM_ASSERT(new_node != NULL); new_node->rsc_discover_mode = this_node->rsc_discover_mode; new_node->weight = this_node->weight; new_node->fixed = this_node->fixed; new_node->details = this_node->details; return new_node; } /* any node in list1 or list2 and not in the other gets a score of -INFINITY */ void node_list_exclude(GHashTable * hash, GList *list, gboolean merge_scores) { GHashTable *result = hash; pe_node_t *other_node = NULL; GList *gIter = list; GHashTableIter iter; pe_node_t *node = NULL; g_hash_table_iter_init(&iter, hash); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { other_node = pe_find_node_id(list, node->details->id); if (other_node == NULL) { node->weight = -INFINITY; } else if (merge_scores) { node->weight = pe__add_scores(node->weight, other_node->weight); } } for (; gIter != NULL; gIter = gIter->next) { pe_node_t *node = (pe_node_t *) gIter->data; other_node = pe_hash_table_lookup(result, node->details->id); if (other_node == NULL) { pe_node_t *new_node = pe__copy_node(node); new_node->weight = -INFINITY; g_hash_table_insert(result, (gpointer) new_node->details->id, new_node); } } } /*! * \internal * \brief Create a node hash table from a node list * * \param[in] list Node list * * \return Hash table equivalent of node list */ GHashTable * pe__node_list2table(GList *list) { GHashTable *result = NULL; result = pcmk__strkey_table(NULL, free); for (GList *gIter = list; gIter != NULL; gIter = gIter->next) { pe_node_t *new_node = pe__copy_node((pe_node_t *) gIter->data); g_hash_table_insert(result, (gpointer) new_node->details->id, new_node); } return result; } gint sort_node_uname(gconstpointer a, gconstpointer b) { return pcmk__numeric_strcasecmp(((const pe_node_t *) a)->details->uname, ((const pe_node_t *) b)->details->uname); } /*! * \internal * \brief Output node weights to stdout * * \param[in] rsc Use allowed nodes for this resource * \param[in] comment Text description to prefix lines with * \param[in] nodes If rsc is not specified, use these nodes */ static void pe__output_node_weights(pe_resource_t *rsc, const char *comment, GHashTable *nodes, pe_working_set_t *data_set) { pcmk__output_t *out = data_set->priv; char score[128]; // Stack-allocated since this is called frequently // Sort the nodes so the output is consistent for regression tests GList *list = g_list_sort(g_hash_table_get_values(nodes), sort_node_uname); for (GList *gIter = list; gIter != NULL; gIter = gIter->next) { pe_node_t *node = (pe_node_t *) gIter->data; score2char_stack(node->weight, score, sizeof(score)); out->message(out, "node-weight", rsc, comment, node->details->uname, score); } g_list_free(list); } /*! * \internal * \brief Log node weights at trace level * * \param[in] file Caller's filename * \param[in] function Caller's function name * \param[in] line Caller's line number * \param[in] rsc Use allowed nodes for this resource * \param[in] comment Text description to prefix lines with * \param[in] nodes If rsc is not specified, use these nodes */ static void pe__log_node_weights(const char *file, const char *function, int line, pe_resource_t *rsc, const char *comment, GHashTable *nodes) { GHashTableIter iter; pe_node_t *node = NULL; char score[128]; // Stack-allocated since this is called frequently // Don't waste time if we're not tracing at this point pcmk__log_else(LOG_TRACE, return); g_hash_table_iter_init(&iter, nodes); while (g_hash_table_iter_next(&iter, NULL, (void **) &node)) { score2char_stack(node->weight, score, sizeof(score)); if (rsc) { qb_log_from_external_source(function, file, "%s: %s allocation score on %s: %s", LOG_TRACE, line, 0, comment, rsc->id, node->details->uname, score); } else { qb_log_from_external_source(function, file, "%s: %s = %s", LOG_TRACE, line, 0, comment, node->details->uname, score); } } } /*! * \internal * \brief Log or output node weights * * \param[in] file Caller's filename * \param[in] function Caller's function name * \param[in] line Caller's line number * \param[in] to_log Log if true, otherwise output * \param[in] rsc Use allowed nodes for this resource * \param[in] comment Text description to prefix lines with * \param[in] nodes Use these nodes */ void pe__show_node_weights_as(const char *file, const char *function, int line, bool to_log, pe_resource_t *rsc, const char *comment, GHashTable *nodes, pe_working_set_t *data_set) { if (rsc != NULL && pcmk_is_set(rsc->flags, pe_rsc_orphan)) { // Don't show allocation scores for orphans return; } if (nodes == NULL) { // Nothing to show return; } if (to_log) { pe__log_node_weights(file, function, line, rsc, comment, nodes); } else { pe__output_node_weights(rsc, comment, nodes, data_set); } // If this resource has children, repeat recursively for each if (rsc && rsc->children) { for (GList *gIter = rsc->children; gIter != NULL; gIter = gIter->next) { pe_resource_t *child = (pe_resource_t *) gIter->data; pe__show_node_weights_as(file, function, line, to_log, child, comment, child->allowed_nodes, data_set); } } } gint sort_rsc_index(gconstpointer a, gconstpointer b) { const pe_resource_t *resource1 = (const pe_resource_t *)a; const pe_resource_t *resource2 = (const pe_resource_t *)b; if (a == NULL && b == NULL) { return 0; } if (a == NULL) { return 1; } if (b == NULL) { return -1; } if (resource1->sort_index > resource2->sort_index) { return -1; } if (resource1->sort_index < resource2->sort_index) { return 1; } return 0; } gint sort_rsc_priority(gconstpointer a, gconstpointer b) { const pe_resource_t *resource1 = (const pe_resource_t *)a; const pe_resource_t *resource2 = (const pe_resource_t *)b; if (a == NULL && b == NULL) { return 0; } if (a == NULL) { return 1; } if (b == NULL) { return -1; } if (resource1->priority > resource2->priority) { return -1; } if (resource1->priority < resource2->priority) { return 1; } return 0; } static enum pe_quorum_policy effective_quorum_policy(pe_resource_t *rsc, pe_working_set_t *data_set) { enum pe_quorum_policy policy = data_set->no_quorum_policy; if (pcmk_is_set(data_set->flags, pe_flag_have_quorum)) { policy = no_quorum_ignore; } else if (data_set->no_quorum_policy == no_quorum_demote) { switch (rsc->role) { case RSC_ROLE_PROMOTED: case RSC_ROLE_UNPROMOTED: if (rsc->next_role > RSC_ROLE_UNPROMOTED) { pe__set_next_role(rsc, RSC_ROLE_UNPROMOTED, "no-quorum-policy=demote"); } policy = no_quorum_ignore; break; default: policy = no_quorum_stop; break; } } return policy; } pe_action_t * custom_action(pe_resource_t * rsc, char *key, const char *task, pe_node_t * on_node, gboolean optional, gboolean save_action, pe_working_set_t * data_set) { pe_action_t *action = NULL; GList *possible_matches = NULL; CRM_CHECK(key != NULL, return NULL); CRM_CHECK(task != NULL, free(key); return NULL); if (save_action && rsc != NULL) { possible_matches = find_actions(rsc->actions, key, on_node); } else if(save_action) { #if 0 action = g_hash_table_lookup(data_set->singletons, key); #else /* More expensive but takes 'node' into account */ possible_matches = find_actions(data_set->actions, key, on_node); #endif } if(data_set->singletons == NULL) { data_set->singletons = pcmk__strkey_table(NULL, NULL); } if (possible_matches != NULL) { if (pcmk__list_of_multiple(possible_matches)) { pe_warn("Action %s for %s on %s exists %d times", task, rsc ? rsc->id : "", on_node ? on_node->details->uname : "", g_list_length(possible_matches)); } action = g_list_nth_data(possible_matches, 0); pe_rsc_trace(rsc, "Found action %d: %s for %s (%s) on %s", action->id, task, (rsc? rsc->id : "no resource"), action->uuid, (on_node? on_node->details->uname : "no node")); g_list_free(possible_matches); } if (action == NULL) { if (save_action) { pe_rsc_trace(rsc, "Creating action %d (%s): %s for %s (%s) on %s", data_set->action_id, (optional? "optional" : "required"), task, (rsc? rsc->id : "no resource"), key, (on_node? on_node->details->uname : "no node")); } action = calloc(1, sizeof(pe_action_t)); if (save_action) { action->id = data_set->action_id++; } else { action->id = 0; } action->rsc = rsc; action->task = strdup(task); if (on_node) { action->node = pe__copy_node(on_node); } action->uuid = strdup(key); if (pcmk__str_eq(task, CRM_OP_LRM_DELETE, pcmk__str_casei)) { // Resource history deletion for a node can be done on the DC pe__set_action_flags(action, pe_action_dc); } pe__set_action_flags(action, pe_action_runnable); if (optional) { pe__set_action_flags(action, pe_action_optional); } else { pe__clear_action_flags(action, pe_action_optional); } action->extra = pcmk__strkey_table(free, free); action->meta = pcmk__strkey_table(free, free); if (save_action) { data_set->actions = g_list_prepend(data_set->actions, action); if(rsc == NULL) { g_hash_table_insert(data_set->singletons, action->uuid, action); } } if (rsc != NULL) { guint interval_ms = 0; action->op_entry = find_rsc_op_entry_helper(rsc, key, TRUE); parse_op_key(key, NULL, NULL, &interval_ms); unpack_operation(action, action->op_entry, rsc->container, data_set, interval_ms); if (save_action) { rsc->actions = g_list_prepend(rsc->actions, action); } } } if (!optional && pcmk_is_set(action->flags, pe_action_optional)) { pe__clear_action_flags(action, pe_action_optional); } if (rsc != NULL) { enum action_tasks a_task = text2task(action->task); enum pe_quorum_policy quorum_policy = effective_quorum_policy(rsc, data_set); int warn_level = LOG_TRACE; if (save_action) { warn_level = LOG_WARNING; } if (!pcmk_is_set(action->flags, pe_action_have_node_attrs) && action->node != NULL && action->op_entry != NULL) { pe_rule_eval_data_t rule_data = { .node_hash = action->node->details->attrs, .role = RSC_ROLE_UNKNOWN, .now = data_set->now, .match_data = NULL, .rsc_data = NULL, .op_data = NULL }; pe__set_action_flags(action, pe_action_have_node_attrs); pe__unpack_dataset_nvpairs(action->op_entry, XML_TAG_ATTR_SETS, &rule_data, action->extra, NULL, FALSE, data_set); } + // Make the action optional if its resource is unmanaged + if (!pcmk_is_set(action->flags, pe_action_pseudo) + && (action->node != NULL) + && !pcmk_is_set(action->rsc->flags, pe_rsc_managed) + && (g_hash_table_lookup(action->meta, + XML_LRM_ATTR_INTERVAL_MS) == NULL)) { + pe_rsc_debug(rsc, "%s on %s is optional (%s is unmanaged)", + action->uuid, action->node->details->uname, + action->rsc->id); + pe__set_action_flags(action, pe_action_optional); + // We shouldn't clear runnable here because ... something + } + + // Make the action runnable or unrunnable as appropriate if (pcmk_is_set(action->flags, pe_action_pseudo)) { /* leave untouched */ } else if (action->node == NULL) { pe_rsc_trace(rsc, "%s is unrunnable (unallocated)", action->uuid); pe__clear_action_flags(action, pe_action_runnable); - } else if (!pcmk_is_set(rsc->flags, pe_rsc_managed) - && g_hash_table_lookup(action->meta, - XML_LRM_ATTR_INTERVAL_MS) == NULL) { - pe_rsc_debug(rsc, "%s on %s is optional (%s is unmanaged)", - action->uuid, action->node->details->uname, rsc->id); - pe__set_action_flags(action, pe_action_optional); - //pe__clear_action_flags(action, pe_action_runnable); - } else if (!pcmk_is_set(action->flags, pe_action_dc) && !(action->node->details->online) && (!pe__is_guest_node(action->node) || action->node->details->remote_requires_reset)) { pe__clear_action_flags(action, pe_action_runnable); do_crm_log(warn_level, "%s on %s is unrunnable (node is offline)", action->uuid, action->node->details->uname); if (pcmk_is_set(action->rsc->flags, pe_rsc_managed) && save_action && a_task == stop_rsc && action->node->details->unclean == FALSE) { pe_fence_node(data_set, action->node, "resource actions are unrunnable", FALSE); } } else if (!pcmk_is_set(action->flags, pe_action_dc) && action->node->details->pending) { pe__clear_action_flags(action, pe_action_runnable); do_crm_log(warn_level, "Action %s on %s is unrunnable (node is pending)", action->uuid, action->node->details->uname); } else if (action->needs == rsc_req_nothing) { pe_action_set_reason(action, NULL, TRUE); if (pe__is_guest_node(action->node) && !pe_can_fence(data_set, action->node)) { /* An action that requires nothing usually does not require any * fencing in order to be runnable. However, there is an * exception: an action cannot be completed if it is on a guest * node whose host is unclean and cannot be fenced. */ pe_rsc_debug(rsc, "%s on %s is unrunnable " "(node's host cannot be fenced)", action->uuid, action->node->details->uname); pe__clear_action_flags(action, pe_action_runnable); } else { pe_rsc_trace(rsc, "%s on %s does not require fencing or quorum", action->uuid, action->node->details->uname); pe__set_action_flags(action, pe_action_runnable); } #if 0 /* * No point checking this * - if we don't have quorum we can't stonith anyway */ } else if (action->needs == rsc_req_stonith) { crm_trace("Action %s requires only stonith", action->uuid); action->runnable = TRUE; #endif } else if (quorum_policy == no_quorum_stop) { pe_rsc_debug(rsc, "%s on %s is unrunnable (no quorum)", action->uuid, action->node->details->uname); pe_action_set_flag_reason(__func__, __LINE__, action, NULL, "no quorum", pe_action_runnable, TRUE); } else if (quorum_policy == no_quorum_freeze) { if (rsc->fns->active(rsc, TRUE) == FALSE || rsc->next_role > rsc->role) { pe_rsc_debug(rsc, "%s on %s is unrunnable (no quorum)", action->uuid, action->node->details->uname); pe_action_set_flag_reason(__func__, __LINE__, action, NULL, "quorum freeze", pe_action_runnable, TRUE); } } else { //pe_action_set_reason(action, NULL, TRUE); pe__set_action_flags(action, pe_action_runnable); } if (save_action) { switch (a_task) { case stop_rsc: pe__set_resource_flags(rsc, pe_rsc_stopping); break; case start_rsc: pe__clear_resource_flags(rsc, pe_rsc_starting); if (pcmk_is_set(action->flags, pe_action_runnable)) { pe__set_resource_flags(rsc, pe_rsc_starting); } break; default: break; } } } free(key); return action; } static bool valid_stop_on_fail(const char *value) { return !pcmk__strcase_any_of(value, "standby", "demote", "stop", NULL); } static const char * unpack_operation_on_fail(pe_action_t * action) { const char *name = NULL; const char *role = NULL; const char *on_fail = NULL; const char *interval_spec = NULL; const char *enabled = NULL; const char *value = g_hash_table_lookup(action->meta, XML_OP_ATTR_ON_FAIL); if (pcmk__str_eq(action->task, CRMD_ACTION_STOP, pcmk__str_casei) && !valid_stop_on_fail(value)) { pcmk__config_err("Resetting '" XML_OP_ATTR_ON_FAIL "' for %s stop " "action to default value because '%s' is not " "allowed for stop", action->rsc->id, value); return NULL; } else if (pcmk__str_eq(action->task, CRMD_ACTION_DEMOTE, pcmk__str_casei) && !value) { // demote on_fail defaults to monitor value for promoted role if present xmlNode *operation = NULL; CRM_CHECK(action->rsc != NULL, return NULL); for (operation = pcmk__xe_first_child(action->rsc->ops_xml); (operation != NULL) && (value == NULL); operation = pcmk__xe_next(operation)) { if (!pcmk__str_eq((const char *)operation->name, "op", pcmk__str_none)) { continue; } name = crm_element_value(operation, "name"); role = crm_element_value(operation, "role"); on_fail = crm_element_value(operation, XML_OP_ATTR_ON_FAIL); enabled = crm_element_value(operation, "enabled"); interval_spec = crm_element_value(operation, XML_LRM_ATTR_INTERVAL); if (!on_fail) { continue; } else if (enabled && !crm_is_true(enabled)) { continue; } else if (!pcmk__str_eq(name, "monitor", pcmk__str_casei) || !pcmk__strcase_any_of(role, RSC_ROLE_PROMOTED_S, RSC_ROLE_PROMOTED_LEGACY_S, NULL)) { continue; } else if (crm_parse_interval_spec(interval_spec) == 0) { continue; } else if (pcmk__str_eq(on_fail, "demote", pcmk__str_casei)) { continue; } value = on_fail; } } else if (pcmk__str_eq(action->task, CRM_OP_LRM_DELETE, pcmk__str_casei)) { value = "ignore"; } else if (pcmk__str_eq(value, "demote", pcmk__str_casei)) { name = crm_element_value(action->op_entry, "name"); role = crm_element_value(action->op_entry, "role"); interval_spec = crm_element_value(action->op_entry, XML_LRM_ATTR_INTERVAL); if (!pcmk__str_eq(name, CRMD_ACTION_PROMOTE, pcmk__str_casei) && (!pcmk__str_eq(name, CRMD_ACTION_STATUS, pcmk__str_casei) || !pcmk__strcase_any_of(role, RSC_ROLE_PROMOTED_S, RSC_ROLE_PROMOTED_LEGACY_S, NULL) || (crm_parse_interval_spec(interval_spec) == 0))) { pcmk__config_err("Resetting '" XML_OP_ATTR_ON_FAIL "' for %s %s " "action to default value because 'demote' is not " "allowed for it", action->rsc->id, name); return NULL; } } return value; } static xmlNode * find_min_interval_mon(pe_resource_t * rsc, gboolean include_disabled) { guint interval_ms = 0; guint min_interval_ms = G_MAXUINT; const char *name = NULL; const char *value = NULL; const char *interval_spec = NULL; xmlNode *op = NULL; xmlNode *operation = NULL; for (operation = pcmk__xe_first_child(rsc->ops_xml); operation != NULL; operation = pcmk__xe_next(operation)) { if (pcmk__str_eq((const char *)operation->name, "op", pcmk__str_none)) { name = crm_element_value(operation, "name"); interval_spec = crm_element_value(operation, XML_LRM_ATTR_INTERVAL); value = crm_element_value(operation, "enabled"); if (!include_disabled && value && crm_is_true(value) == FALSE) { continue; } if (!pcmk__str_eq(name, RSC_STATUS, pcmk__str_casei)) { continue; } interval_ms = crm_parse_interval_spec(interval_spec); if (interval_ms && (interval_ms < min_interval_ms)) { min_interval_ms = interval_ms; op = operation; } } } return op; } static int unpack_start_delay(const char *value, GHashTable *meta) { int start_delay = 0; if (value != NULL) { start_delay = crm_get_msec(value); if (start_delay < 0) { start_delay = 0; } if (meta) { g_hash_table_replace(meta, strdup(XML_OP_ATTR_START_DELAY), pcmk__itoa(start_delay)); } } return start_delay; } // true if value contains valid, non-NULL interval origin for recurring op static bool unpack_interval_origin(const char *value, xmlNode *xml_obj, guint interval_ms, crm_time_t *now, long long *start_delay) { long long result = 0; guint interval_sec = interval_ms / 1000; crm_time_t *origin = NULL; // Ignore unspecified values and non-recurring operations if ((value == NULL) || (interval_ms == 0) || (now == NULL)) { return false; } // Parse interval origin from text origin = crm_time_new(value); if (origin == NULL) { pcmk__config_err("Ignoring '" XML_OP_ATTR_ORIGIN "' for operation " "'%s' because '%s' is not valid", (ID(xml_obj)? ID(xml_obj) : "(missing ID)"), value); return false; } // Get seconds since origin (negative if origin is in the future) result = crm_time_get_seconds(now) - crm_time_get_seconds(origin); crm_time_free(origin); // Calculate seconds from closest interval to now result = result % interval_sec; // Calculate seconds remaining until next interval result = ((result <= 0)? 0 : interval_sec) - result; crm_info("Calculated a start delay of %llds for operation '%s'", result, (ID(xml_obj)? ID(xml_obj) : "(unspecified)")); if (start_delay != NULL) { *start_delay = result * 1000; // milliseconds } return true; } static int unpack_timeout(const char *value) { int timeout_ms = crm_get_msec(value); if (timeout_ms < 0) { timeout_ms = crm_get_msec(CRM_DEFAULT_OP_TIMEOUT_S); } return timeout_ms; } int pe_get_configured_timeout(pe_resource_t *rsc, const char *action, pe_working_set_t *data_set) { xmlNode *child = NULL; GHashTable *action_meta = NULL; const char *timeout_spec = NULL; int timeout_ms = 0; pe_rule_eval_data_t rule_data = { .node_hash = NULL, .role = RSC_ROLE_UNKNOWN, .now = data_set->now, .match_data = NULL, .rsc_data = NULL, .op_data = NULL }; for (child = first_named_child(rsc->ops_xml, XML_ATTR_OP); child != NULL; child = crm_next_same_xml(child)) { if (pcmk__str_eq(action, crm_element_value(child, XML_NVPAIR_ATTR_NAME), pcmk__str_casei)) { timeout_spec = crm_element_value(child, XML_ATTR_TIMEOUT); break; } } if (timeout_spec == NULL && data_set->op_defaults) { action_meta = pcmk__strkey_table(free, free); pe__unpack_dataset_nvpairs(data_set->op_defaults, XML_TAG_META_SETS, &rule_data, action_meta, NULL, FALSE, data_set); timeout_spec = g_hash_table_lookup(action_meta, XML_ATTR_TIMEOUT); } // @TODO check meta-attributes (including versioned meta-attributes) // @TODO maybe use min-interval monitor timeout as default for monitors timeout_ms = crm_get_msec(timeout_spec); if (timeout_ms < 0) { timeout_ms = crm_get_msec(CRM_DEFAULT_OP_TIMEOUT_S); } if (action_meta != NULL) { g_hash_table_destroy(action_meta); } return timeout_ms; } #if ENABLE_VERSIONED_ATTRS static void unpack_versioned_meta(xmlNode *versioned_meta, xmlNode *xml_obj, guint interval_ms, crm_time_t *now) { xmlNode *attrs = NULL; xmlNode *attr = NULL; for (attrs = pcmk__xe_first_child(versioned_meta); attrs != NULL; attrs = pcmk__xe_next(attrs)) { for (attr = pcmk__xe_first_child(attrs); attr != NULL; attr = pcmk__xe_next(attr)) { const char *name = crm_element_value(attr, XML_NVPAIR_ATTR_NAME); const char *value = crm_element_value(attr, XML_NVPAIR_ATTR_VALUE); if (pcmk__str_eq(name, XML_OP_ATTR_START_DELAY, pcmk__str_casei)) { int start_delay = unpack_start_delay(value, NULL); crm_xml_add_int(attr, XML_NVPAIR_ATTR_VALUE, start_delay); } else if (pcmk__str_eq(name, XML_OP_ATTR_ORIGIN, pcmk__str_casei)) { long long start_delay = 0; if (unpack_interval_origin(value, xml_obj, interval_ms, now, &start_delay)) { crm_xml_add(attr, XML_NVPAIR_ATTR_NAME, XML_OP_ATTR_START_DELAY); crm_xml_add_ll(attr, XML_NVPAIR_ATTR_VALUE, start_delay); } } else if (pcmk__str_eq(name, XML_ATTR_TIMEOUT, pcmk__str_casei)) { int timeout_ms = unpack_timeout(value); crm_xml_add_int(attr, XML_NVPAIR_ATTR_VALUE, timeout_ms); } } } } #endif /*! * \brief Unpack operation XML into an action structure * * Unpack an operation's meta-attributes (normalizing the interval, timeout, * and start delay values as integer milliseconds), requirements, and * failure policy. * * \param[in,out] action Action to unpack into * \param[in] xml_obj Operation XML (or NULL if all defaults) * \param[in] container Resource that contains affected resource, if any * \param[in] data_set Cluster state * \param[in] interval_ms How frequently to perform the operation */ static void unpack_operation(pe_action_t * action, xmlNode * xml_obj, pe_resource_t * container, pe_working_set_t * data_set, guint interval_ms) { int timeout_ms = 0; const char *value = NULL; bool is_probe = pcmk__str_eq(action->task, RSC_STATUS, pcmk__str_casei) && (interval_ms == 0); #if ENABLE_VERSIONED_ATTRS pe_rsc_action_details_t *rsc_details = NULL; #endif pe_rsc_eval_data_t rsc_rule_data = { .standard = crm_element_value(action->rsc->xml, XML_AGENT_ATTR_CLASS), .provider = crm_element_value(action->rsc->xml, XML_AGENT_ATTR_PROVIDER), .agent = crm_element_value(action->rsc->xml, XML_EXPR_ATTR_TYPE) }; pe_op_eval_data_t op_rule_data = { .op_name = action->task, .interval = interval_ms }; pe_rule_eval_data_t rule_data = { .node_hash = NULL, .role = RSC_ROLE_UNKNOWN, .now = data_set->now, .match_data = NULL, .rsc_data = &rsc_rule_data, .op_data = &op_rule_data }; CRM_CHECK(action && action->rsc, return); // Cluster-wide pe__unpack_dataset_nvpairs(data_set->op_defaults, XML_TAG_META_SETS, &rule_data, action->meta, NULL, FALSE, data_set); // Determine probe default timeout differently if (is_probe) { xmlNode *min_interval_mon = find_min_interval_mon(action->rsc, FALSE); if (min_interval_mon) { value = crm_element_value(min_interval_mon, XML_ATTR_TIMEOUT); if (value) { crm_trace("\t%s: Setting default timeout to minimum-interval " "monitor's timeout '%s'", action->uuid, value); g_hash_table_replace(action->meta, strdup(XML_ATTR_TIMEOUT), strdup(value)); } } } if (xml_obj) { xmlAttrPtr xIter = NULL; // take precedence over defaults pe__unpack_dataset_nvpairs(xml_obj, XML_TAG_META_SETS, &rule_data, action->meta, NULL, TRUE, data_set); #if ENABLE_VERSIONED_ATTRS rsc_details = pe_rsc_action_details(action); pe_eval_versioned_attributes(data_set->input, xml_obj, XML_TAG_ATTR_SETS, &rule_data, rsc_details->versioned_parameters, NULL); pe_eval_versioned_attributes(data_set->input, xml_obj, XML_TAG_META_SETS, &rule_data, rsc_details->versioned_meta, NULL); #endif /* Anything set as an XML property has highest precedence. * This ensures we use the name and interval from the tag. */ for (xIter = xml_obj->properties; xIter; xIter = xIter->next) { const char *prop_name = (const char *)xIter->name; const char *prop_value = crm_element_value(xml_obj, prop_name); g_hash_table_replace(action->meta, strdup(prop_name), strdup(prop_value)); } } g_hash_table_remove(action->meta, "id"); // Normalize interval to milliseconds if (interval_ms > 0) { g_hash_table_replace(action->meta, strdup(XML_LRM_ATTR_INTERVAL), crm_strdup_printf("%u", interval_ms)); } else { g_hash_table_remove(action->meta, XML_LRM_ATTR_INTERVAL); } /* * Timeout order of precedence: * 1. pcmk_monitor_timeout (if rsc has pcmk_ra_cap_fence_params * and task is start or a probe; pcmk_monitor_timeout works * by default for a recurring monitor) * 2. explicit op timeout on the primitive * 3. default op timeout * a. if probe, then min-interval monitor's timeout * b. else, in XML_CIB_TAG_OPCONFIG * 4. CRM_DEFAULT_OP_TIMEOUT_S * * #1 overrides general rule of XML property having highest * precedence. */ if (pcmk_is_set(pcmk_get_ra_caps(rsc_rule_data.standard), pcmk_ra_cap_fence_params) && (pcmk__str_eq(action->task, RSC_START, pcmk__str_casei) || is_probe)) { GHashTable *params = pe_rsc_params(action->rsc, action->node, data_set); value = g_hash_table_lookup(params, "pcmk_monitor_timeout"); if (value) { crm_trace("\t%s: Setting timeout to pcmk_monitor_timeout '%s', " "overriding default", action->uuid, value); g_hash_table_replace(action->meta, strdup(XML_ATTR_TIMEOUT), strdup(value)); } } // Normalize timeout to positive milliseconds value = g_hash_table_lookup(action->meta, XML_ATTR_TIMEOUT); timeout_ms = unpack_timeout(value); g_hash_table_replace(action->meta, strdup(XML_ATTR_TIMEOUT), pcmk__itoa(timeout_ms)); if (!pcmk__strcase_any_of(action->task, RSC_START, RSC_PROMOTE, NULL)) { action->needs = rsc_req_nothing; value = "nothing (not start or promote)"; } else if (pcmk_is_set(action->rsc->flags, pe_rsc_needs_fencing)) { action->needs = rsc_req_stonith; value = "fencing"; } else if (pcmk_is_set(action->rsc->flags, pe_rsc_needs_quorum)) { action->needs = rsc_req_quorum; value = "quorum"; } else { action->needs = rsc_req_nothing; value = "nothing"; } pe_rsc_trace(action->rsc, "%s requires %s", action->uuid, value); value = unpack_operation_on_fail(action); if (value == NULL) { } else if (pcmk__str_eq(value, "block", pcmk__str_casei)) { action->on_fail = action_fail_block; g_hash_table_insert(action->meta, strdup(XML_OP_ATTR_ON_FAIL), strdup("block")); value = "block"; // The above could destroy the original string } else if (pcmk__str_eq(value, "fence", pcmk__str_casei)) { action->on_fail = action_fail_fence; value = "node fencing"; if (!pcmk_is_set(data_set->flags, pe_flag_stonith_enabled)) { pcmk__config_err("Resetting '" XML_OP_ATTR_ON_FAIL "' for " "operation '%s' to 'stop' because 'fence' is not " "valid when fencing is disabled", action->uuid); action->on_fail = action_fail_stop; action->fail_role = RSC_ROLE_STOPPED; value = "stop resource"; } } else if (pcmk__str_eq(value, "standby", pcmk__str_casei)) { action->on_fail = action_fail_standby; value = "node standby"; } else if (pcmk__strcase_any_of(value, "ignore", "nothing", NULL)) { action->on_fail = action_fail_ignore; value = "ignore"; } else if (pcmk__str_eq(value, "migrate", pcmk__str_casei)) { action->on_fail = action_fail_migrate; value = "force migration"; } else if (pcmk__str_eq(value, "stop", pcmk__str_casei)) { action->on_fail = action_fail_stop; action->fail_role = RSC_ROLE_STOPPED; value = "stop resource"; } else if (pcmk__str_eq(value, "restart", pcmk__str_casei)) { action->on_fail = action_fail_recover; value = "restart (and possibly migrate)"; } else if (pcmk__str_eq(value, "restart-container", pcmk__str_casei)) { if (container) { action->on_fail = action_fail_restart_container; value = "restart container (and possibly migrate)"; } else { value = NULL; } } else if (pcmk__str_eq(value, "demote", pcmk__str_casei)) { action->on_fail = action_fail_demote; value = "demote instance"; } else { pe_err("Resource %s: Unknown failure type (%s)", action->rsc->id, value); value = NULL; } /* defaults */ if (value == NULL && container) { action->on_fail = action_fail_restart_container; value = "restart container (and possibly migrate) (default)"; /* For remote nodes, ensure that any failure that results in dropping an * active connection to the node results in fencing of the node. * * There are only two action failures that don't result in fencing. * 1. probes - probe failures are expected. * 2. start - a start failure indicates that an active connection does not already * exist. The user can set op on-fail=fence if they really want to fence start * failures. */ } else if (((value == NULL) || !pcmk_is_set(action->rsc->flags, pe_rsc_managed)) && pe__resource_is_remote_conn(action->rsc, data_set) && !(pcmk__str_eq(action->task, CRMD_ACTION_STATUS, pcmk__str_casei) && (interval_ms == 0)) && !pcmk__str_eq(action->task, CRMD_ACTION_START, pcmk__str_casei)) { if (!pcmk_is_set(action->rsc->flags, pe_rsc_managed)) { action->on_fail = action_fail_stop; action->fail_role = RSC_ROLE_STOPPED; value = "stop unmanaged remote node (enforcing default)"; } else { if (pcmk_is_set(data_set->flags, pe_flag_stonith_enabled)) { value = "fence remote node (default)"; } else { value = "recover remote node connection (default)"; } if (action->rsc->remote_reconnect_ms) { action->fail_role = RSC_ROLE_STOPPED; } action->on_fail = action_fail_reset_remote; } } else if (value == NULL && pcmk__str_eq(action->task, CRMD_ACTION_STOP, pcmk__str_casei)) { if (pcmk_is_set(data_set->flags, pe_flag_stonith_enabled)) { action->on_fail = action_fail_fence; value = "resource fence (default)"; } else { action->on_fail = action_fail_block; value = "resource block (default)"; } } else if (value == NULL) { action->on_fail = action_fail_recover; value = "restart (and possibly migrate) (default)"; } pe_rsc_trace(action->rsc, "%s failure handling: %s", action->uuid, value); value = NULL; if (xml_obj != NULL) { value = g_hash_table_lookup(action->meta, "role_after_failure"); if (value) { pe_warn_once(pe_wo_role_after, "Support for role_after_failure is deprecated and will be removed in a future release"); } } if (value != NULL && action->fail_role == RSC_ROLE_UNKNOWN) { action->fail_role = text2role(value); } /* defaults */ if (action->fail_role == RSC_ROLE_UNKNOWN) { if (pcmk__str_eq(action->task, CRMD_ACTION_PROMOTE, pcmk__str_casei)) { action->fail_role = RSC_ROLE_UNPROMOTED; } else { action->fail_role = RSC_ROLE_STARTED; } } pe_rsc_trace(action->rsc, "%s failure results in: %s", action->uuid, role2text(action->fail_role)); value = g_hash_table_lookup(action->meta, XML_OP_ATTR_START_DELAY); if (value) { unpack_start_delay(value, action->meta); } else { long long start_delay = 0; value = g_hash_table_lookup(action->meta, XML_OP_ATTR_ORIGIN); if (unpack_interval_origin(value, xml_obj, interval_ms, data_set->now, &start_delay)) { g_hash_table_replace(action->meta, strdup(XML_OP_ATTR_START_DELAY), crm_strdup_printf("%lld", start_delay)); } } #if ENABLE_VERSIONED_ATTRS unpack_versioned_meta(rsc_details->versioned_meta, xml_obj, interval_ms, data_set->now); #endif } static xmlNode * find_rsc_op_entry_helper(pe_resource_t * rsc, const char *key, gboolean include_disabled) { guint interval_ms = 0; gboolean do_retry = TRUE; char *local_key = NULL; const char *name = NULL; const char *value = NULL; const char *interval_spec = NULL; char *match_key = NULL; xmlNode *op = NULL; xmlNode *operation = NULL; retry: for (operation = pcmk__xe_first_child(rsc->ops_xml); operation != NULL; operation = pcmk__xe_next(operation)) { if (pcmk__str_eq((const char *)operation->name, "op", pcmk__str_none)) { name = crm_element_value(operation, "name"); interval_spec = crm_element_value(operation, XML_LRM_ATTR_INTERVAL); value = crm_element_value(operation, "enabled"); if (!include_disabled && value && crm_is_true(value) == FALSE) { continue; } interval_ms = crm_parse_interval_spec(interval_spec); match_key = pcmk__op_key(rsc->id, name, interval_ms); if (pcmk__str_eq(key, match_key, pcmk__str_casei)) { op = operation; } free(match_key); if (rsc->clone_name) { match_key = pcmk__op_key(rsc->clone_name, name, interval_ms); if (pcmk__str_eq(key, match_key, pcmk__str_casei)) { op = operation; } free(match_key); } if (op != NULL) { free(local_key); return op; } } } free(local_key); if (do_retry == FALSE) { return NULL; } do_retry = FALSE; if (strstr(key, CRMD_ACTION_MIGRATE) || strstr(key, CRMD_ACTION_MIGRATED)) { local_key = pcmk__op_key(rsc->id, "migrate", 0); key = local_key; goto retry; } else if (strstr(key, "_notify_")) { local_key = pcmk__op_key(rsc->id, "notify", 0); key = local_key; goto retry; } return NULL; } xmlNode * find_rsc_op_entry(pe_resource_t * rsc, const char *key) { return find_rsc_op_entry_helper(rsc, key, FALSE); } /* * Used by the HashTable for-loop */ void print_str_str(gpointer key, gpointer value, gpointer user_data) { crm_trace("%s%s %s ==> %s", user_data == NULL ? "" : (char *)user_data, user_data == NULL ? "" : ": ", (char *)key, (char *)value); } void pe_free_action(pe_action_t * action) { if (action == NULL) { return; } g_list_free_full(action->actions_before, free); /* pe_action_wrapper_t* */ g_list_free_full(action->actions_after, free); /* pe_action_wrapper_t* */ if (action->extra) { g_hash_table_destroy(action->extra); } if (action->meta) { g_hash_table_destroy(action->meta); } #if ENABLE_VERSIONED_ATTRS if (action->rsc) { pe_free_rsc_action_details(action); } #endif free(action->cancel_task); free(action->reason); free(action->task); free(action->uuid); free(action->node); free(action); } GList * find_recurring_actions(GList *input, pe_node_t * not_on_node) { const char *value = NULL; GList *result = NULL; GList *gIter = input; CRM_CHECK(input != NULL, return NULL); for (; gIter != NULL; gIter = gIter->next) { pe_action_t *action = (pe_action_t *) gIter->data; value = g_hash_table_lookup(action->meta, XML_LRM_ATTR_INTERVAL_MS); if (value == NULL) { /* skip */ } else if (pcmk__str_eq(value, "0", pcmk__str_casei)) { /* skip */ } else if (pcmk__str_eq(CRMD_ACTION_CANCEL, action->task, pcmk__str_casei)) { /* skip */ } else if (not_on_node == NULL) { crm_trace("(null) Found: %s", action->uuid); result = g_list_prepend(result, action); } else if (action->node == NULL) { /* skip */ } else if (action->node->details != not_on_node->details) { crm_trace("Found: %s", action->uuid); result = g_list_prepend(result, action); } } return result; } enum action_tasks get_complex_task(pe_resource_t * rsc, const char *name, gboolean allow_non_atomic) { enum action_tasks task = text2task(name); if (rsc == NULL) { return task; } else if (allow_non_atomic == FALSE || rsc->variant == pe_native) { switch (task) { case stopped_rsc: case started_rsc: case action_demoted: case action_promoted: crm_trace("Folding %s back into its atomic counterpart for %s", name, rsc->id); return task - 1; default: break; } } return task; } pe_action_t * find_first_action(GList *input, const char *uuid, const char *task, pe_node_t * on_node) { GList *gIter = NULL; CRM_CHECK(uuid || task, return NULL); for (gIter = input; gIter != NULL; gIter = gIter->next) { pe_action_t *action = (pe_action_t *) gIter->data; if (uuid != NULL && !pcmk__str_eq(uuid, action->uuid, pcmk__str_casei)) { continue; } else if (task != NULL && !pcmk__str_eq(task, action->task, pcmk__str_casei)) { continue; } else if (on_node == NULL) { return action; } else if (action->node == NULL) { continue; } else if (on_node->details == action->node->details) { return action; } } return NULL; } GList * find_actions(GList *input, const char *key, const pe_node_t *on_node) { GList *gIter = input; GList *result = NULL; CRM_CHECK(key != NULL, return NULL); for (; gIter != NULL; gIter = gIter->next) { pe_action_t *action = (pe_action_t *) gIter->data; if (!pcmk__str_eq(key, action->uuid, pcmk__str_casei)) { crm_trace("%s does not match action %s", key, action->uuid); continue; } else if (on_node == NULL) { crm_trace("Action %s matches (ignoring node)", key); result = g_list_prepend(result, action); } else if (action->node == NULL) { crm_trace("Action %s matches (unallocated, assigning to %s)", key, on_node->details->uname); action->node = pe__copy_node(on_node); result = g_list_prepend(result, action); } else if (on_node->details == action->node->details) { crm_trace("Action %s on %s matches", key, on_node->details->uname); result = g_list_prepend(result, action); } else { crm_trace("Action %s on node %s does not match requested node %s", key, action->node->details->uname, on_node->details->uname); } } return result; } GList * find_actions_exact(GList *input, const char *key, const pe_node_t *on_node) { GList *result = NULL; CRM_CHECK(key != NULL, return NULL); if (on_node == NULL) { crm_trace("Not searching for action %s because node not specified", key); return NULL; } for (GList *gIter = input; gIter != NULL; gIter = gIter->next) { pe_action_t *action = (pe_action_t *) gIter->data; if (action->node == NULL) { crm_trace("Skipping comparison of %s vs action %s without node", key, action->uuid); } else if (!pcmk__str_eq(key, action->uuid, pcmk__str_casei)) { crm_trace("Desired action %s doesn't match %s", key, action->uuid); } else if (!pcmk__str_eq(on_node->details->id, action->node->details->id, pcmk__str_casei)) { crm_trace("Action %s desired node ID %s doesn't match %s", key, on_node->details->id, action->node->details->id); } else { crm_trace("Action %s matches", key); result = g_list_prepend(result, action); } } return result; } /*! * \brief Find all actions of given type for a resource * * \param[in] rsc Resource to search * \param[in] node Find only actions scheduled on this node * \param[in] task Action name to search for * \param[in] require_node If TRUE, NULL node or action node will not match * * \return List of actions found (or NULL if none) * \note If node is not NULL and require_node is FALSE, matching actions * without a node will be assigned to node. */ GList * pe__resource_actions(const pe_resource_t *rsc, const pe_node_t *node, const char *task, bool require_node) { GList *result = NULL; char *key = pcmk__op_key(rsc->id, task, 0); if (require_node) { result = find_actions_exact(rsc->actions, key, node); } else { result = find_actions(rsc->actions, key, node); } free(key); return result; } static void resource_node_score(pe_resource_t * rsc, pe_node_t * node, int score, const char *tag) { pe_node_t *match = NULL; if ((rsc->exclusive_discover || (node->rsc_discover_mode == pe_discover_never)) && pcmk__str_eq(tag, "symmetric_default", pcmk__str_casei)) { /* This string comparision may be fragile, but exclusive resources and * exclusive nodes should not have the symmetric_default constraint * applied to them. */ return; } else if (rsc->children) { GList *gIter = rsc->children; for (; gIter != NULL; gIter = gIter->next) { pe_resource_t *child_rsc = (pe_resource_t *) gIter->data; resource_node_score(child_rsc, node, score, tag); } } pe_rsc_trace(rsc, "Setting %s for %s on %s: %d", tag, rsc->id, node->details->uname, score); match = pe_hash_table_lookup(rsc->allowed_nodes, node->details->id); if (match == NULL) { match = pe__copy_node(node); g_hash_table_insert(rsc->allowed_nodes, (gpointer) match->details->id, match); } match->weight = pe__add_scores(match->weight, score); } void resource_location(pe_resource_t * rsc, pe_node_t * node, int score, const char *tag, pe_working_set_t * data_set) { if (node != NULL) { resource_node_score(rsc, node, score, tag); } else if (data_set != NULL) { GList *gIter = data_set->nodes; for (; gIter != NULL; gIter = gIter->next) { pe_node_t *node_iter = (pe_node_t *) gIter->data; resource_node_score(rsc, node_iter, score, tag); } } else { GHashTableIter iter; pe_node_t *node_iter = NULL; g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node_iter)) { resource_node_score(rsc, node_iter, score, tag); } } if (node == NULL && score == -INFINITY) { if (rsc->allocated_to) { crm_info("Deallocating %s from %s", rsc->id, rsc->allocated_to->details->uname); free(rsc->allocated_to); rsc->allocated_to = NULL; } } } #define sort_return(an_int, why) do { \ free(a_uuid); \ free(b_uuid); \ crm_trace("%s (%d) %c %s (%d) : %s", \ a_xml_id, a_call_id, an_int>0?'>':an_int<0?'<':'=', \ b_xml_id, b_call_id, why); \ return an_int; \ } while(0) gint sort_op_by_callid(gconstpointer a, gconstpointer b) { int a_call_id = -1; int b_call_id = -1; char *a_uuid = NULL; char *b_uuid = NULL; const xmlNode *xml_a = a; const xmlNode *xml_b = b; const char *a_xml_id = crm_element_value(xml_a, XML_ATTR_ID); const char *b_xml_id = crm_element_value(xml_b, XML_ATTR_ID); if (pcmk__str_eq(a_xml_id, b_xml_id, pcmk__str_casei)) { /* We have duplicate lrm_rsc_op entries in the status * section which is unlikely to be a good thing * - we can handle it easily enough, but we need to get * to the bottom of why it's happening. */ pe_err("Duplicate lrm_rsc_op entries named %s", a_xml_id); sort_return(0, "duplicate"); } crm_element_value_int(xml_a, XML_LRM_ATTR_CALLID, &a_call_id); crm_element_value_int(xml_b, XML_LRM_ATTR_CALLID, &b_call_id); if (a_call_id == -1 && b_call_id == -1) { /* both are pending ops so it doesn't matter since * stops are never pending */ sort_return(0, "pending"); } else if (a_call_id >= 0 && a_call_id < b_call_id) { sort_return(-1, "call id"); } else if (b_call_id >= 0 && a_call_id > b_call_id) { sort_return(1, "call id"); } else if (b_call_id >= 0 && a_call_id == b_call_id) { /* * The op and last_failed_op are the same * Order on last-rc-change */ time_t last_a = -1; time_t last_b = -1; crm_element_value_epoch(xml_a, XML_RSC_OP_LAST_CHANGE, &last_a); crm_element_value_epoch(xml_b, XML_RSC_OP_LAST_CHANGE, &last_b); crm_trace("rc-change: %lld vs %lld", (long long) last_a, (long long) last_b); if (last_a >= 0 && last_a < last_b) { sort_return(-1, "rc-change"); } else if (last_b >= 0 && last_a > last_b) { sort_return(1, "rc-change"); } sort_return(0, "rc-change"); } else { /* One of the inputs is a pending operation * Attempt to use XML_ATTR_TRANSITION_MAGIC to determine its age relative to the other */ int a_id = -1; int b_id = -1; const char *a_magic = crm_element_value(xml_a, XML_ATTR_TRANSITION_MAGIC); const char *b_magic = crm_element_value(xml_b, XML_ATTR_TRANSITION_MAGIC); CRM_CHECK(a_magic != NULL && b_magic != NULL, sort_return(0, "No magic")); if (!decode_transition_magic(a_magic, &a_uuid, &a_id, NULL, NULL, NULL, NULL)) { sort_return(0, "bad magic a"); } if (!decode_transition_magic(b_magic, &b_uuid, &b_id, NULL, NULL, NULL, NULL)) { sort_return(0, "bad magic b"); } /* try to determine the relative age of the operation... * some pending operations (e.g. a start) may have been superseded * by a subsequent stop * * [a|b]_id == -1 means it's a shutdown operation and _always_ comes last */ if (!pcmk__str_eq(a_uuid, b_uuid, pcmk__str_casei) || a_id == b_id) { /* * some of the logic in here may be redundant... * * if the UUID from the TE doesn't match then one better * be a pending operation. * pending operations don't survive between elections and joins * because we query the LRM directly */ if (b_call_id == -1) { sort_return(-1, "transition + call"); } else if (a_call_id == -1) { sort_return(1, "transition + call"); } } else if ((a_id >= 0 && a_id < b_id) || b_id == -1) { sort_return(-1, "transition"); } else if ((b_id >= 0 && a_id > b_id) || a_id == -1) { sort_return(1, "transition"); } } /* we should never end up here */ CRM_CHECK(FALSE, sort_return(0, "default")); } time_t get_effective_time(pe_working_set_t * data_set) { if(data_set) { if (data_set->now == NULL) { crm_trace("Recording a new 'now'"); data_set->now = crm_time_new(NULL); } return crm_time_get_seconds_since_epoch(data_set->now); } crm_trace("Defaulting to 'now'"); return time(NULL); } gboolean get_target_role(pe_resource_t * rsc, enum rsc_role_e * role) { enum rsc_role_e local_role = RSC_ROLE_UNKNOWN; const char *value = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_TARGET_ROLE); CRM_CHECK(role != NULL, return FALSE); if (pcmk__str_eq(value, "started", pcmk__str_null_matches | pcmk__str_casei) || pcmk__str_eq("default", value, pcmk__str_casei)) { return FALSE; } local_role = text2role(value); if (local_role == RSC_ROLE_UNKNOWN) { pcmk__config_err("Ignoring '" XML_RSC_ATTR_TARGET_ROLE "' for %s " "because '%s' is not valid", rsc->id, value); return FALSE; } else if (local_role > RSC_ROLE_STARTED) { if (pcmk_is_set(uber_parent(rsc)->flags, pe_rsc_promotable)) { if (local_role > RSC_ROLE_UNPROMOTED) { /* This is what we'd do anyway, just leave the default to avoid messing up the placement algorithm */ return FALSE; } } else { pcmk__config_err("Ignoring '" XML_RSC_ATTR_TARGET_ROLE "' for %s " "because '%s' only makes sense for promotable " "clones", rsc->id, value); return FALSE; } } *role = local_role; return TRUE; } gboolean order_actions(pe_action_t * lh_action, pe_action_t * rh_action, enum pe_ordering order) { GList *gIter = NULL; pe_action_wrapper_t *wrapper = NULL; GList *list = NULL; if (order == pe_order_none) { return FALSE; } if (lh_action == NULL || rh_action == NULL) { return FALSE; } crm_trace("Ordering Action %s before %s", lh_action->uuid, rh_action->uuid); /* Ensure we never create a dependency on ourselves... it's happened */ CRM_ASSERT(lh_action != rh_action); /* Filter dups, otherwise update_action_states() has too much work to do */ gIter = lh_action->actions_after; for (; gIter != NULL; gIter = gIter->next) { pe_action_wrapper_t *after = (pe_action_wrapper_t *) gIter->data; if (after->action == rh_action && (after->type & order)) { return FALSE; } } wrapper = calloc(1, sizeof(pe_action_wrapper_t)); wrapper->action = rh_action; wrapper->type = order; list = lh_action->actions_after; list = g_list_prepend(list, wrapper); lh_action->actions_after = list; wrapper = calloc(1, sizeof(pe_action_wrapper_t)); wrapper->action = lh_action; wrapper->type = order; list = rh_action->actions_before; list = g_list_prepend(list, wrapper); rh_action->actions_before = list; return TRUE; } pe_action_t * get_pseudo_op(const char *name, pe_working_set_t * data_set) { pe_action_t *op = NULL; if(data_set->singletons) { op = g_hash_table_lookup(data_set->singletons, name); } if (op == NULL) { op = custom_action(NULL, strdup(name), name, NULL, TRUE, TRUE, data_set); pe__set_action_flags(op, pe_action_pseudo|pe_action_runnable); } return op; } void destroy_ticket(gpointer data) { pe_ticket_t *ticket = data; if (ticket->state) { g_hash_table_destroy(ticket->state); } free(ticket->id); free(ticket); } pe_ticket_t * ticket_new(const char *ticket_id, pe_working_set_t * data_set) { pe_ticket_t *ticket = NULL; if (pcmk__str_empty(ticket_id)) { return NULL; } if (data_set->tickets == NULL) { data_set->tickets = pcmk__strkey_table(free, destroy_ticket); } ticket = g_hash_table_lookup(data_set->tickets, ticket_id); if (ticket == NULL) { ticket = calloc(1, sizeof(pe_ticket_t)); if (ticket == NULL) { crm_err("Cannot allocate ticket '%s'", ticket_id); return NULL; } crm_trace("Creaing ticket entry for %s", ticket_id); ticket->id = strdup(ticket_id); ticket->granted = FALSE; ticket->last_granted = -1; ticket->standby = FALSE; ticket->state = pcmk__strkey_table(free, free); g_hash_table_insert(data_set->tickets, strdup(ticket->id), ticket); } return ticket; } const char *rsc_printable_id(pe_resource_t *rsc) { if (!pcmk_is_set(rsc->flags, pe_rsc_unique)) { return ID(rsc->xml); } return rsc->id; } void pe__clear_resource_flags_recursive(pe_resource_t *rsc, uint64_t flags) { pe__clear_resource_flags(rsc, flags); for (GList *gIter = rsc->children; gIter != NULL; gIter = gIter->next) { pe__clear_resource_flags_recursive((pe_resource_t *) gIter->data, flags); } } void pe__clear_resource_flags_on_all(pe_working_set_t *data_set, uint64_t flag) { for (GList *lpc = data_set->resources; lpc != NULL; lpc = lpc->next) { pe_resource_t *r = (pe_resource_t *) lpc->data; pe__clear_resource_flags_recursive(r, flag); } } void pe__set_resource_flags_recursive(pe_resource_t *rsc, uint64_t flags) { pe__set_resource_flags(rsc, flags); for (GList *gIter = rsc->children; gIter != NULL; gIter = gIter->next) { pe__set_resource_flags_recursive((pe_resource_t *) gIter->data, flags); } } static GList * find_unfencing_devices(GList *candidates, GList *matches) { for (GList *gIter = candidates; gIter != NULL; gIter = gIter->next) { pe_resource_t *candidate = gIter->data; const char *provides = g_hash_table_lookup(candidate->meta, PCMK_STONITH_PROVIDES); const char *requires = g_hash_table_lookup(candidate->meta, XML_RSC_ATTR_REQUIRES); if(candidate->children) { matches = find_unfencing_devices(candidate->children, matches); } else if (!pcmk_is_set(candidate->flags, pe_rsc_fence_device)) { continue; } else if (pcmk__str_eq(provides, "unfencing", pcmk__str_casei) || pcmk__str_eq(requires, "unfencing", pcmk__str_casei)) { matches = g_list_prepend(matches, candidate); } } return matches; } static int node_priority_fencing_delay(pe_node_t * node, pe_working_set_t * data_set) { int member_count = 0; int online_count = 0; int top_priority = 0; int lowest_priority = 0; GList *gIter = NULL; // `priority-fencing-delay` is disabled if (data_set->priority_fencing_delay <= 0) { return 0; } /* No need to request a delay if the fencing target is not a normal cluster * member, for example if it's a remote node or a guest node. */ if (node->details->type != node_member) { return 0; } // No need to request a delay if the fencing target is in our partition if (node->details->online) { return 0; } for (gIter = data_set->nodes; gIter != NULL; gIter = gIter->next) { pe_node_t *n = gIter->data; if (n->details->type != node_member) { continue; } member_count ++; if (n->details->online) { online_count++; } if (member_count == 1 || n->details->priority > top_priority) { top_priority = n->details->priority; } if (member_count == 1 || n->details->priority < lowest_priority) { lowest_priority = n->details->priority; } } // No need to delay if we have more than half of the cluster members if (online_count > member_count / 2) { return 0; } /* All the nodes have equal priority. * Any configured corresponding `pcmk_delay_base/max` will be applied. */ if (lowest_priority == top_priority) { return 0; } if (node->details->priority < top_priority) { return 0; } return data_set->priority_fencing_delay; } pe_action_t * pe_fence_op(pe_node_t * node, const char *op, bool optional, const char *reason, bool priority_delay, pe_working_set_t * data_set) { char *op_key = NULL; pe_action_t *stonith_op = NULL; if(op == NULL) { op = data_set->stonith_action; } op_key = crm_strdup_printf("%s-%s-%s", CRM_OP_FENCE, node->details->uname, op); if(data_set->singletons) { stonith_op = g_hash_table_lookup(data_set->singletons, op_key); } if(stonith_op == NULL) { stonith_op = custom_action(NULL, op_key, CRM_OP_FENCE, node, TRUE, TRUE, data_set); add_hash_param(stonith_op->meta, XML_LRM_ATTR_TARGET, node->details->uname); add_hash_param(stonith_op->meta, XML_LRM_ATTR_TARGET_UUID, node->details->id); add_hash_param(stonith_op->meta, "stonith_action", op); if (pe__is_guest_or_remote_node(node) && pcmk_is_set(data_set->flags, pe_flag_enable_unfencing)) { /* Extra work to detect device changes on remotes * * We may do this for all nodes in the future, but for now * the check_action_definition() based stuff works fine. */ long max = 1024; long digests_all_offset = 0; long digests_secure_offset = 0; char *digests_all = calloc(max, sizeof(char)); char *digests_secure = calloc(max, sizeof(char)); GList *matches = find_unfencing_devices(data_set->resources, NULL); for (GList *gIter = matches; gIter != NULL; gIter = gIter->next) { pe_resource_t *match = gIter->data; const char *agent = g_hash_table_lookup(match->meta, XML_ATTR_TYPE); op_digest_cache_t *data = NULL; data = pe__compare_fencing_digest(match, agent, node, data_set); if(data->rc == RSC_DIGEST_ALL) { optional = FALSE; crm_notice("Unfencing %s (remote): because the definition of %s changed", node->details->uname, match->id); if (!pcmk__is_daemon && data_set->priv != NULL) { pcmk__output_t *out = data_set->priv; out->info(out, "notice: Unfencing %s (remote): because the definition of %s changed", node->details->uname, match->id); } } digests_all_offset += snprintf( digests_all+digests_all_offset, max-digests_all_offset, "%s:%s:%s,", match->id, agent, data->digest_all_calc); digests_secure_offset += snprintf( digests_secure+digests_secure_offset, max-digests_secure_offset, "%s:%s:%s,", match->id, agent, data->digest_secure_calc); } g_hash_table_insert(stonith_op->meta, strdup(XML_OP_ATTR_DIGESTS_ALL), digests_all); g_hash_table_insert(stonith_op->meta, strdup(XML_OP_ATTR_DIGESTS_SECURE), digests_secure); } } else { free(op_key); } if (data_set->priority_fencing_delay > 0 /* It's a suitable case where `priority-fencing-delay` applies. * At least add `priority-fencing-delay` field as an indicator. */ && (priority_delay /* Re-calculate priority delay for the suitable case when * pe_fence_op() is called again by stage6() after node priority has * been actually calculated with native_add_running() */ || g_hash_table_lookup(stonith_op->meta, XML_CONFIG_ATTR_PRIORITY_FENCING_DELAY) != NULL)) { /* Add `priority-fencing-delay` to the fencing op even if it's 0 for * the targeting node. So that it takes precedence over any possible * `pcmk_delay_base/max`. */ char *delay_s = pcmk__itoa(node_priority_fencing_delay(node, data_set)); g_hash_table_insert(stonith_op->meta, strdup(XML_CONFIG_ATTR_PRIORITY_FENCING_DELAY), delay_s); } if(optional == FALSE && pe_can_fence(data_set, node)) { pe_action_required(stonith_op, NULL, reason); } else if(reason && stonith_op->reason == NULL) { stonith_op->reason = strdup(reason); } return stonith_op; } void trigger_unfencing( pe_resource_t * rsc, pe_node_t *node, const char *reason, pe_action_t *dependency, pe_working_set_t * data_set) { if (!pcmk_is_set(data_set->flags, pe_flag_enable_unfencing)) { /* No resources require it */ return; } else if ((rsc != NULL) && !pcmk_is_set(rsc->flags, pe_rsc_fence_device)) { /* Wasn't a stonith device */ return; } else if(node && node->details->online && node->details->unclean == FALSE && node->details->shutdown == FALSE) { pe_action_t *unfence = pe_fence_op(node, "on", FALSE, reason, FALSE, data_set); if(dependency) { order_actions(unfence, dependency, pe_order_optional); } } else if(rsc) { GHashTableIter iter; g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { if(node->details->online && node->details->unclean == FALSE && node->details->shutdown == FALSE) { trigger_unfencing(rsc, node, reason, dependency, data_set); } } } } gboolean add_tag_ref(GHashTable * tags, const char * tag_name, const char * obj_ref) { pe_tag_t *tag = NULL; GList *gIter = NULL; gboolean is_existing = FALSE; CRM_CHECK(tags && tag_name && obj_ref, return FALSE); tag = g_hash_table_lookup(tags, tag_name); if (tag == NULL) { tag = calloc(1, sizeof(pe_tag_t)); if (tag == NULL) { return FALSE; } tag->id = strdup(tag_name); tag->refs = NULL; g_hash_table_insert(tags, strdup(tag_name), tag); } for (gIter = tag->refs; gIter != NULL; gIter = gIter->next) { const char *existing_ref = (const char *) gIter->data; if (pcmk__str_eq(existing_ref, obj_ref, pcmk__str_none)){ is_existing = TRUE; break; } } if (is_existing == FALSE) { tag->refs = g_list_append(tag->refs, strdup(obj_ref)); crm_trace("Added: tag=%s ref=%s", tag->id, obj_ref); } return TRUE; } void pe_action_set_flag_reason(const char *function, long line, pe_action_t *action, pe_action_t *reason, const char *text, enum pe_action_flags flags, bool overwrite) { bool unset = FALSE; bool update = FALSE; const char *change = NULL; if (pcmk_is_set(flags, pe_action_runnable)) { unset = TRUE; change = "unrunnable"; } else if (pcmk_is_set(flags, pe_action_optional)) { unset = TRUE; change = "required"; } else if (pcmk_is_set(flags, pe_action_migrate_runnable)) { unset = TRUE; overwrite = TRUE; change = "unrunnable"; } else if (pcmk_is_set(flags, pe_action_dangle)) { change = "dangling"; } else if (pcmk_is_set(flags, pe_action_requires_any)) { change = "required"; } else { crm_err("Unknown flag change to %x by %s: 0x%s", flags, action->uuid, (reason? reason->uuid : "0")); } if(unset) { if (pcmk_is_set(action->flags, flags)) { pe__clear_action_flags_as(function, line, action, flags); update = TRUE; } } else { if (!pcmk_is_set(action->flags, flags)) { pe__set_action_flags_as(function, line, action, flags); update = TRUE; } } if((change && update) || text) { char *reason_text = NULL; if(reason == NULL) { pe_action_set_reason(action, text, overwrite); } else if(reason->rsc == NULL) { reason_text = crm_strdup_printf("%s %s%c %s", change, reason->task, text?':':0, text?text:""); } else { reason_text = crm_strdup_printf("%s %s %s%c %s", change, reason->rsc->id, reason->task, text?':':0, text?text:"NA"); } if(reason_text && action->rsc != reason->rsc) { pe_action_set_reason(action, reason_text, overwrite); } free(reason_text); } } void pe_action_set_reason(pe_action_t *action, const char *reason, bool overwrite) { if (action->reason != NULL && overwrite) { pe_rsc_trace(action->rsc, "Changing %s reason from '%s' to '%s'", action->uuid, action->reason, crm_str(reason)); free(action->reason); } else if (action->reason == NULL) { pe_rsc_trace(action->rsc, "Set %s reason to '%s'", action->uuid, crm_str(reason)); } else { // crm_assert(action->reason != NULL && !overwrite); return; } if (reason != NULL) { action->reason = strdup(reason); } else { action->reason = NULL; } } /*! * \internal * \brief Check whether shutdown has been requested for a node * * \param[in] node Node to check * * \return TRUE if node has shutdown attribute set and nonzero, FALSE otherwise * \note This differs from simply using node->details->shutdown in that it can * be used before that has been determined (and in fact to determine it), * and it can also be used to distinguish requested shutdown from implicit * shutdown of remote nodes by virtue of their connection stopping. */ bool pe__shutdown_requested(pe_node_t *node) { const char *shutdown = pe_node_attribute_raw(node, XML_CIB_ATTR_SHUTDOWN); return !pcmk__str_eq(shutdown, "0", pcmk__str_null_matches); } /*! * \internal * \brief Update a data set's "recheck by" time * * \param[in] recheck Epoch time when recheck should happen * \param[in,out] data_set Current working set */ void pe__update_recheck_time(time_t recheck, pe_working_set_t *data_set) { if ((recheck > get_effective_time(data_set)) && ((data_set->recheck_by == 0) || (data_set->recheck_by > recheck))) { data_set->recheck_by = recheck; } } /*! * \internal * \brief Wrapper for pe_unpack_nvpairs() using a cluster working set */ void pe__unpack_dataset_nvpairs(xmlNode *xml_obj, const char *set_name, pe_rule_eval_data_t *rule_data, GHashTable *hash, const char *always_first, gboolean overwrite, pe_working_set_t *data_set) { crm_time_t *next_change = crm_time_new_undefined(); pe_eval_nvpairs(data_set->input, xml_obj, set_name, rule_data, hash, always_first, overwrite, next_change); if (crm_time_is_defined(next_change)) { time_t recheck = (time_t) crm_time_get_seconds_since_epoch(next_change); pe__update_recheck_time(recheck, data_set); } crm_time_free(next_change); } bool pe__resource_is_disabled(pe_resource_t *rsc) { const char *target_role = NULL; CRM_CHECK(rsc != NULL, return false); target_role = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_TARGET_ROLE); if (target_role) { enum rsc_role_e target_role_e = text2role(target_role); if ((target_role_e == RSC_ROLE_STOPPED) || ((target_role_e == RSC_ROLE_UNPROMOTED) && pcmk_is_set(uber_parent(rsc)->flags, pe_rsc_promotable))) { return true; } } return false; } /*! * \internal * \brief Create an action to clear a resource's history from CIB * * \param[in] rsc Resource to clear * \param[in] node Node to clear history on * * \return New action to clear resource history */ pe_action_t * pe__clear_resource_history(pe_resource_t *rsc, pe_node_t *node, pe_working_set_t *data_set) { char *key = NULL; CRM_ASSERT(rsc && node); key = pcmk__op_key(rsc->id, CRM_OP_LRM_DELETE, 0); return custom_action(rsc, key, CRM_OP_LRM_DELETE, node, FALSE, TRUE, data_set); } bool pe__rsc_running_on_any(pe_resource_t *rsc, GList *node_list) { for (GList *ele = rsc->running_on; ele; ele = ele->next) { pe_node_t *node = (pe_node_t *) ele->data; if (pcmk__str_in_list(node_list, node->details->uname, pcmk__str_casei)) { return true; } } return false; } bool pcmk__rsc_filtered_by_node(pe_resource_t *rsc, GList *only_node) { return (rsc->fns->active(rsc, FALSE) && !pe__rsc_running_on_any(rsc, only_node)); } GList * pe__filter_rsc_list(GList *rscs, GList *filter) { GList *retval = NULL; for (GList *gIter = rscs; gIter; gIter = gIter->next) { pe_resource_t *rsc = (pe_resource_t *) gIter->data; /* I think the second condition is safe here for all callers of this * function. If not, it needs to move into pe__node_text. */ if (pcmk__str_in_list(filter, rsc_printable_id(rsc), pcmk__str_none) || (rsc->parent && pcmk__str_in_list(filter, rsc_printable_id(rsc->parent), pcmk__str_none))) { retval = g_list_prepend(retval, rsc); } } return retval; } GList * pe__build_node_name_list(pe_working_set_t *data_set, const char *s) { GList *nodes = NULL; if (pcmk__str_eq(s, "*", pcmk__str_null_matches)) { /* Nothing was given so return a list of all node names. Or, '*' was * given. This would normally fall into the pe__unames_with_tag branch * where it will return an empty list. Catch it here instead. */ nodes = g_list_prepend(nodes, strdup("*")); } else { pe_node_t *node = pe_find_node(data_set->nodes, s); if (node) { /* The given string was a valid uname for a node. Return a * singleton list containing just that uname. */ nodes = g_list_prepend(nodes, strdup(s)); } else { /* The given string was not a valid uname. It's either a tag or * it's a typo or something. In the first case, we'll return a * list of all the unames of the nodes with the given tag. In the * second case, we'll return a NULL pointer and nothing will * get displayed. */ nodes = pe__unames_with_tag(data_set, s); } } return nodes; } GList * pe__build_rsc_list(pe_working_set_t *data_set, const char *s) { GList *resources = NULL; if (pcmk__str_eq(s, "*", pcmk__str_null_matches)) { resources = g_list_prepend(resources, strdup("*")); } else { pe_resource_t *rsc = pe_find_resource_with_flags(data_set->resources, s, pe_find_renamed|pe_find_any); if (rsc) { /* A colon in the name we were given means we're being asked to filter * on a specific instance of a cloned resource. Put that exact string * into the filter list. Otherwise, use the printable ID of whatever * resource was found that matches what was asked for. */ if (strstr(s, ":") != NULL) { resources = g_list_prepend(resources, strdup(rsc->id)); } else { resources = g_list_prepend(resources, strdup(rsc_printable_id(rsc))); } } else { /* The given string was not a valid resource name. It's either * a tag or it's a typo or something. See build_uname_list for * more detail. */ resources = pe__rscs_with_tag(data_set, s); } } return resources; }