diff --git a/BasicSanity.sh b/BasicSanity.sh
index 6ae3e22acd..2485953ea3 100755
--- a/BasicSanity.sh
+++ b/BasicSanity.sh
@@ -1,79 +1,86 @@
#!/bin/bash
test_home=`dirname $0`
valgrind=""
verbose=""
tests=""
if [ "$test_home" = "." ]; then
test_home="$PWD"
fi
function info() {
printf "$*\n"
}
function error() {
printf " * ERROR: $*\n"
}
info "Test home is:\t$test_home"
while true ; do
case "$1" in
- all) tests="pengine lrmd fencing cli"; shift;;
- pengine|lrmd|fencing|cli) tests="$tests $1"; shift;;
- -V|--verbose) verbose="-V"; shift;;
- -v|--valgrind) valgrind="-v"; shift;;
- --) shift ; break ;;
- "") break;;
- *) echo "unknown option: $1"; exit 1;;
+ all) tests="pengine lrmd fencing cli"; shift;;
+ pengine|lrmd|pacemaker_remote|fencing|cli) tests="$tests $1"; shift;;
+ -V|--verbose) verbose="-V"; shift;;
+ -v|--valgrind) valgrind="-v"; shift;;
+ --) shift ; break ;;
+ "") break;;
+ *) echo "unknown option: $1"; exit 1;;
esac
done
if [ -z "$tests" ]; then
tests="pengine lrmd cli"
fi
failed=""
for t in $tests; do
info "Executing the $t regression tests"
info "============================================================"
if [ -e $test_home/$t/regression.py ]; then
- # Fencing, lrmd need root access
- chmod a+x $test_home/$t/regression.py
- echo "Enter the root password..."
- su root -c "$test_home/$t/regression.py $verbose"
- rc=$?
+ # Fencing, lrmd need root access
+ chmod a+x $test_home/$t/regression.py
+ echo "Enter the root password..."
+ su root -c "$test_home/$t/regression.py $verbose"
+ rc=$?
+
+ elif [ $t == "pacemaker_remote" ] && [ -e $test_home/lrmd/regression.py ]; then
+ # pacemaker_remote
+ chmod a+x $test_home/lrmd/regression.py
+ echo "Enter the root password..."
+ su root -c "$test_home/lrmd/regression.py -R $verbose"
+ rc=$?
elif [ -e $test_home/$t ]; then
- # pengine, cli
- $test_home/$t/regression.sh $verbose $valgrind
- rc=$?
+ # pengine, cli
+ $test_home/$t/regression.sh $verbose $valgrind
+ rc=$?
elif [ $t = cli -a -e $test_home/tools ]; then
- # Running cli tests from the source tree
- $test_home/tools/regression.sh $verbose $valgrind
- rc=$?
+ # Running cli tests from the source tree
+ $test_home/tools/regression.sh $verbose $valgrind
+ rc=$?
else
- error "Cannot find $t test in $test_home"
- rc=1
+ error "Cannot find $t test in $test_home"
+ rc=1
fi
if [ $rc != 0 ]; then
- info "$t regression tests failed: $rc"
- failed="$failed $t"
+ info "$t regression tests failed: $rc"
+ failed="$failed $t"
fi
info "============================================================"
info ""
info ""
done
if [ -z "$failed" ]; then
exit 0
fi
error "regression tests for $failed failed"
exit 1
diff --git a/fencing/regression.py.in b/fencing/regression.py.in
index 851ae17a74..6b203a288d 100644
--- a/fencing/regression.py.in
+++ b/fencing/regression.py.in
@@ -1,1029 +1,1029 @@
#!/usr/bin/python
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
import os
import sys
import subprocess
import shlex
import time
def output_from_command(command):
test = subprocess.Popen(shlex.split(command), stdout=subprocess.PIPE, stderr=subprocess.PIPE)
test.wait()
return test.communicate()[0].split("\n")
class Test:
def __init__(self, name, description, verbose = 0, with_cpg = 0):
self.name = name
self.description = description
self.cmds = []
self.verbose = verbose
self.result_txt = ""
self.cmd_tool_output = ""
self.result_exitcode = 0;
self.stonith_options = "-s"
self.enable_corosync = 0
if with_cpg:
self.stonith_options = "-c"
self.enable_corosync = 1
self.stonith_process = None
self.stonith_output = ""
self.stonith_patterns = []
self.negative_stonith_patterns = []
self.executed = 0
rsc_classes = output_from_command("crm_resource --list-standards")
def __new_cmd(self, cmd, args, exitcode, stdout_match = "", no_wait = 0, stdout_negative_match = "", kill=None):
self.cmds.append(
{
"cmd" : cmd,
"kill" : kill,
"args" : args,
"expected_exitcode" : exitcode,
"stdout_match" : stdout_match,
"stdout_negative_match" : stdout_negative_match,
"no_wait" : no_wait,
}
)
def stop_pacemaker(self):
cmd = shlex.split("killall -9 -q pacemakerd")
test = subprocess.Popen(cmd, stdout=subprocess.PIPE)
test.wait()
def start_environment(self):
### make sure we are in full control here ###
self.stop_pacemaker()
cmd = shlex.split("killall -9 -q stonithd")
test = subprocess.Popen(cmd, stdout=subprocess.PIPE)
test.wait()
if self.verbose:
print "Starting stonithd with %s" % self.stonith_options
self.stonith_process = subprocess.Popen(
shlex.split("@CRM_DAEMON_DIR@/stonithd %s -V" % self.stonith_options),
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
time.sleep(1)
def clean_environment(self):
if self.stonith_process:
self.stonith_process.terminate()
self.stonith_output = self.stonith_process.communicate()[1]
self.stonith_process = None
if self.verbose:
print self.stonith_output
def add_stonith_log_pattern(self, pattern):
self.stonith_patterns.append(pattern)
def add_stonith_negative_log_pattern(self, pattern):
self.negative_stonith_patterns.append(pattern)
def add_cmd(self, cmd, args):
self.__new_cmd(cmd, args, 0, "")
def add_cmd_no_wait(self, cmd, args):
self.__new_cmd(cmd, args, 0, "", 1)
def add_cmd_check_stdout(self, cmd, args, match, no_match = ""):
self.__new_cmd(cmd, args, 0, match, 0, no_match)
def add_expected_fail_cmd(self, cmd, args, exitcode = 255):
self.__new_cmd(cmd, args, exitcode, "")
def get_exitcode(self):
return self.result_exitcode
def print_result(self, filler):
print "%s%s" % (filler, self.result_txt)
def run_cmd(self, args):
cmd = shlex.split(args['args'])
cmd.insert(0, args['cmd'])
if self.verbose:
print "\n\nRunning: "+" ".join(cmd)
test = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
if args['kill']:
if self.verbose:
print "Also running: "+args['kill']
subprocess.Popen(shlex.split(args['kill']))
if args['no_wait'] == 0:
test.wait()
else:
return 0
output_res = test.communicate()
output = output_res[0] + output_res[1]
if self.verbose:
print output
if args['stdout_match'] != "" and output.count(args['stdout_match']) == 0:
test.returncode = -2
print "STDOUT string '%s' was not found in cmd output: %s" % (args['stdout_match'], output)
if args['stdout_negative_match'] != "" and output.count(args['stdout_negative_match']) != 0:
test.returncode = -2
print "STDOUT string '%s' was found in cmd output: %s" % (args['stdout_negative_match'], output)
return test.returncode;
def count_negative_matches(self, outline):
count = 0
for line in self.negative_stonith_patterns:
if outline.count(line):
count = 1
if self.verbose:
print "This pattern should not have matched = '%s" % (line)
return count
def match_stonith_patterns(self):
negative_matches = 0
cur = 0
pats = self.stonith_patterns
total_patterns = len(self.stonith_patterns)
if len(self.stonith_patterns) == 0:
return
for line in self.stonith_output.split("\n"):
negative_matches = negative_matches + self.count_negative_matches(line)
if len(pats) == 0:
continue
cur = -1
for p in pats:
cur = cur + 1
if line.count(pats[cur]):
del pats[cur]
break
if len(pats) > 0 or negative_matches:
if self.verbose:
for p in pats:
print "Pattern Not Matched = '%s'" % p
self.result_txt = "FAILURE - '%s' failed. %d patterns out of %d not matched. %d negative matches." % (self.name, len(pats), total_patterns, negative_matches)
self.result_exitcode = -1
def run(self):
res = 0
i = 1
self.start_environment()
if self.verbose:
print "\n--- START TEST - %s" % self.name
self.result_txt = "SUCCESS - '%s'" % (self.name)
self.result_exitcode = 0
for cmd in self.cmds:
res = self.run_cmd(cmd)
if res != cmd['expected_exitcode']:
print "Step %d FAILED - command returned %d, expected %d" % (i, res, cmd['expected_exitcode'])
self.result_txt = "FAILURE - '%s' failed at step %d. Command: %s %s" % (self.name, i, cmd['cmd'], cmd['args'])
self.result_exitcode = -1
break
else:
if self.verbose:
print "Step %d SUCCESS" % (i)
i = i + 1
self.clean_environment()
if self.result_exitcode == 0:
self.match_stonith_patterns()
print self.result_txt
if self.verbose:
print "--- END TEST - %s\n" % self.name
self.executed = 1
return res
class Tests:
def __init__(self, verbose = 0):
self.tests = []
self.verbose = verbose
self.autogen_corosync_cfg = 0
if not os.path.exists("/etc/corosync/corosync.conf"):
self.autogen_corosync_cfg = 1
def new_test(self, name, description, with_cpg = 0):
test = Test(name, description, self.verbose, with_cpg)
self.tests.append(test)
return test
def print_list(self):
print "\n==== %d TESTS FOUND ====" % (len(self.tests))
print "%35s - %s" % ("TEST NAME", "TEST DESCRIPTION")
print "%35s - %s" % ("--------------------", "--------------------")
for test in self.tests:
print "%35s - %s" % (test.name, test.description)
print "==== END OF LIST ====\n"
def start_corosync(self):
if self.verbose:
print "Starting corosync"
test = subprocess.Popen("corosync", stdout=subprocess.PIPE)
test.wait()
time.sleep(10)
def stop_corosync(self):
cmd = shlex.split("killall -9 -q corosync")
test = subprocess.Popen(cmd, stdout=subprocess.PIPE)
test.wait()
def run_single(self, name):
for test in self.tests:
if test.name == name:
test.run()
break;
def run_tests_matching(self, pattern):
for test in self.tests:
if test.name.count(pattern) != 0:
test.run()
def run_cpg_only(self):
for test in self.tests:
if test.enable_corosync:
test.run()
def run_no_cpg(self):
for test in self.tests:
if not test.enable_corosync:
test.run()
def run_tests(self):
for test in self.tests:
test.run()
def exit(self):
for test in self.tests:
if test.executed == 0:
continue
if test.get_exitcode() != 0:
sys.exit(-1)
sys.exit(0)
def print_results(self):
failures = 0;
success = 0;
print "\n\n======= FINAL RESULTS =========="
print "\n--- FAILURE RESULTS:"
for test in self.tests:
if test.executed == 0:
continue
if test.get_exitcode() != 0:
failures = failures + 1
test.print_result(" ")
else:
success = success + 1
if failures == 0:
print " None"
print "\n--- TOTALS\n Pass:%d\n Fail:%d\n" % (success, failures)
def build_api_sanity_tests(self):
verbose_arg = ""
if self.verbose:
verbose_arg = "-V"
test = self.new_test("standalone_low_level_api_test", "Sanity test client api in standalone mode.")
test.add_cmd("@CRM_DAEMON_DIR@/stonith-test", "-t %s" % (verbose_arg))
test = self.new_test("cpg_low_level_api_test", "Sanity test client api using mainloop and cpg.", 1)
test.add_cmd("@CRM_DAEMON_DIR@/stonith-test", "-m %s" % (verbose_arg))
def build_custom_timeout_tests(self):
# custom timeout without topology
test = self.new_test("cpg_custom_timeout_1",
"Verify per device timeouts work as expected without using topology.", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\" -o \"pcmk_off_timeout=1\"")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node3\" -o \"pcmk_off_timeout=4\"")
test.add_cmd("stonith_admin", "-F node3 -t 2")
# timeout is 2+1+4 = 7
test.add_stonith_log_pattern("remote op timeout set to 7")
# custom timeout _WITH_ topology
test = self.new_test("cpg_custom_timeout_2",
"Verify per device timeouts work as expected _WITH_ topology.", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\" -o \"pcmk_off_timeout=1\"")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node3\" -o \"pcmk_off_timeout=4000\"")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v true1")
test.add_cmd("stonith_admin", "-r node3 -i 3 -v false2")
test.add_cmd("stonith_admin", "-F node3 -t 2")
# timeout is 2+1+4000 = 4003
test.add_stonith_log_pattern("remote op timeout set to 4003")
def build_fence_merge_tests(self):
### Simple test that overlapping fencing operations get merged
test = self.new_test("cpg_custom_merge_single",
"Verify overlapping identical fencing operations are merged, no fencing levels used.", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\" ")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node3\"")
test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10")
test.add_cmd("stonith_admin", "-F node3 -t 10")
### one merger will happen
test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client")
### the pattern below signifies that both the original and duplicate operation completed
test.add_stonith_log_pattern("Operation off of node3 by")
test.add_stonith_log_pattern("Operation off of node3 by")
### Test that multiple mergers occur
test = self.new_test("cpg_custom_merge_multiple",
"Verify multiple overlapping identical fencing operations are merged", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\" ")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node3\"")
test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10")
test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10")
test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10")
test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10")
test.add_cmd("stonith_admin", "-F node3 -t 10")
### 4 mergers should occur
test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client")
test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client")
test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client")
test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client")
### the pattern below signifies that both the original and duplicate operation completed
test.add_stonith_log_pattern("Operation off of node3 by")
test.add_stonith_log_pattern("Operation off of node3 by")
test.add_stonith_log_pattern("Operation off of node3 by")
test.add_stonith_log_pattern("Operation off of node3 by")
test.add_stonith_log_pattern("Operation off of node3 by")
### Test that multiple mergers occur with topologies used
test = self.new_test("cpg_custom_merge_with_topology",
"Verify multiple overlapping identical fencing operations are merged with fencing levels.", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\" ")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v false2")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v true1")
test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10")
test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10")
test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10")
test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10")
test.add_cmd("stonith_admin", "-F node3 -t 10")
### 4 mergers should occur
test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client")
test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client")
test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client")
test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client")
### the pattern below signifies that both the original and duplicate operation completed
test.add_stonith_log_pattern("Operation off of node3 by")
test.add_stonith_log_pattern("Operation off of node3 by")
test.add_stonith_log_pattern("Operation off of node3 by")
test.add_stonith_log_pattern("Operation off of node3 by")
test.add_stonith_log_pattern("Operation off of node3 by")
test = self.new_test("cpg_custom_no_merge",
"Verify differing fencing operations are not merged", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node3 node2\"")
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3 node2\" ")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node3 node2\"")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v false2")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v true1")
test.add_cmd_no_wait("stonith_admin", "-F node2 -t 10")
test.add_cmd("stonith_admin", "-F node3 -t 10")
test.add_stonith_negative_log_pattern("Merging stonith action off for node node3 originating from client")
def build_standalone_tests(self):
test_types = [
{
"prefix" : "standalone" ,
"use_cpg" : 0,
},
{
"prefix" : "cpg" ,
"use_cpg" : 1,
},
]
# test what happens when all devices timeout
for test_type in test_types:
test = self.new_test("%s_fence_multi_device_failure" % test_type["prefix"],
"Verify that all devices timeout, a fencing failure is returned.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R false3 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_expected_fail_cmd("stonith_admin", "-F node3 -t 2", 194)
if test_type["use_cpg"] == 1:
test.add_stonith_log_pattern("remote op timeout set to 6")
test.add_stonith_log_pattern("for host 'node3' with device 'false1' returned: -62")
test.add_stonith_log_pattern("for host 'node3' with device 'false2' returned: -62")
test.add_stonith_log_pattern("for host 'node3' with device 'false3' returned: -62")
# test what happens when multiple devices can fence a node, but the first device fails.
for test_type in test_types:
test = self.new_test("%s_fence_device_failure_rollover" % test_type["prefix"],
"Verify that when one fence device fails for a node, the others are tried.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-F node3 -t 2")
if test_type["use_cpg"] == 1:
test.add_stonith_log_pattern("remote op timeout set to 6")
# simple topology test for one device
for test_type in test_types:
if test_type["use_cpg"] == 0:
continue
test = self.new_test("%s_topology_simple" % test_type["prefix"],
"Verify all fencing devices at a level are used.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v true")
test.add_cmd("stonith_admin", "-F node3 -t 2")
test.add_stonith_log_pattern("remote op timeout set to 2")
test.add_stonith_log_pattern("for host 'node3' with device 'true' returned: 0")
# test what happens when the first fencing level has multiple devices.
for test_type in test_types:
if test_type["use_cpg"] == 0:
continue
test = self.new_test("%s_topology_device_fails" % test_type["prefix"],
"Verify if one device in a level fails, the other is tried.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R false -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v false")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v true")
test.add_cmd("stonith_admin", "-F node3 -t 20")
test.add_stonith_log_pattern("remote op timeout set to 4")
test.add_stonith_log_pattern("for host 'node3' with device 'false' returned: -62")
test.add_stonith_log_pattern("for host 'node3' with device 'true' returned: 0")
# test what happens when the first fencing level fails.
for test_type in test_types:
if test_type["use_cpg"] == 0:
continue
test = self.new_test("%s_topology_multi_level_fails" % test_type["prefix"],
"Verify if one level fails, the next leve is tried.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true2 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true3 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true4 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v true1")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v true2")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v false2")
test.add_cmd("stonith_admin", "-r node3 -i 3 -v true3")
test.add_cmd("stonith_admin", "-r node3 -i 3 -v true4")
test.add_cmd("stonith_admin", "-F node3 -t 2")
test.add_stonith_log_pattern("remote op timeout set to 12")
test.add_stonith_log_pattern("for host 'node3' with device 'false1' returned: -62")
test.add_stonith_log_pattern("for host 'node3' with device 'false2' returned: -62")
test.add_stonith_log_pattern("for host 'node3' with device 'true3' returned: 0")
test.add_stonith_log_pattern("for host 'node3' with device 'true4' returned: 0")
# test what happens when the first fencing level had devices that no one has registered
for test_type in test_types:
if test_type["use_cpg"] == 0:
continue
test = self.new_test("%s_topology_missing_devices" % test_type["prefix"],
"Verify topology can continue with missing devices.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true2 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true3 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true4 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v true1")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v true2")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v false2")
test.add_cmd("stonith_admin", "-r node3 -i 3 -v true3")
test.add_cmd("stonith_admin", "-r node3 -i 3 -v true4")
test.add_cmd("stonith_admin", "-F node3 -t 2")
# Test what happens if multiple fencing levels are defined, and then the first one is removed.
for test_type in test_types:
if test_type["use_cpg"] == 0:
continue
test = self.new_test("%s_topology_level_removal" % test_type["prefix"],
"Verify level removal works.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true2 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true3 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true4 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1")
test.add_cmd("stonith_admin", "-r node3 -i 1 -v true1")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v true2")
test.add_cmd("stonith_admin", "-r node3 -i 2 -v false2")
test.add_cmd("stonith_admin", "-r node3 -i 3 -v true3")
test.add_cmd("stonith_admin", "-r node3 -i 3 -v true4")
# Now remove level 2, verify none of the devices in level two are hit.
test.add_cmd("stonith_admin", "-d node3 -i 2")
test.add_cmd("stonith_admin", "-F node3 -t 20")
test.add_stonith_log_pattern("remote op timeout set to 8")
test.add_stonith_log_pattern("for host 'node3' with device 'false1' returned: -62")
test.add_stonith_negative_log_pattern("for host 'node3' with device 'false2' returned: -62")
test.add_stonith_negative_log_pattern("for host 'node3' with device 'false2' returned: -1001")
test.add_stonith_log_pattern("for host 'node3' with device 'true3' returned: 0")
test.add_stonith_log_pattern("for host 'node3' with device 'true4' returned: 0")
# test the stonith builds the correct list of devices that can fence a node.
for test_type in test_types:
test = self.new_test("%s_list_devices" % test_type["prefix"],
"Verify list of devices that can fence a node is correct", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-R true2 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd("stonith_admin", "-R true3 -a fence_true -o \"pcmk_host_list=node1 node2 node3\"")
test.add_cmd_check_stdout("stonith_admin", "-l node1 -V", "true2", "true1")
test.add_cmd_check_stdout("stonith_admin", "-l node1 -V", "true3", "true1")
# simple test of device monitor
for test_type in test_types:
test = self.new_test("%s_monitor" % test_type["prefix"],
"Verify device is reachable", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-Q true1")
test.add_cmd("stonith_admin", "-Q false1")
test.add_expected_fail_cmd("stonith_admin", "-Q true2", 237)
# Verify monitor occurs for duration of timeout period on failure
for test_type in test_types:
test = self.new_test("%s_monitor_timeout" % test_type["prefix"],
"Verify monitor uses duration of timeout period given.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_monitor_fail -o \"pcmk_host_list=node3\"")
- test.add_expected_fail_cmd("stonith_admin", "-Q true1 -t 5", 23)
+ test.add_expected_fail_cmd("stonith_admin", "-Q true1 -t 5", 195)
test.add_stonith_log_pattern("Attempt 2 to execute")
# Verify monitor occurs for duration of timeout period on failure, but stops at max retries
for test_type in test_types:
test = self.new_test("%s_monitor_timeout_max_retries" % test_type["prefix"],
"Verify monitor retries until max retry value or timeout is hit.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_monitor_fail -o \"pcmk_host_list=node3\"")
- test.add_expected_fail_cmd("stonith_admin", "-Q true1 -t 15", 23)
+ test.add_expected_fail_cmd("stonith_admin", "-Q true1 -t 15",195)
test.add_stonith_log_pattern("Attempted to execute agent fence_dummy_monitor_fail (list) the maximum number of times")
# simple register test
for test_type in test_types:
test = self.new_test("%s_register" % test_type["prefix"],
"Verify devices can be registered and un-registered", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-Q true1")
test.add_cmd("stonith_admin", "-D true1")
test.add_expected_fail_cmd("stonith_admin", "-Q true1", 237)
# simple reboot test
for test_type in test_types:
test = self.new_test("%s_reboot" % test_type["prefix"],
"Verify devices can be rebooted", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-B node3 -t 2")
test.add_cmd("stonith_admin", "-D true1")
test.add_expected_fail_cmd("stonith_admin", "-Q true1", 237)
# test fencing history.
for test_type in test_types:
if test_type["use_cpg"] == 0:
continue
test = self.new_test("%s_fence_history" % test_type["prefix"],
"Verify last fencing operation is returned.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-F node3 -t 2 -V")
test.add_cmd_check_stdout("stonith_admin", "-H node3", "was able to turn off node node3", "")
# simple test of dynamic list query
for test_type in test_types:
test = self.new_test("%s_dynamic_list_query" % test_type["prefix"],
"Verify dynamic list of fencing devices can be retrieved.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true")
test.add_cmd("stonith_admin", "-R true2 -a fence_true")
test.add_cmd("stonith_admin", "-R true3 -a fence_true")
test.add_cmd_check_stdout("stonith_admin", "-l fake_port_1", "3 devices found")
# fence using dynamic list query
for test_type in test_types:
test = self.new_test("%s_fence_dynamic_list_query" % test_type["prefix"],
"Verify dynamic list of fencing devices can be retrieved.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true")
test.add_cmd("stonith_admin", "-R true2 -a fence_true")
test.add_cmd("stonith_admin", "-R true3 -a fence_true")
test.add_cmd("stonith_admin", "-F fake_port_1 -t 5 -V");
# simple test of query using status action
for test_type in test_types:
test = self.new_test("%s_status_query" % test_type["prefix"],
"Verify dynamic list of fencing devices can be retrieved.", test_type["use_cpg"])
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_check=status\"")
test.add_cmd("stonith_admin", "-R true2 -a fence_true -o \"pcmk_host_check=status\"")
test.add_cmd("stonith_admin", "-R true3 -a fence_true -o \"pcmk_host_check=status\"")
test.add_cmd_check_stdout("stonith_admin", "-l fake_port_1", "3 devices found")
def build_nodeid_tests(self):
our_uname = output_from_command("uname -n")
if our_uname:
our_uname = our_uname[0]
### verify nodeid is supplied when nodeid is in the metadata parameters
test = self.new_test("cpg_supply_nodeid",
"Verify nodeid is given when fence agent has nodeid as parameter", 1)
test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_metadata_helper -o \"pcmk_host_list=%s\"" % (our_uname))
test.add_cmd("stonith_admin", "-F %s -t 3" % (our_uname))
test.add_stonith_log_pattern("For stonith action (off) for victim %s, adding nodeid" % (our_uname))
### verify nodeid is _NOT_ supplied when nodeid is not in the metadata parameters
test = self.new_test("cpg_do_not_supply_nodeid",
"Verify nodeid is _NOT_ given when fence agent does not have nodeid as parameter", 1)
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=%s\"" % (our_uname))
test.add_cmd("stonith_admin", "-F %s -t 3" % (our_uname))
test.add_stonith_negative_log_pattern("For stonith action (off) for victim %s, adding nodeid" % (our_uname))
### verify nodeid use doesn't explode standalone mode
test = self.new_test("standalone_do_not_supply_nodeid",
"Verify nodeid in metadata parameter list doesn't kill standalone mode", 0)
test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_metadata_helper -o \"pcmk_host_list=%s\"" % (our_uname))
test.add_cmd("stonith_admin", "-F %s -t 3" % (our_uname))
test.add_stonith_negative_log_pattern("For stonith action (off) for victim %s, adding nodeid" % (our_uname))
def build_unfence_tests(self):
our_uname = output_from_command("uname -n")
if our_uname:
our_uname = our_uname[0]
### Simple test unfencing works
test = self.new_test("cpg_unfence_simple",
"Verify simple unfencing.", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-R true1 -a fence_true -o \"pcmk_host_list=node3\" ")
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=node3\"")
test.add_cmd("stonith_admin", "-U node3 -t 3")
### verify unfencing using on_target device
test = self.new_test("cpg_unfence_on_target_1",
"Verify unfencing with on_target = true", 1)
test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_metadata_helper -o \"pcmk_host_list=%s\"" % (our_uname))
test.add_cmd("stonith_admin", "-U %s -t 3" % (our_uname))
test.add_stonith_log_pattern("(on) to be executed on the target node")
### verify failure of unfencing using on_target device
test = self.new_test("cpg_unfence_on_target_2",
"Verify failure unfencing with on_target = true", 1)
test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_metadata_helper -o \"pcmk_host_list=%s node_fake_1234\"" % (our_uname))
- test.add_expected_fail_cmd("stonith_admin", "-U node_fake_1234 -t 3", 194)
+ test.add_expected_fail_cmd("stonith_admin", "-U node_fake_1234 -t 3", 143)
test.add_stonith_log_pattern("(on) to be executed on the target node")
### verify unfencing using on_target device with topology
test = self.new_test("cpg_unfence_on_target_3",
"Verify unfencing with on_target = true using topology", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=%s node3\"" % (our_uname))
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=%s node3\"" % (our_uname))
test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_metadata_helper -o \"pcmk_host_list=%s node3\"" % (our_uname))
test.add_cmd("stonith_admin", "-r %s -i 1 -v false1" % (our_uname))
test.add_cmd("stonith_admin", "-r %s -i 2 -v false2" % (our_uname))
test.add_cmd("stonith_admin", "-r %s -i 3 -v true1" % (our_uname))
test.add_cmd("stonith_admin", "-U %s -t 3" % (our_uname))
test.add_stonith_log_pattern("(on) to be executed on the target node")
### verify unfencing using on_target device with topology fails
test = self.new_test("cpg_unfence_on_target_4",
"Verify unfencing failure with on_target = true using topology", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=%s node_fake\"" % (our_uname))
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=%s node_fake\"" % (our_uname))
test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_metadata_helper -o \"pcmk_host_list=%s node_fake\"" % (our_uname))
test.add_cmd("stonith_admin", "-r node_fake -i 1 -v false1")
test.add_cmd("stonith_admin", "-r node_fake -i 2 -v false2")
test.add_cmd("stonith_admin", "-r node_fake -i 3 -v true1")
test.add_expected_fail_cmd("stonith_admin", "-U node_fake -t 3", 194)
test.add_stonith_log_pattern("(on) to be executed on the target node")
### verify use of on_target = true for "on" action does not interfere with "off" action
test = self.new_test("cpg_unfence_on_target_ignored",
"Verify on target is ignored for other actions", 1)
test.add_cmd("stonith_admin", "-R false1 -a fence_false -o \"pcmk_host_list=%s node_fake\"" % (our_uname))
test.add_cmd("stonith_admin", "-R false2 -a fence_false -o \"pcmk_host_list=%s node_fake\"" % (our_uname))
test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_metadata_helper -o \"pcmk_host_list=%s node_fake\"" % (our_uname))
test.add_cmd("stonith_admin", "-r node_fake -i 1 -v false1")
test.add_cmd("stonith_admin", "-r node_fake -i 2 -v false2")
test.add_cmd("stonith_admin", "-r node_fake -i 3 -v true1")
test.add_cmd("stonith_admin", "-F node_fake -t 3")
test.add_stonith_log_pattern("(on) to be executed on the target node")
def setup_environment(self, use_corosync):
if self.autogen_corosync_cfg and use_corosync:
corosync_conf = ("""
totem {
version: 2
crypto_cipher: none
crypto_hash: none
nodeid: 101
secauth: off
interface {
ttl: 1
ringnumber: 0
mcastport: 6666
mcastaddr: 226.94.1.1
bindnetaddr: 127.0.0.1
}
}
logging {
debug: off
fileline: off
to_syslog: no
to_stderr: no
syslog_facility: daemon
timestamp: on
to_logfile: yes
logfile: /var/log/corosync.log
logfile_priority: info
}
""")
os.system("cat <<-END >>/etc/corosync/corosync.conf\n%s\nEND" % (corosync_conf))
if use_corosync:
### make sure we are in control ###
self.stop_corosync()
self.start_corosync()
monitor_fail_agent = ("""#!/usr/bin/python
import sys
def main():
for line in sys.stdin.readlines():
if line.count("monitor") > 0:
sys.exit(-1);
sys.exit(-1)
if __name__ == "__main__":
main()
""")
on_target_agent = ("""#!/usr/bin/python
import sys
def main():
for line in sys.stdin.readlines():
off_hit = 0
nodeid_found = 0
if line.count("monitor") > 0:
sys.exit(0)
if line.count("metadata") > 0:
print ''
print ' dummy description.'
print ' http://www.example.com'
print ' '
print ' '
print ' '
print ' '
print ' Fencing Action'
print ' '
print ' '
print ' '
print ' Corosync nodeid of the fence victim'
print ' '
print ' '
print ' '
print ' '
print ' Physical plug number or name of virtual machine'
print ' '
print ' '
print ' '
print ' '
print ' '
print ' '
print ' '
print ' '
print ''
sys.exit(0)
if line.count("on") > 0:
sys.exit(0)
if line.count("off") > 0:
off_hit = 1
if line.count("nodeid") > 0:
nodeid_found = 1
if off_hit and nodeid_found:
sys.exit(0)
sys.exit(-1)
if __name__ == "__main__":
main()
""")
os.system("cat <<-END >>/usr/sbin/fence_dummy_metadata_helper\n%s\nEND" % (on_target_agent))
os.system("chmod 711 /usr/sbin/fence_dummy_metadata_helper")
os.system("cat <<-END >>/usr/sbin/fence_dummy_monitor_fail\n%s\nEND" % (monitor_fail_agent))
os.system("chmod 711 /usr/sbin/fence_dummy_monitor_fail")
os.system("cp /usr/share/pacemaker/tests/cts/fence_false /usr/sbin/fence_false")
os.system("cp /usr/share/pacemaker/tests/cts/fence_true /usr/sbin/fence_true")
def cleanup_environment(self, use_corosync):
if use_corosync:
self.stop_corosync()
if self.verbose and os.path.exists('/var/log/corosync.log'):
print "Daemon output"
f = open('/var/log/corosync.log', 'r')
for line in f.readlines():
print line.strip()
os.remove('/var/log/corosync.log')
if self.autogen_corosync_cfg:
os.system("rm -f /etc/corosync/corosync.conf")
os.system("rm -f /usr/sbin/fence_dummy_metadata_helper")
os.system("rm -f /usr/sbin/fence_dummy_monitor_fail")
class TestOptions:
def __init__(self):
self.options = {}
self.options['list-tests'] = 0
self.options['run-all'] = 1
self.options['run-only'] = ""
self.options['run-only-pattern'] = ""
self.options['verbose'] = 0
self.options['invalid-arg'] = ""
self.options['cpg-only'] = 0
self.options['no-cpg'] = 0
self.options['show-usage'] = 0
def build_options(self, argv):
args = argv[1:]
skip = 0
for i in range(0, len(args)):
if skip:
skip = 0
continue
elif args[i] == "-h" or args[i] == "--help":
self.options['show-usage'] = 1
elif args[i] == "-l" or args[i] == "--list-tests":
self.options['list-tests'] = 1
elif args[i] == "-V" or args[i] == "--verbose":
self.options['verbose'] = 1
elif args[i] == "-n" or args[i] == "--no-cpg":
self.options['no-cpg'] = 1
elif args[i] == "-c" or args[i] == "--cpg-only":
self.options['cpg-only'] = 1
elif args[i] == "-r" or args[i] == "--run-only":
self.options['run-only'] = args[i+1]
skip = 1
elif args[i] == "-p" or args[i] == "--run-only-pattern":
self.options['run-only-pattern'] = args[i+1]
skip = 1
def show_usage(self):
print "usage: " + sys.argv[0] + " [options]"
print "If no options are provided, all tests will run"
print "Options:"
print "\t [--help | -h] Show usage"
print "\t [--list-tests | -l] Print out all registered tests."
print "\t [--cpg-only | -c] Only run tests that require corosync."
print "\t [--no-cpg | -n] Only run tests that do not require corosync"
print "\t [--run-only | -r 'testname'] Run a specific test"
print "\t [--verbose | -V] Verbose output"
print "\t [--run-only-pattern | -p 'string'] Run only tests containing the string value"
print "\n\tExample: Run only the test 'start_top'"
print "\t\t python ./regression.py --run-only start_stop"
print "\n\tExample: Run only the tests with the string 'systemd' present in them"
print "\t\t python ./regression.py --run-only-pattern systemd"
def main(argv):
o = TestOptions()
o.build_options(argv)
use_corosync = 1
tests = Tests(o.options['verbose'])
tests.build_standalone_tests()
tests.build_custom_timeout_tests()
tests.build_api_sanity_tests()
tests.build_fence_merge_tests()
tests.build_unfence_tests()
tests.build_nodeid_tests()
if o.options['list-tests']:
tests.print_list()
sys.exit(0)
elif o.options['show-usage']:
o.show_usage()
sys.exit(0)
print "Starting ..."
if o.options['no-cpg']:
use_corosync = 0
tests.setup_environment(use_corosync)
if o.options['run-only-pattern'] != "":
tests.run_tests_matching(o.options['run-only-pattern'])
tests.print_results()
elif o.options['run-only'] != "":
tests.run_single(o.options['run-only'])
tests.print_results()
elif o.options['no-cpg']:
tests.run_no_cpg()
tests.print_results()
elif o.options['cpg-only']:
tests.run_cpg_only()
tests.print_results()
else:
tests.run_tests()
tests.print_results()
tests.cleanup_environment(use_corosync)
tests.exit()
if __name__=="__main__":
main(sys.argv)
diff --git a/fencing/remote.c b/fencing/remote.c
index 5a8156af53..18c13bce0a 100644
--- a/fencing/remote.c
+++ b/fencing/remote.c
@@ -1,1358 +1,1358 @@
/*
* Copyright (C) 2009 Andrew Beekhof
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This software is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#define TIMEOUT_MULTIPLY_FACTOR 1.2
typedef struct st_query_result_s {
char *host;
int devices;
/* only try peers for non-topology based operations once */
gboolean tried;
GListPtr device_list;
GHashTable *custom_action_timeouts;
/* Subset of devices that peer has verified connectivity on */
GHashTable *verified_devices;
} st_query_result_t;
GHashTable *remote_op_list = NULL;
void call_remote_stonith(remote_fencing_op_t * op, st_query_result_t * peer);
static void remote_op_done(remote_fencing_op_t * op, xmlNode * data, int rc, int dup);
extern xmlNode *stonith_create_op(int call_id, const char *token, const char *op, xmlNode * data,
int call_options);
static void report_timeout_period(remote_fencing_op_t * op, int op_timeout);
static int get_op_total_timeout(remote_fencing_op_t * op, st_query_result_t * chosen_peer,
int default_timeout);
static void
free_remote_query(gpointer data)
{
if (data) {
st_query_result_t *query = data;
crm_trace("Free'ing query result from %s", query->host);
free(query->host);
g_list_free_full(query->device_list, free);
g_hash_table_destroy(query->custom_action_timeouts);
g_hash_table_destroy(query->verified_devices);
free(query);
}
}
static void
clear_remote_op_timers(remote_fencing_op_t * op)
{
if (op->query_timer) {
g_source_remove(op->query_timer);
op->query_timer = 0;
}
if (op->op_timer_total) {
g_source_remove(op->op_timer_total);
op->op_timer_total = 0;
}
if (op->op_timer_one) {
g_source_remove(op->op_timer_one);
op->op_timer_one = 0;
}
}
static void
free_remote_op(gpointer data)
{
remote_fencing_op_t *op = data;
crm_trace("Free'ing op %s for %s", op->id, op->target);
crm_log_xml_debug(op->request, "Destroying");
clear_remote_op_timers(op);
free(op->id);
free(op->action);
free(op->target);
free(op->client_id);
free(op->client_name);
free(op->originator);
if (op->query_results) {
g_list_free_full(op->query_results, free_remote_query);
}
if (op->request) {
free_xml(op->request);
op->request = NULL;
}
free(op);
}
static xmlNode *
create_op_done_notify(remote_fencing_op_t * op, int rc)
{
xmlNode *notify_data = create_xml_node(NULL, T_STONITH_NOTIFY_FENCE);
crm_xml_add_int(notify_data, "state", op->state);
crm_xml_add_int(notify_data, F_STONITH_RC, rc);
crm_xml_add(notify_data, F_STONITH_TARGET, op->target);
crm_xml_add(notify_data, F_STONITH_ACTION, op->action);
crm_xml_add(notify_data, F_STONITH_DELEGATE, op->delegate);
crm_xml_add(notify_data, F_STONITH_REMOTE_OP_ID, op->id);
crm_xml_add(notify_data, F_STONITH_ORIGIN, op->originator);
crm_xml_add(notify_data, F_STONITH_CLIENTID, op->client_id);
crm_xml_add(notify_data, F_STONITH_CLIENTNAME, op->client_name);
return notify_data;
}
static void
bcast_result_to_peers(remote_fencing_op_t * op, int rc)
{
static int count = 0;
xmlNode *bcast = create_xml_node(NULL, T_STONITH_REPLY);
xmlNode *notify_data = create_op_done_notify(op, rc);
count++;
crm_trace("Broadcasting result to peers");
crm_xml_add(bcast, F_TYPE, T_STONITH_NOTIFY);
crm_xml_add(bcast, F_SUBTYPE, "broadcast");
crm_xml_add(bcast, F_STONITH_OPERATION, T_STONITH_NOTIFY);
crm_xml_add_int(bcast, "count", count);
add_message_xml(bcast, F_STONITH_CALLDATA, notify_data);
send_cluster_message(NULL, crm_msg_stonith_ng, bcast, FALSE);
free_xml(notify_data);
free_xml(bcast);
return;
}
static void
handle_local_reply_and_notify(remote_fencing_op_t * op, xmlNode * data, int rc)
{
xmlNode *notify_data = NULL;
xmlNode *reply = NULL;
if (op->notify_sent == TRUE) {
/* nothing to do */
return;
}
/* Do notification with a clean data object */
notify_data = create_op_done_notify(op, rc);
crm_xml_add_int(data, "state", op->state);
crm_xml_add(data, F_STONITH_TARGET, op->target);
crm_xml_add(data, F_STONITH_OPERATION, op->action);
reply = stonith_construct_reply(op->request, NULL, data, rc);
crm_xml_add(reply, F_STONITH_DELEGATE, op->delegate);
/* Send fencing OP reply to local client that initiated fencing */
do_local_reply(reply, op->client_id, op->call_options & st_opt_sync_call, FALSE);
/* bcast to all local clients that the fencing operation happend */
do_stonith_notify(0, T_STONITH_NOTIFY_FENCE, rc, notify_data);
/* mark this op as having notify's already sent */
op->notify_sent = TRUE;
free_xml(reply);
free_xml(notify_data);
}
static void
handle_duplicates(remote_fencing_op_t * op, xmlNode * data, int rc)
{
GListPtr iter = NULL;
for (iter = op->duplicates; iter != NULL; iter = iter->next) {
remote_fencing_op_t *other = iter->data;
if (other->state == st_duplicate) {
/* Ie. it hasn't timed out already */
other->state = op->state;
crm_debug("Peforming duplicate notification for %s@%s.%.8s = %s", other->client_name,
other->originator, other->id, pcmk_strerror(rc));
remote_op_done(other, data, rc, TRUE);
} else {
crm_err("Skipping duplicate notification for %s@%s - %d", other->client_name,
other->originator, other->state);
}
}
}
/*!
* \internal
* \brief Finalize a remote operation.
*
* \description This function has two code paths.
*
* Path 1. This node is the owner of the operation and needs
* to notify the cpg group via a broadcast as to the operation's
* results.
*
* Path 2. The cpg broadcast is received. All nodes notify their local
* stonith clients the operation results.
*
* So, The owner of the operation first notifies the cluster of the result,
* and once that cpg notify is received back it notifies all the local clients.
*
* Nodes that are passive watchers of the operation will receive the
* broadcast and only need to notify their local clients the operation finished.
*
* \param op, The fencing operation to finalize
* \param data, The xml msg reply (if present) of the last delegated fencing
* operation.
* \param dup, Is this operation a duplicate, if so treat it a little differently
* making sure the broadcast is not sent out.
*/
static void
remote_op_done(remote_fencing_op_t * op, xmlNode * data, int rc, int dup)
{
int level = LOG_ERR;
const char *subt = NULL;
xmlNode *local_data = NULL;
op->completed = time(NULL);
clear_remote_op_timers(op);
if (op->notify_sent == TRUE) {
crm_err("Already sent notifications for '%s of %s by %s' (for=%s@%s.%.8s, state=%d): %s",
op->action, op->target, op->delegate ? op->delegate : "",
op->client_name, op->originator, op->id, op->state, pcmk_strerror(rc));
goto remote_op_done_cleanup;
}
if (!op->delegate && data) {
op->delegate = crm_element_value_copy(data, F_ORIG);
}
if (data == NULL) {
data = create_xml_node(NULL, "remote-op");
local_data = data;
}
/* Tell everyone the operation is done, we will continue
* with doing the local notifications once we receive
* the broadcast back. */
subt = crm_element_value(data, F_SUBTYPE);
if (dup == FALSE && safe_str_neq(subt, "broadcast")) {
/* Defer notification until the bcast message arrives */
bcast_result_to_peers(op, rc);
goto remote_op_done_cleanup;
}
if (rc == pcmk_ok || dup) {
level = LOG_NOTICE;
} else if (safe_str_neq(op->originator, stonith_our_uname)) {
level = LOG_NOTICE;
}
do_crm_log(level,
"Operation %s of %s by %s for %s@%s.%.8s: %s",
op->action, op->target, op->delegate ? op->delegate : "",
op->client_name, op->originator, op->id, pcmk_strerror(rc));
handle_local_reply_and_notify(op, data, rc);
if (dup == FALSE) {
handle_duplicates(op, data, rc);
}
/* Free non-essential parts of the record
* Keep the record around so we can query the history
*/
if (op->query_results) {
g_list_free_full(op->query_results, free_remote_query);
op->query_results = NULL;
}
if (op->request) {
free_xml(op->request);
op->request = NULL;
}
remote_op_done_cleanup:
free_xml(local_data);
}
static gboolean
remote_op_timeout_one(gpointer userdata)
{
remote_fencing_op_t *op = userdata;
op->op_timer_one = 0;
crm_notice("Remote %s operation on %s for %s.%8s timed out",
op->action, op->target, op->client_name, op->id);
call_remote_stonith(op, NULL);
return FALSE;
}
static gboolean
remote_op_timeout(gpointer userdata)
{
remote_fencing_op_t *op = userdata;
op->op_timer_total = 0;
if (op->state == st_done) {
crm_debug("Action %s (%s) for %s (%s) already completed",
op->action, op->id, op->target, op->client_name);
return FALSE;
}
crm_debug("Action %s (%s) for %s (%s) timed out",
op->action, op->id, op->target, op->client_name);
op->state = st_failed;
remote_op_done(op, NULL, -ETIME, FALSE);
return FALSE;
}
static gboolean
remote_op_query_timeout(gpointer data)
{
remote_fencing_op_t *op = data;
op->query_timer = 0;
if (op->state == st_done) {
crm_debug("Operation %s for %s already completed", op->id, op->target);
} else if (op->state == st_exec) {
crm_debug("Operation %s for %s already in progress", op->id, op->target);
} else if (op->query_results) {
crm_debug("Query %s for %s complete: %d", op->id, op->target, op->state);
call_remote_stonith(op, NULL);
} else {
crm_debug("Query %s for %s timed out: %d", op->id, op->target, op->state);
if (op->op_timer_total) {
g_source_remove(op->op_timer_total);
op->op_timer_total = 0;
}
remote_op_timeout(op);
}
return FALSE;
}
static int
stonith_topology_next(remote_fencing_op_t * op)
{
stonith_topology_t *tp = NULL;
if (op->target) {
/* Queries don't have a target set */
tp = g_hash_table_lookup(topology, op->target);
}
if (tp == NULL) {
return pcmk_ok;
}
set_bit(op->call_options, st_opt_topology);
do {
op->level++;
} while (op->level < ST_LEVEL_MAX && tp->levels[op->level] == NULL);
if (op->level < ST_LEVEL_MAX) {
crm_trace("Attempting fencing level %d for %s (%d devices) - %s@%s.%.8s",
op->level, op->target, g_list_length(tp->levels[op->level]),
op->client_name, op->originator, op->id);
op->devices = tp->levels[op->level];
return pcmk_ok;
}
crm_notice("All fencing options to fence %s for %s@%s.%.8s failed",
op->target, op->client_name, op->originator, op->id);
return -EINVAL;
}
/*!
* \brief Check to see if this operation is a duplicate of another in flight
* operation. If so merge this operation into the inflight operation, and mark
* it as a duplicate.
*/
static void
merge_duplicates(remote_fencing_op_t * op)
{
GHashTableIter iter;
remote_fencing_op_t *other = NULL;
time_t now = time(NULL);
g_hash_table_iter_init(&iter, remote_op_list);
while (g_hash_table_iter_next(&iter, NULL, (void **)&other)) {
crm_node_t *peer = NULL;
if (other->state > st_exec) {
/* Must be in-progress */
continue;
} else if (safe_str_neq(op->target, other->target)) {
/* Must be for the same node */
continue;
} else if (safe_str_neq(op->action, other->action)) {
crm_trace("Must be for the same action: %s vs. ", op->action, other->action);
continue;
} else if (safe_str_eq(op->client_name, other->client_name)) {
crm_trace("Must be for different clients: %s", op->client_name);
continue;
} else if (safe_str_eq(other->target, other->originator)) {
crm_trace("Can't be a suicide operation: %s", other->target);
continue;
}
peer = crm_get_peer(0, other->originator);
if(fencing_peer_active(peer) == FALSE) {
crm_notice("Failing stonith action %s for node %s originating from %s@%s.%.8s: Originator is dead",
other->action, other->target, other->client_name, other->originator, other->id);
other->state = st_failed;
continue;
} else if(other->total_timeout > 0 && now > (other->total_timeout + other->created)) {
crm_info("Stonith action %s for node %s originating from %s@%s.%.8s is too old: %d vs. %d + %d",
other->action, other->target, other->client_name, other->originator, other->id,
now, other->created, other->total_timeout);
continue;
}
/* There is another in-flight request to fence the same host
* Piggyback on that instead. If it fails, so do we.
*/
other->duplicates = g_list_append(other->duplicates, op);
if (other->total_timeout == 0) {
crm_trace("Making a best-guess as to the timeout used");
other->total_timeout = op->total_timeout =
TIMEOUT_MULTIPLY_FACTOR * get_op_total_timeout(op, NULL, op->base_timeout);
}
crm_notice
("Merging stonith action %s for node %s originating from client %s.%.8s with identical request from %s@%s.%.8s (%ds)",
op->action, op->target, op->client_name, op->id, other->client_name, other->originator,
other->id, other->total_timeout);
report_timeout_period(op, other->total_timeout);
op->state = st_duplicate;
}
}
static uint32_t fencing_active_peers(void)
{
uint32_t count = 0;
crm_node_t *entry;
GHashTableIter gIter;
g_hash_table_iter_init(&gIter, crm_peer_cache);
while (g_hash_table_iter_next(&gIter, NULL, (void **)&entry)) {
if(fencing_peer_active(entry)) {
count++;
}
}
return count;
}
/*!
* \internal
* \brief Create a new remote stonith op
* \param client, he local stonith client id that initaited the operation
* \param request, The request from the client that started the operation
* \param peer, Is this operation owned by another stonith peer? Operations
* owned by other peers are stored on all the stonith nodes, but only the
* owner executes the operation. All the nodes get the results to the operation
* once the owner finishes executing it.
*/
void *
create_remote_stonith_op(const char *client, xmlNode * request, gboolean peer)
{
remote_fencing_op_t *op = NULL;
xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, request, LOG_TRACE);
if (remote_op_list == NULL) {
remote_op_list = g_hash_table_new_full(crm_str_hash, g_str_equal, NULL, free_remote_op);
}
/* If this operation is owned by another node, check to make
* sure we haven't already created this operation. */
if (peer && dev) {
const char *op_id = crm_element_value(dev, F_STONITH_REMOTE_OP_ID);
CRM_CHECK(op_id != NULL, return NULL);
op = g_hash_table_lookup(remote_op_list, op_id);
if (op) {
crm_debug("%s already exists", op_id);
return op;
}
}
op = calloc(1, sizeof(remote_fencing_op_t));
crm_element_value_int(request, F_STONITH_TIMEOUT, (int *)&(op->base_timeout));
if (peer && dev) {
op->id = crm_element_value_copy(dev, F_STONITH_REMOTE_OP_ID);
} else {
op->id = crm_generate_uuid();
}
g_hash_table_replace(remote_op_list, op->id, op);
CRM_LOG_ASSERT(g_hash_table_lookup(remote_op_list, op->id) != NULL);
op->state = st_query;
op->replies_expected = fencing_active_peers();
op->action = crm_element_value_copy(dev, F_STONITH_ACTION);
op->originator = crm_element_value_copy(dev, F_STONITH_ORIGIN);
op->created = time(NULL);
if (op->originator == NULL) {
/* Local or relayed request */
op->originator = strdup(stonith_our_uname);
}
CRM_LOG_ASSERT(client != NULL);
if (client) {
op->client_id = strdup(client);
}
op->client_name = crm_element_value_copy(request, F_STONITH_CLIENTNAME);
op->target = crm_element_value_copy(dev, F_STONITH_TARGET);
op->request = copy_xml(request); /* TODO: Figure out how to avoid this */
crm_element_value_int(request, F_STONITH_CALLOPTS, (int *)&(op->call_options));
crm_trace("%s new stonith op: %s - %s of %s for %s",
(peer
&& dev) ? "Recorded" : "Generated", op->id, op->action, op->target, op->client_name);
if (op->call_options & st_opt_cs_nodeid) {
int nodeid = crm_atoi(op->target, NULL);
crm_node_t *node = crm_get_peer(nodeid, NULL);
/* Ensure the conversion only happens once */
op->call_options &= ~st_opt_cs_nodeid;
if (node && node->uname) {
free(op->target);
op->target = strdup(node->uname);
} else {
crm_warn("Could not expand nodeid '%s' into a host name (%p)", op->target, node);
}
}
/* check to see if this is a duplicate operation of another in-flight operation */
merge_duplicates(op);
return op;
}
remote_fencing_op_t *
initiate_remote_stonith_op(crm_client_t * client, xmlNode * request, gboolean manual_ack)
{
xmlNode *query = NULL;
const char *client_id = NULL;
remote_fencing_op_t *op = NULL;
if (client) {
client_id = client->id;
} else {
client_id = crm_element_value(request, F_STONITH_CLIENTID);
}
CRM_LOG_ASSERT(client_id != NULL);
op = create_remote_stonith_op(client_id, request, FALSE);
op->owner = TRUE;
CRM_CHECK(op->action, return NULL);
if (stonith_topology_next(op) != pcmk_ok) {
op->state = st_failed;
}
switch (op->state) {
case st_failed:
crm_warn("Initiation of remote operation %s for %s: failed (%s)", op->action,
op->target, op->id);
remote_op_done(op, NULL, -EINVAL, FALSE);
return op;
case st_duplicate:
crm_info("Initiating remote operation %s for %s: %s (duplicate)", op->action,
op->target, op->id);
return op;
default:
crm_notice("Initiating remote operation %s for %s: %s (%d)", op->action, op->target,
op->id, op->state);
}
query = stonith_create_op(0, op->id, STONITH_OP_QUERY, NULL, 0);
if (!manual_ack) {
int query_timeout = op->base_timeout * TIMEOUT_MULTIPLY_FACTOR;
op->query_timer = g_timeout_add((1000 * query_timeout), remote_op_query_timeout, op);
} else {
crm_xml_add(query, F_STONITH_DEVICE, "manual_ack");
}
crm_xml_add(query, F_STONITH_REMOTE_OP_ID, op->id);
crm_xml_add(query, F_STONITH_TARGET, op->target);
crm_xml_add(query, F_STONITH_ACTION, op->action);
crm_xml_add(query, F_STONITH_ORIGIN, op->originator);
crm_xml_add(query, F_STONITH_CLIENTID, op->client_id);
crm_xml_add(query, F_STONITH_CLIENTNAME, op->client_name);
crm_xml_add_int(query, F_STONITH_TIMEOUT, op->base_timeout);
send_cluster_message(NULL, crm_msg_stonith_ng, query, FALSE);
free_xml(query);
return op;
}
static gint
sort_strings(gconstpointer a, gconstpointer b)
{
return strcmp(a, b);
}
enum find_best_peer_options {
/*! Skip checking the target peer for capable fencing devices */
FIND_PEER_SKIP_TARGET = 0x0001,
/*! Only check the target peer for capable fencing devices */
FIND_PEER_TARGET_ONLY = 0x0002,
/*! Skip peers and devices that are not verified */
FIND_PEER_VERIFIED_ONLY = 0x0004,
};
static st_query_result_t *
find_best_peer(const char *device, remote_fencing_op_t * op, enum find_best_peer_options options)
{
GListPtr iter = NULL;
gboolean verified_devices_only = (options & FIND_PEER_VERIFIED_ONLY) ? TRUE : FALSE;
if (!device && is_set(op->call_options, st_opt_topology)) {
return NULL;
}
for (iter = op->query_results; iter != NULL; iter = iter->next) {
st_query_result_t *peer = iter->data;
if ((options & FIND_PEER_SKIP_TARGET) && safe_str_eq(peer->host, op->target)) {
continue;
}
if ((options & FIND_PEER_TARGET_ONLY) && safe_str_neq(peer->host, op->target)) {
continue;
}
if (is_set(op->call_options, st_opt_topology)) {
/* Do they have the next device of the current fencing level? */
GListPtr match = NULL;
if (verified_devices_only && !g_hash_table_lookup(peer->verified_devices, device)) {
continue;
}
match = g_list_find_custom(peer->device_list, device, sort_strings);
if (match) {
crm_trace("Removing %s from %s (%d remaining)", (char *)match->data, peer->host,
g_list_length(peer->device_list));
peer->device_list = g_list_remove(peer->device_list, match->data);
return peer;
}
} else if (peer->devices > 0 && peer->tried == FALSE) {
if (verified_devices_only && !g_hash_table_size(peer->verified_devices)) {
continue;
}
/* No topology: Use the current best peer */
crm_trace("Simple fencing");
return peer;
}
}
return NULL;
}
static st_query_result_t *
stonith_choose_peer(remote_fencing_op_t * op)
{
st_query_result_t *peer = NULL;
const char *device = NULL;
do {
if (op->devices) {
device = op->devices->data;
crm_trace("Checking for someone to fence %s with %s", op->target,
(char *)op->devices->data);
} else {
crm_trace("Checking for someone to fence %s", op->target);
}
if ((peer = find_best_peer(device, op, FIND_PEER_SKIP_TARGET | FIND_PEER_VERIFIED_ONLY))) {
return peer;
} else if ((peer = find_best_peer(device, op, FIND_PEER_SKIP_TARGET))) {
return peer;
} else if ((peer = find_best_peer(device, op, FIND_PEER_TARGET_ONLY))) {
return peer;
}
/* Try the next fencing level if there is one */
} while (is_set(op->call_options, st_opt_topology)
&& stonith_topology_next(op) == pcmk_ok);
if (op->devices) {
crm_debug("Couldn't find anyone to fence %s with %s", op->target,
(char *)op->devices->data);
} else {
crm_debug("Couldn't find anyone to fence %s", op->target);
}
return NULL;
}
static int
get_device_timeout(st_query_result_t * peer, const char *device, int default_timeout)
{
gpointer res;
if (!peer || !device) {
return default_timeout;
}
res = g_hash_table_lookup(peer->custom_action_timeouts, device);
return res ? GPOINTER_TO_INT(res) : default_timeout;
}
static int
get_peer_timeout(st_query_result_t * peer, int default_timeout)
{
int total_timeout = 0;
GListPtr cur = NULL;
for (cur = peer->device_list; cur; cur = cur->next) {
total_timeout += get_device_timeout(peer, cur->data, default_timeout);
}
return total_timeout ? total_timeout : default_timeout;
}
static int
get_op_total_timeout(remote_fencing_op_t * op, st_query_result_t * chosen_peer, int default_timeout)
{
stonith_topology_t *tp = g_hash_table_lookup(topology, op->target);
int total_timeout = 0;
if (is_set(op->call_options, st_opt_topology) && tp) {
int i;
GListPtr device_list = NULL;
GListPtr iter = NULL;
/* Yep, this looks scary, nested loops all over the place.
* Here is what is going on.
* Loop1: Iterate through fencing levels.
* Loop2: If a fencing level has devices, loop through each device
* Loop3: For each device in a fencing level, see what peer owns it
* and what that peer has reported the timeout is for the device.
*/
for (i = 0; i < ST_LEVEL_MAX; i++) {
if (!tp->levels[i]) {
continue;
}
for (device_list = tp->levels[i]; device_list; device_list = device_list->next) {
for (iter = op->query_results; iter != NULL; iter = iter->next) {
st_query_result_t *peer = iter->data;
if (g_list_find_custom(peer->device_list, device_list->data, sort_strings)) {
total_timeout +=
get_device_timeout(peer, device_list->data, default_timeout);
break;
}
} /* End Loop3: match device with peer that owns device, find device's timeout period */
} /* End Loop2: iterate through devices at a specific level */
} /*End Loop1: iterate through fencing levels */
} else if (chosen_peer) {
total_timeout = get_peer_timeout(chosen_peer, default_timeout);
} else {
total_timeout = default_timeout;
}
return total_timeout ? total_timeout : default_timeout;
}
static void
report_timeout_period(remote_fencing_op_t * op, int op_timeout)
{
GListPtr iter = NULL;
xmlNode *update = NULL;
const char *client_node = NULL;
const char *client_id = NULL;
const char *call_id = NULL;
if (op->call_options & st_opt_sync_call) {
/* There is no reason to report the timeout for a syncronous call. It
* is impossible to use the reported timeout to do anything when the client
* is blocking for the response. This update is only important for
* async calls that require a callback to report the results in. */
return;
} else if (!op->request) {
return;
}
crm_trace("Reporting timeout for %s.%.8s", op->client_name, op->id);
client_node = crm_element_value(op->request, F_STONITH_CLIENTNODE);
call_id = crm_element_value(op->request, F_STONITH_CALLID);
client_id = crm_element_value(op->request, F_STONITH_CLIENTID);
if (!client_node || !call_id || !client_id) {
return;
}
if (safe_str_eq(client_node, stonith_our_uname)) {
/* The client is connected to this node, send the update direclty to them */
do_stonith_async_timeout_update(client_id, call_id, op_timeout);
return;
}
/* The client is connected to another node, relay this update to them */
update = stonith_create_op(0, op->id, STONITH_OP_TIMEOUT_UPDATE, NULL, 0);
crm_xml_add(update, F_STONITH_REMOTE_OP_ID, op->id);
crm_xml_add(update, F_STONITH_CLIENTID, client_id);
crm_xml_add(update, F_STONITH_CALLID, call_id);
crm_xml_add_int(update, F_STONITH_TIMEOUT, op_timeout);
send_cluster_message(crm_get_peer(0, client_node), crm_msg_stonith_ng, update, FALSE);
free_xml(update);
for (iter = op->duplicates; iter != NULL; iter = iter->next) {
remote_fencing_op_t *dup = iter->data;
crm_trace("Reporting timeout for duplicate %s.%.8s", dup->client_name, dup->id);
report_timeout_period(iter->data, op_timeout);
}
}
void
call_remote_stonith(remote_fencing_op_t * op, st_query_result_t * peer)
{
const char *device = NULL;
int timeout = op->base_timeout;
crm_trace("State for %s.%.8s: %s %d", op->target, op->client_name, op->id, op->state);
if (peer == NULL && !is_set(op->call_options, st_opt_topology)) {
peer = stonith_choose_peer(op);
}
if (!op->op_timer_total) {
int total_timeout = get_op_total_timeout(op, peer, op->base_timeout);
op->total_timeout = TIMEOUT_MULTIPLY_FACTOR * total_timeout;
op->op_timer_total = g_timeout_add(1000 * op->total_timeout, remote_op_timeout, op);
report_timeout_period(op, op->total_timeout);
crm_info("Total remote op timeout set to %d for fencing of node %s for %s.%.8s",
total_timeout, op->target, op->client_name, op->id);
}
if (is_set(op->call_options, st_opt_topology) && op->devices) {
/* Ignore any preference, they might not have the device we need */
/* When using topology, the stonith_choose_peer function pops off
* the peer from the op's query results. Make sure to calculate
* the op_timeout before calling this function when topology is in use */
peer = stonith_choose_peer(op);
device = op->devices->data;
timeout = get_device_timeout(peer, device, op->base_timeout);
}
if (peer) {
int timeout_one = 0;
xmlNode *query = stonith_create_op(0, op->id, STONITH_OP_FENCE, NULL, 0);
crm_xml_add(query, F_STONITH_REMOTE_OP_ID, op->id);
crm_xml_add(query, F_STONITH_TARGET, op->target);
crm_xml_add(query, F_STONITH_ACTION, op->action);
crm_xml_add(query, F_STONITH_ORIGIN, op->originator);
crm_xml_add(query, F_STONITH_CLIENTID, op->client_id);
crm_xml_add(query, F_STONITH_CLIENTNAME, op->client_name);
crm_xml_add_int(query, F_STONITH_TIMEOUT, timeout);
if (device) {
timeout_one =
TIMEOUT_MULTIPLY_FACTOR * get_device_timeout(peer, device, op->base_timeout);
crm_info("Requesting that %s perform op %s %s with %s for %s (%ds)", peer->host,
op->action, op->target, device, op->client_name, timeout_one);
crm_xml_add(query, F_STONITH_DEVICE, device);
crm_xml_add(query, F_STONITH_MODE, "slave");
} else {
timeout_one = TIMEOUT_MULTIPLY_FACTOR * get_peer_timeout(peer, op->base_timeout);
crm_info("Requesting that %s perform op %s %s for %s (%ds)",
peer->host, op->action, op->target, op->client_name, timeout_one);
crm_xml_add(query, F_STONITH_MODE, "smart");
}
op->state = st_exec;
if (op->op_timer_one) {
g_source_remove(op->op_timer_one);
}
op->op_timer_one = g_timeout_add((1000 * timeout_one), remote_op_timeout_one, op);
send_cluster_message(crm_get_peer(0, peer->host), crm_msg_stonith_ng, query, FALSE);
peer->tried = TRUE;
free_xml(query);
return;
} else if (op->owner == FALSE) {
crm_err("The termination of %s for %s is not ours to control", op->target, op->client_name);
} else if (op->query_timer == 0) {
/* We've exhausted all available peers */
crm_info("No remaining peers capable of terminating %s for %s (%d)", op->target,
op->client_name, op->state);
CRM_LOG_ASSERT(op->state < st_done);
remote_op_timeout(op);
} else if(op->replies >= op->replies_expected || op->replies >= fencing_active_peers()) {
- crm_info("None of the %d peers is capable of terminating %s for %s (%d)",
+ crm_info("None of the %d peers are capable of terminating %s for %s (%d)",
op->replies, op->target, op->client_name, op->state);
op->state = st_failed;
remote_op_done(op, NULL, -EHOSTUNREACH, FALSE);
} else if (device) {
crm_info("Waiting for additional peers capable of terminating %s with %s for %s.%.8s",
op->target, device, op->client_name, op->id);
} else {
crm_info("Waiting for additional peers capable of terminating %s for %s%.8s",
op->target, op->client_name, op->id);
}
}
static gint
sort_peers(gconstpointer a, gconstpointer b)
{
const st_query_result_t *peer_a = a;
const st_query_result_t *peer_b = a;
if (peer_a->devices > peer_b->devices) {
return -1;
} else if (peer_a->devices > peer_b->devices) {
return 1;
}
return 0;
}
/*!
* \internal
* \brief Determine if all the devices in the topology are found or not
*/
static gboolean
all_topology_devices_found(remote_fencing_op_t * op)
{
GListPtr device = NULL;
GListPtr iter = NULL;
GListPtr match = NULL;
stonith_topology_t *tp = NULL;
gboolean skip_target = FALSE;
int i;
tp = g_hash_table_lookup(topology, op->target);
if (!tp) {
return FALSE;
}
if (safe_str_eq(op->action, "off") || safe_str_eq(op->action, "reboot")) {
/* Don't count the devices on the target node if we are killing
* the target node. */
skip_target = TRUE;
}
for (i = 0; i < ST_LEVEL_MAX; i++) {
for (device = tp->levels[i]; device; device = device->next) {
match = FALSE;
for (iter = op->query_results; iter != NULL; iter = iter->next) {
st_query_result_t *peer = iter->data;
if (skip_target && safe_str_eq(peer->host, op->target)) {
continue;
}
match = g_list_find_custom(peer->device_list, device->data, sort_strings);
}
if (!match) {
return FALSE;
}
}
}
return TRUE;
}
int
process_remote_stonith_query(xmlNode * msg)
{
int devices = 0;
gboolean host_is_target = FALSE;
const char *id = NULL;
const char *host = NULL;
remote_fencing_op_t *op = NULL;
st_query_result_t *result = NULL;
uint32_t active = fencing_active_peers();
xmlNode *dev = get_xpath_object("//@" F_STONITH_REMOTE_OP_ID, msg, LOG_ERR);
xmlNode *child = NULL;
CRM_CHECK(dev != NULL, return -EPROTO);
id = crm_element_value(dev, F_STONITH_REMOTE_OP_ID);
CRM_CHECK(id != NULL, return -EPROTO);
dev = get_xpath_object("//@st-available-devices", msg, LOG_ERR);
CRM_CHECK(dev != NULL, return -EPROTO);
crm_element_value_int(dev, "st-available-devices", &devices);
op = g_hash_table_lookup(remote_op_list, id);
if (op == NULL) {
crm_debug("Unknown or expired remote op: %s", id);
return -EOPNOTSUPP;
}
op->replies++;
host = crm_element_value(msg, F_ORIG);
host_is_target = safe_str_eq(host, op->target);
if (devices <= 0) {
/* If we're doing 'known' then we might need to fire anyway */
crm_trace("Query result from %s (%d devices)", host, devices);
if(op->state == st_query && (op->replies >= op->replies_expected || op->replies >= active)) {
crm_info("All queries have arrived, continuing (%d, %d, %d) ", op->replies_expected, active, op->replies);
call_remote_stonith(op, NULL);
}
return pcmk_ok;
} else if (host_is_target) {
if (op->call_options & st_opt_allow_suicide) {
crm_trace("Allowing %s to potentialy fence itself", op->target);
} else {
crm_info("Ignoring reply from %s, hosts are not permitted to commit suicide",
op->target);
return pcmk_ok;
}
}
crm_info("Query result %d of %d from %s (%d devices)", op->replies, op->replies_expected, host, devices);
result = calloc(1, sizeof(st_query_result_t));
result->host = strdup(host);
result->devices = devices;
result->custom_action_timeouts = g_hash_table_new_full(crm_str_hash, g_str_equal, free, NULL);
result->verified_devices = g_hash_table_new_full(crm_str_hash, g_str_equal, free, NULL);
for (child = __xml_first_child(dev); child != NULL; child = __xml_next(child)) {
const char *device = ID(child);
int action_timeout = 0;
int verified = 0;
if (device) {
result->device_list = g_list_prepend(result->device_list, strdup(device));
crm_element_value_int(child, F_STONITH_ACTION_TIMEOUT, &action_timeout);
crm_element_value_int(child, F_STONITH_DEVICE_VERIFIED, &verified);
if (action_timeout) {
crm_trace("Peer %s with device %s returned action timeout %d",
result->host, device, action_timeout);
g_hash_table_insert(result->custom_action_timeouts,
strdup(device), GINT_TO_POINTER(action_timeout));
}
if (verified) {
crm_trace("Peer %s has confirmed a verified device %s", result->host, device);
g_hash_table_insert(result->verified_devices,
strdup(device), GINT_TO_POINTER(verified));
}
}
}
CRM_CHECK(devices == g_list_length(result->device_list),
crm_err("Mis-match: Query claimed to have %d devices but %d found", devices,
g_list_length(result->device_list)));
op->query_results = g_list_insert_sorted(op->query_results, result, sort_peers);
if (is_set(op->call_options, st_opt_topology)) {
/* If we start the fencing before all the topology results are in,
* it is possible fencing levels will be skipped because of the missing
* query results. */
if (op->state == st_query && all_topology_devices_found(op)) {
/* All the query results are in for the topology, start the fencing ops. */
crm_trace("All topology devices found");
call_remote_stonith(op, result);
} else if(op->state == st_query && (op->replies >= op->replies_expected || op->replies >= active)) {
crm_info("All topology queries have arrived, continuing (%d, %d, %d) ", op->replies_expected, active, op->replies);
call_remote_stonith(op, NULL);
}
} else if (op->state == st_query) {
/* We have a result for a non-topology fencing op that looks promising,
* go ahead and start fencing before query timeout */
if (host_is_target == FALSE && g_hash_table_size(result->verified_devices)) {
/* we have a verified device living on a peer that is not the target */
crm_trace("Found %d verified devices", g_hash_table_size(result->verified_devices));
call_remote_stonith(op, result);
} else if (safe_str_eq(op->action, "on")) {
crm_trace("Unfencing %s", op->target);
call_remote_stonith(op, result);
} else if(op->replies >= op->replies_expected || op->replies >= active) {
crm_info("All queries have arrived, continuing (%d, %d, %d) ", op->replies_expected, active, op->replies);
call_remote_stonith(op, NULL);
} else {
crm_trace("Waiting for more peer results before launching fencing operation");
}
} else if (op->state == st_done) {
crm_info("Discarding query result from %s (%d devices): Operation is in state %d",
result->host, result->devices, op->state);
}
return pcmk_ok;
}
int
process_remote_stonith_exec(xmlNode * msg)
{
int rc = 0;
const char *id = NULL;
const char *device = NULL;
remote_fencing_op_t *op = NULL;
xmlNode *dev = get_xpath_object("//@" F_STONITH_REMOTE_OP_ID, msg, LOG_ERR);
CRM_CHECK(dev != NULL, return -EPROTO);
id = crm_element_value(dev, F_STONITH_REMOTE_OP_ID);
CRM_CHECK(id != NULL, return -EPROTO);
dev = get_xpath_object("//@" F_STONITH_RC, msg, LOG_ERR);
CRM_CHECK(dev != NULL, return -EPROTO);
crm_element_value_int(dev, F_STONITH_RC, &rc);
device = crm_element_value(dev, F_STONITH_DEVICE);
if (remote_op_list) {
op = g_hash_table_lookup(remote_op_list, id);
}
if (op == NULL && rc == pcmk_ok) {
/* Record successful fencing operations */
const char *client_id = crm_element_value(dev, F_STONITH_CLIENTID);
op = create_remote_stonith_op(client_id, dev, TRUE);
}
if (op == NULL) {
/* Could be for an event that began before we started */
/* TODO: Record the op for later querying */
crm_info("Unknown or expired remote op: %s", id);
return -EOPNOTSUPP;
}
if (op->devices && device && safe_str_neq(op->devices->data, device)) {
crm_err
("Received outdated reply for device %s (instead of %s) to %s node %s. Operation already timed out at remote level.",
device, op->devices->data, op->action, op->target);
return rc;
}
if (safe_str_eq(crm_element_value(msg, F_SUBTYPE), "broadcast")) {
crm_debug("Marking call to %s for %s on behalf of %s@%s.%.8s: %s (%d)",
op->action, op->target, op->client_name, op->id, op->originator,
pcmk_strerror(rc), rc);
if (rc == pcmk_ok) {
op->state = st_done;
} else {
op->state = st_failed;
}
remote_op_done(op, msg, rc, FALSE);
return pcmk_ok;
} else if (safe_str_neq(op->originator, stonith_our_uname)) {
/* If this isn't a remote level broadcast, and we are not the
* originator of the operation, we should not be receiving this msg. */
crm_err
("%s received non-broadcast fencing result for operation it does not own (device %s targeting %s)",
stonith_our_uname, device, op->target);
return rc;
}
if (is_set(op->call_options, st_opt_topology)) {
const char *device = crm_element_value(msg, F_STONITH_DEVICE);
crm_notice("Call to %s for %s on behalf of %s@%s: %s (%d)",
device, op->target, op->client_name, op->originator,
pcmk_strerror(rc), rc);
/* We own the op, and it is complete. broadcast the result to all nodes
* and notify our local clients. */
if (op->state == st_done) {
remote_op_done(op, msg, rc, FALSE);
return rc;
}
/* An operation completed succesfully but has not yet been marked as done.
* Continue the topology if more devices exist at the current level, otherwise
* mark as done. */
if (rc == pcmk_ok) {
if (op->devices) {
/* Success, are there any more? */
op->devices = op->devices->next;
}
/* if no more devices at this fencing level, we are done,
* else we need to contine with executing the next device in the list */
if (op->devices == NULL) {
crm_trace("Marking complex fencing op for %s as complete", op->target);
op->state = st_done;
remote_op_done(op, msg, rc, FALSE);
return rc;
}
} else {
/* This device failed, time to try another topology level. If no other
* levels are available, mark this operation as failed and report results. */
if (stonith_topology_next(op) != pcmk_ok) {
op->state = st_failed;
remote_op_done(op, msg, rc, FALSE);
return rc;
}
}
} else if (rc == pcmk_ok && op->devices == NULL) {
crm_trace("All done for %s", op->target);
op->state = st_done;
remote_op_done(op, msg, rc, FALSE);
return rc;
}
/* Retry on failure or execute the rest of the topology */
crm_trace("Next for %s on behalf of %s@%s (rc was %d)", op->target, op->originator,
op->client_name, rc);
call_remote_stonith(op, NULL);
return rc;
}
int
stonith_fence_history(xmlNode * msg, xmlNode ** output)
{
int rc = 0;
const char *target = NULL;
xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, msg, LOG_TRACE);
if (dev) {
int options = 0;
target = crm_element_value(dev, F_STONITH_TARGET);
crm_element_value_int(msg, F_STONITH_CALLOPTS, &options);
if (target && (options & st_opt_cs_nodeid)) {
int nodeid = crm_atoi(target, NULL);
crm_node_t *node = crm_get_peer(nodeid, NULL);
if (node) {
target = node->uname;
}
}
}
*output = create_xml_node(NULL, F_STONITH_HISTORY_LIST);
if (remote_op_list) {
GHashTableIter iter;
remote_fencing_op_t *op = NULL;
g_hash_table_iter_init(&iter, remote_op_list);
while (g_hash_table_iter_next(&iter, NULL, (void **)&op)) {
xmlNode *entry = NULL;
if (target && strcmp(op->target, target) != 0) {
continue;
}
rc = 0;
entry = create_xml_node(*output, STONITH_OP_EXEC);
crm_xml_add(entry, F_STONITH_TARGET, op->target);
crm_xml_add(entry, F_STONITH_ACTION, op->action);
crm_xml_add(entry, F_STONITH_ORIGIN, op->originator);
crm_xml_add(entry, F_STONITH_DELEGATE, op->delegate);
crm_xml_add_int(entry, F_STONITH_DATE, op->completed);
crm_xml_add_int(entry, F_STONITH_STATE, op->state);
}
}
return rc;
}
gboolean
stonith_check_fence_tolerance(int tolerance, const char *target, const char *action)
{
GHashTableIter iter;
time_t now = time(NULL);
remote_fencing_op_t *rop = NULL;
crm_trace("tolerance=%d, remote_op_list=%p", tolerance, remote_op_list);
if (tolerance <= 0 || !remote_op_list || target == NULL || action == NULL) {
return FALSE;
}
g_hash_table_iter_init(&iter, remote_op_list);
while (g_hash_table_iter_next(&iter, NULL, (void **)&rop)) {
if (strcmp(rop->target, target) != 0) {
continue;
} else if (rop->state != st_done) {
continue;
} else if (strcmp(rop->action, action) != 0) {
continue;
} else if ((rop->completed + tolerance) < now) {
continue;
}
crm_notice("Target %s was fenced (%s) less than %ds ago by %s on behalf of %s",
target, action, tolerance, rop->delegate, rop->originator);
return TRUE;
}
return FALSE;
}
diff --git a/fencing/test.c b/fencing/test.c
index 5ae83f5b57..7ca876524b 100644
--- a/fencing/test.c
+++ b/fencing/test.c
@@ -1,651 +1,651 @@
/*
* Copyright (C) 2009 Andrew Beekhof
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This software is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
GMainLoop *mainloop = NULL;
crm_trigger_t *trig = NULL;
int mainloop_iter = 0;
int callback_rc = 0;
typedef void (*mainloop_test_iteration_cb) (int check_event);
#define MAINLOOP_DEFAULT_TIMEOUT 2
#define mainloop_test_done(pass) \
if (pass) { \
crm_info("SUCCESS - %s", __PRETTY_FUNCTION__); \
mainloop_iter++; \
mainloop_set_trigger(trig); \
} else { \
- crm_info("FAILURE = %s async_callback %d", __PRETTY_FUNCTION__, callback_rc); \
+ crm_err("FAILURE = %s async_callback %d", __PRETTY_FUNCTION__, callback_rc); \
crm_exit(-1); \
} \
callback_rc = 0; \
/* *INDENT-OFF* */
enum test_modes {
/* class dev test using a very specific environment */
test_standard = 0,
/* watch notifications only */
test_passive,
/* sanity test stonith client api using fence_true and fence_false */
test_api_sanity,
/* sanity test mainloop code with async respones. */
test_api_mainloop,
};
static struct crm_option long_options[] = {
{"verbose", 0, 0, 'V'},
{"version", 0, 0, '$'},
{"help", 0, 0, '?'},
{"passive", 0, 0, 'p'},
{"api_test", 0, 0, 't'},
{"mainloop_api_test", 0, 0, 'm'},
{0, 0, 0, 0}
};
/* *INDENT-ON* */
stonith_t *st = NULL;
struct pollfd pollfd;
int st_opts = st_opt_sync_call;
int expected_notifications = 0;
int verbose = 0;
static void
dispatch_helper(int timeout)
{
int rc;
crm_debug("Looking for notification");
pollfd.events = POLLIN;
while (true) {
rc = poll(&pollfd, 1, timeout); /* wait 10 minutes, -1 forever */
if (rc > 0) {
if (!stonith_dispatch(st)) {
break;
}
} else {
break;
}
}
}
static void
st_callback(stonith_t * st, stonith_event_t * e)
{
if (st->state == stonith_disconnected) {
crm_exit(1);
}
crm_notice("Operation %s requested by %s %s for peer %s. %s reported: %s (ref=%s)",
e->operation, e->origin, e->result == pcmk_ok ? "completed" : "failed",
e->target, e->executioner ? e->executioner : "",
pcmk_strerror(e->result), e->id);
if (expected_notifications) {
expected_notifications--;
}
}
static void
st_global_callback(stonith_t * stonith, stonith_callback_data_t * data)
{
crm_notice("Call id %d completed with rc %d", data->call_id, data->rc);
}
static void
passive_test(void)
{
int rc = 0;
rc = st->cmds->connect(st, crm_system_name, &pollfd.fd);
crm_debug("Connect: %d", rc);
st->cmds->register_notification(st, T_STONITH_NOTIFY_DISCONNECT, st_callback);
st->cmds->register_notification(st, T_STONITH_NOTIFY_FENCE, st_callback);
st->cmds->register_notification(st, STONITH_OP_DEVICE_ADD, st_callback);
st->cmds->register_notification(st, STONITH_OP_DEVICE_DEL, st_callback);
st->cmds->register_callback(st, 0, 120, st_opt_timeout_updates, NULL, "st_global_callback",
st_global_callback);
dispatch_helper(600 * 1000);
}
#define single_test(cmd, str, num_notifications, expected_rc) \
{ \
int rc = 0; \
rc = cmd; \
expected_notifications = 0; \
if (num_notifications) { \
expected_notifications = num_notifications; \
dispatch_helper(500); \
} \
if (rc != expected_rc) { \
- crm_info("FAILURE - expected rc %d != %d(%s) for cmd - %s\n", expected_rc, rc, pcmk_strerror(rc), str); \
+ crm_err("FAILURE - expected rc %d != %d(%s) for cmd - %s\n", expected_rc, rc, pcmk_strerror(rc), str); \
crm_exit(-1); \
} else if (expected_notifications) { \
- crm_info("FAILURE - expected %d notifications, got only %d for cmd - %s\n", \
+ crm_err("FAILURE - expected %d notifications, got only %d for cmd - %s\n", \
num_notifications, num_notifications - expected_notifications, str); \
crm_exit(-1); \
} else { \
if (verbose) { \
crm_info("SUCCESS - %s: %d", str, rc); \
} else { \
crm_debug("SUCCESS - %s: %d", str, rc); \
} \
} \
}\
static void
run_fence_failure_test(void)
{
stonith_key_value_t *params = NULL;
params = stonith_key_value_add(params, "pcmk_host_map", "false_1_node1=1,2 false_1_node2=3,4");
single_test(st->
cmds->register_device(st, st_opts, "test-id1", "stonith-ng", "fence_false", params),
"Register device1 for failure test", 1, 0);
single_test(st->cmds->fence(st, st_opts, "false_1_node2", "off", 3, 0),
"Fence failure results off", 1, -62);
single_test(st->cmds->fence(st, st_opts, "false_1_node2", "reboot", 3, 0),
"Fence failure results reboot", 1, -62);
single_test(st->cmds->remove_device(st, st_opts, "test-id1"),
"Remove device1 for failure test", 1, 0);
stonith_key_value_freeall(params, 1, 1);
}
static void
run_fence_failure_rollover_test(void)
{
stonith_key_value_t *params = NULL;
params = stonith_key_value_add(params, "pcmk_host_map", "false_1_node1=1,2 false_1_node2=3,4");
single_test(st->
cmds->register_device(st, st_opts, "test-id1", "stonith-ng", "fence_false", params),
"Register device1 for rollover test", 1, 0);
single_test(st->
cmds->register_device(st, st_opts, "test-id2", "stonith-ng", "fence_true", params),
"Register device2 for rollover test", 1, 0);
single_test(st->cmds->fence(st, st_opts, "false_1_node2", "off", 3, 0),
"Fence rollover results off", 1, 0);
single_test(st->cmds->fence(st, st_opts, "false_1_node2", "on", 3, 0),
"Fence rollover results on", 1, 0);
single_test(st->cmds->remove_device(st, st_opts, "test-id1"),
"Remove device1 for rollover tests", 1, 0);
single_test(st->cmds->remove_device(st, st_opts, "test-id2"),
"Remove device2 for rollover tests", 1, 0);
stonith_key_value_freeall(params, 1, 1);
}
static void
run_standard_test(void)
{
stonith_key_value_t *params = NULL;
params = stonith_key_value_add(params, "pcmk_host_map", "false_1_node1=1,2 false_1_node2=3,4");
single_test(st->
cmds->register_device(st, st_opts, "test-id", "stonith-ng", "fence_true", params),
"Register", 1, 0);
single_test(st->cmds->list(st, st_opts, "test-id", NULL, 1), "list", 1, 0);
single_test(st->cmds->monitor(st, st_opts, "test-id", 1), "Monitor", 1, 0);
single_test(st->cmds->status(st, st_opts, "test-id", "false_1_node2", 1),
"Status false_1_node2", 1, 0);
single_test(st->cmds->status(st, st_opts, "test-id", "false_1_node1", 1),
"Status false_1_node1", 1, 0);
single_test(st->cmds->fence(st, st_opts, "unknown-host", "off", 1, 0),
- "Fence unknown-host (expected failure)", 0, -113);
+ "Fence unknown-host (expected failure)", 0, -19);
single_test(st->cmds->fence(st, st_opts, "false_1_node1", "off", 1, 0),
"Fence false_1_node1", 1, 0);
single_test(st->cmds->fence(st, st_opts, "false_1_node1", "on", 1, 0),
"Unfence false_1_node1", 1, 0);
single_test(st->cmds->remove_device(st, st_opts, "test-id"), "Remove test-id", 1, 0);
stonith_key_value_freeall(params, 1, 1);
}
static void
sanity_tests(void)
{
int rc = 0;
rc = st->cmds->connect(st, crm_system_name, &pollfd.fd);
crm_debug("Connect: %d", rc);
st->cmds->register_notification(st, T_STONITH_NOTIFY_DISCONNECT, st_callback);
st->cmds->register_notification(st, T_STONITH_NOTIFY_FENCE, st_callback);
st->cmds->register_notification(st, STONITH_OP_DEVICE_ADD, st_callback);
st->cmds->register_notification(st, STONITH_OP_DEVICE_DEL, st_callback);
st->cmds->register_callback(st, 0, 120, st_opt_timeout_updates, NULL, "st_global_callback",
st_global_callback);
crm_info("Starting API Sanity Tests");
run_standard_test();
run_fence_failure_test();
run_fence_failure_rollover_test();
crm_info("Sanity Tests Passed");
}
static void
standard_dev_test(void)
{
int rc = 0;
char *tmp = NULL;
stonith_key_value_t *params = NULL;
rc = st->cmds->connect(st, crm_system_name, &pollfd.fd);
crm_debug("Connect: %d", rc);
params = stonith_key_value_add(params, "pcmk_host_map", "some-host=pcmk-7 true_1_node1=3,4");
rc = st->cmds->register_device(st, st_opts, "test-id", "stonith-ng", "fence_xvm", params);
crm_debug("Register: %d", rc);
rc = st->cmds->list(st, st_opts, "test-id", &tmp, 10);
crm_debug("List: %d output: %s\n", rc, tmp ? tmp : "");
rc = st->cmds->monitor(st, st_opts, "test-id", 10);
crm_debug("Monitor: %d", rc);
rc = st->cmds->status(st, st_opts, "test-id", "false_1_node2", 10);
crm_debug("Status false_1_node2: %d", rc);
rc = st->cmds->status(st, st_opts, "test-id", "false_1_node1", 10);
crm_debug("Status false_1_node1: %d", rc);
rc = st->cmds->fence(st, st_opts, "unknown-host", "off", 60, 0);
crm_debug("Fence unknown-host: %d", rc);
rc = st->cmds->status(st, st_opts, "test-id", "false_1_node1", 10);
crm_debug("Status false_1_node1: %d", rc);
rc = st->cmds->fence(st, st_opts, "false_1_node1", "off", 60, 0);
crm_debug("Fence false_1_node1: %d", rc);
rc = st->cmds->status(st, st_opts, "test-id", "false_1_node1", 10);
crm_debug("Status false_1_node1: %d", rc);
rc = st->cmds->fence(st, st_opts, "false_1_node1", "on", 10, 0);
crm_debug("Unfence false_1_node1: %d", rc);
rc = st->cmds->status(st, st_opts, "test-id", "false_1_node1", 10);
crm_debug("Status false_1_node1: %d", rc);
rc = st->cmds->fence(st, st_opts, "some-host", "off", 10, 0);
crm_debug("Fence alias: %d", rc);
rc = st->cmds->status(st, st_opts, "test-id", "some-host", 10);
crm_debug("Status alias: %d", rc);
rc = st->cmds->fence(st, st_opts, "false_1_node1", "on", 10, 0);
crm_debug("Unfence false_1_node1: %d", rc);
rc = st->cmds->remove_device(st, st_opts, "test-id");
crm_debug("Remove test-id: %d", rc);
stonith_key_value_freeall(params, 1, 1);
}
static void
iterate_mainloop_tests(gboolean event_ready);
static void
mainloop_callback(stonith_t * stonith, stonith_callback_data_t * data)
{
callback_rc = data->rc;
iterate_mainloop_tests(TRUE);
}
static int
register_callback_helper(int callid)
{
return st->cmds->register_callback(st,
callid,
MAINLOOP_DEFAULT_TIMEOUT,
st_opt_timeout_updates, NULL, "callback", mainloop_callback);
}
static void
test_async_fence_pass(int check_event)
{
int rc = 0;
if (check_event) {
if (callback_rc != 0) {
mainloop_test_done(FALSE);
} else {
mainloop_test_done(TRUE);
}
return;
}
rc = st->cmds->fence(st, 0, "true_1_node1", "off", MAINLOOP_DEFAULT_TIMEOUT, 0);
if (rc < 0) {
crm_err("fence failed with rc %d", rc);
mainloop_test_done(FALSE);
}
register_callback_helper(rc);
/* wait for event */
}
#define CUSTOM_TIMEOUT_ADDITION 10
static void
test_async_fence_custom_timeout(int check_event)
{
int rc = 0;
static time_t begin = 0;
if (check_event) {
uint32_t diff = (time(NULL) - begin);
if (callback_rc != -ETIME) {
mainloop_test_done(FALSE);
} else if (diff < CUSTOM_TIMEOUT_ADDITION + MAINLOOP_DEFAULT_TIMEOUT) {
crm_err
("Custom timeout test failed, callback expiration should be updated to %d, actual timeout was %d",
CUSTOM_TIMEOUT_ADDITION + MAINLOOP_DEFAULT_TIMEOUT, diff);
mainloop_test_done(FALSE);
} else {
mainloop_test_done(TRUE);
}
return;
}
begin = time(NULL);
rc = st->cmds->fence(st, 0, "custom_timeout_node1", "off", MAINLOOP_DEFAULT_TIMEOUT, 0);
if (rc < 0) {
crm_err("fence failed with rc %d", rc);
mainloop_test_done(FALSE);
}
register_callback_helper(rc);
/* wait for event */
}
static void
test_async_fence_timeout(int check_event)
{
int rc = 0;
if (check_event) {
- if (callback_rc != -ETIME) {
+ if (callback_rc != -EHOSTUNREACH) {
mainloop_test_done(FALSE);
} else {
mainloop_test_done(TRUE);
}
return;
}
rc = st->cmds->fence(st, 0, "false_1_node2", "off", MAINLOOP_DEFAULT_TIMEOUT, 0);
if (rc < 0) {
crm_err("fence failed with rc %d", rc);
mainloop_test_done(FALSE);
}
register_callback_helper(rc);
/* wait for event */
}
static void
test_async_monitor(int check_event)
{
int rc = 0;
if (check_event) {
if (callback_rc) {
mainloop_test_done(FALSE);
} else {
mainloop_test_done(TRUE);
}
return;
}
rc = st->cmds->monitor(st, 0, "false_1", MAINLOOP_DEFAULT_TIMEOUT);
if (rc < 0) {
crm_err("monitor failed with rc %d", rc);
mainloop_test_done(FALSE);
}
register_callback_helper(rc);
/* wait for event */
}
static void
test_register_async_devices(int check_event)
{
char buf[16] = { 0, };
stonith_key_value_t *params = NULL;
params = stonith_key_value_add(params, "pcmk_host_map", "false_1_node1=1,2");
st->cmds->register_device(st, st_opts, "false_1", "stonith-ng", "fence_false", params);
stonith_key_value_freeall(params, 1, 1);
params = NULL;
params = stonith_key_value_add(params, "pcmk_host_map", "true_1_node1=1,2");
st->cmds->register_device(st, st_opts, "true_1", "stonith-ng", "fence_true", params);
stonith_key_value_freeall(params, 1, 1);
params = NULL;
params = stonith_key_value_add(params, "pcmk_host_map", "custom_timeout_node1=1,2");
snprintf(buf, sizeof(buf) - 1, "%d", MAINLOOP_DEFAULT_TIMEOUT + CUSTOM_TIMEOUT_ADDITION);
params = stonith_key_value_add(params, "pcmk_off_timeout", buf);
st->cmds->register_device(st, st_opts, "false_custom_timeout", "stonith-ng", "fence_false",
params);
stonith_key_value_freeall(params, 1, 1);
mainloop_test_done(TRUE);
}
static void
try_mainloop_connect(int check_event)
{
int tries = 10;
int i = 0;
int rc = 0;
for (i = 0; i < tries; i++) {
rc = st->cmds->connect(st, crm_system_name, NULL);
if (!rc) {
crm_info("stonith client connection established");
mainloop_test_done(TRUE);
return;
} else {
crm_info("stonith client connection failed");
}
sleep(1);
}
crm_err("API CONNECTION FAILURE\n");
mainloop_test_done(FALSE);
}
static void
iterate_mainloop_tests(gboolean event_ready)
{
static mainloop_test_iteration_cb callbacks[] = {
try_mainloop_connect,
test_register_async_devices,
test_async_monitor,
test_async_fence_pass,
test_async_fence_timeout,
test_async_fence_custom_timeout,
};
if (mainloop_iter == (sizeof(callbacks) / sizeof(mainloop_test_iteration_cb))) {
/* all tests ran, everything passed */
crm_info("ALL MAINLOOP TESTS PASSED!");
crm_exit(0);
}
callbacks[mainloop_iter] (event_ready);
}
static gboolean
trigger_iterate_mainloop_tests(gpointer user_data)
{
iterate_mainloop_tests(FALSE);
return TRUE;
}
static void
test_shutdown(int nsig)
{
int rc = 0;
if (st) {
rc = st->cmds->disconnect(st);
crm_info("Disconnect: %d", rc);
crm_debug("Destroy");
stonith_api_delete(st);
}
if (rc) {
crm_exit(-1);
}
}
static void
mainloop_tests(void)
{
trig = mainloop_add_trigger(G_PRIORITY_HIGH, trigger_iterate_mainloop_tests, NULL);
mainloop_set_trigger(trig);
mainloop_add_signal(SIGTERM, test_shutdown);
crm_info("Starting");
mainloop = g_main_new(FALSE);
g_main_run(mainloop);
}
int
main(int argc, char **argv)
{
int argerr = 0;
int flag;
int option_index = 0;
enum test_modes mode = test_standard;
crm_set_options(NULL, "mode [options]", long_options,
"Provides a summary of cluster's current state."
"\n\nOutputs varying levels of detail in a number of different formats.\n");
while (1) {
flag = crm_get_option(argc, argv, &option_index);
if (flag == -1) {
break;
}
switch (flag) {
case 'V':
verbose = 1;
break;
case '$':
case '?':
crm_help(flag, EX_OK);
break;
case 'p':
mode = test_passive;
break;
case 't':
mode = test_api_sanity;
break;
case 'm':
mode = test_api_mainloop;
break;
default:
++argerr;
break;
}
}
crm_log_init("stonith-test", LOG_INFO, TRUE, verbose ? TRUE : FALSE, argc, argv, FALSE);
if (optind > argc) {
++argerr;
}
if (argerr) {
crm_help('?', EX_USAGE);
}
crm_debug("Create");
st = stonith_api_new();
switch (mode) {
case test_standard:
standard_dev_test();
break;
case test_passive:
passive_test();
break;
case test_api_sanity:
sanity_tests();
break;
case test_api_mainloop:
mainloop_tests();
break;
}
test_shutdown(0);
return 0;
}
diff --git a/lrmd/regression.py.in b/lrmd/regression.py.in
index 7db366371c..7b2b02dd32 100755
--- a/lrmd/regression.py.in
+++ b/lrmd/regression.py.in
@@ -1,951 +1,951 @@
#!/usr/bin/python
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
import os
import sys
import subprocess
import shlex
import time
def output_from_command(command, no_wait=0):
test = subprocess.Popen(shlex.split(command), stdout=subprocess.PIPE)
if no_wait == 0:
test.wait()
else:
return 0
return test.communicate()[0].split("\n")
class Test:
def __init__(self, name, description, verbose = 0, tls = 0):
self.name = name
self.description = description
self.cmds = []
if tls:
self.daemon_location = "/usr/sbin/pacemaker_remoted"
else:
self.daemon_location = "@CRM_DAEMON_DIR@/lrmd"
self.test_tool_location = "@CRM_DAEMON_DIR@/lrmd_test"
self.verbose = verbose
self.tls = tls
self.result_txt = ""
self.cmd_tool_output = ""
self.result_exitcode = 0;
self.lrmd_process = None
self.stonith_process = None
self.executed = 0
def __new_cmd(self, cmd, args, exitcode, stdout_match = "", no_wait = 0, stdout_negative_match = "", kill=None):
if self.verbose and cmd == self.test_tool_location:
args = args + " -V "
if (cmd == self.test_tool_location) and self.tls:
args = args + " -S "
self.cmds.append(
{
"cmd" : cmd,
"kill" : kill,
"args" : args,
"expected_exitcode" : exitcode,
"stdout_match" : stdout_match,
"stdout_negative_match" : stdout_negative_match,
"no_wait" : no_wait,
"cmd_output" : "",
}
)
def start_environment(self):
### make sure we are in full control here ###
cmd = shlex.split("killall -q -9 stonithd lrmd lt-lrmd lrmd_test lt-lrmd_test pacemaker_remoted")
test = subprocess.Popen(cmd, stdout=subprocess.PIPE)
test.wait()
additional_args = ""
if self.tls == 0:
self.stonith_process = subprocess.Popen(shlex.split("@CRM_DAEMON_DIR@/stonithd -s"))
if self.verbose:
additional_args = additional_args + " -VVV "
self.lrmd_process = subprocess.Popen(shlex.split("%s %s -l /tmp/lrmd-regression.log" % (self.daemon_location, additional_args)))
time.sleep(1)
def clean_environment(self):
if self.lrmd_process:
self.lrmd_process.terminate()
self.lrmd_process.wait()
if self.verbose:
print "Daemon output"
f = open('/tmp/lrmd-regression.log', 'r')
for line in f.readlines():
print line.strip()
os.remove('/tmp/lrmd-regression.log')
if self.stonith_process:
self.stonith_process.terminate()
self.stonith_process.wait()
self.lrmd_process = None
self.stonith_process = None
def add_sys_cmd(self, cmd, args):
self.__new_cmd(cmd, args, 0, "")
def add_sys_cmd_no_wait(self, cmd, args):
self.__new_cmd(cmd, args, 0, "", 1)
def add_cmd_check_stdout(self, args, match, no_match = ""):
self.__new_cmd(self.test_tool_location, args, 0, match, 0, no_match)
def add_cmd(self, args):
self.__new_cmd(self.test_tool_location, args, 0, "")
def add_cmd_and_kill(self, killProc, args):
self.__new_cmd(self.test_tool_location, args, 0, "", kill=killProc)
def add_expected_fail_cmd(self, args):
self.__new_cmd(self.test_tool_location, args, 255, "")
def get_exitcode(self):
return self.result_exitcode
def print_result(self, filler):
print "%s%s" % (filler, self.result_txt)
def run_cmd(self, args):
cmd = shlex.split(args['args'])
cmd.insert(0, args['cmd'])
if self.verbose:
print "\n\nRunning: "+" ".join(cmd)
test = subprocess.Popen(cmd, stdout=subprocess.PIPE)
if args['kill']:
if self.verbose:
print "Also running: "+args['kill']
### Typically the kill argument is used to detect some sort of
### failure. Without yeilding for a few seconds here the process
### launched earlier that is listening for the failure may not have time
### to connect to the lrmd.
time.sleep(2)
subprocess.Popen(shlex.split(args['kill']))
if args['no_wait'] == 0:
test.wait()
else:
return 0
output = test.communicate()[0]
if args['stdout_match'] != "" and output.count(args['stdout_match']) == 0:
test.returncode = -2
print "STDOUT string '%s' was not found in cmd output" % (args['stdout_match'])
if args['stdout_negative_match'] != "" and output.count(args['stdout_negative_match']) != 0:
test.returncode = -2
print "STDOUT string '%s' was found in cmd output" % (args['stdout_negative_match'])
args['cmd_output'] = output
return test.returncode;
def run(self):
res = 0
i = 1
if self.tls and self.name.count("stonith") != 0:
self.result_txt = "SKIPPED - '%s' - disabled when testing pacemaker_remote" % (self.name)
print self.result_txt
return res
self.start_environment()
if self.verbose:
print "\n--- START TEST - %s" % self.name
self.result_txt = "SUCCESS - '%s'" % (self.name)
self.result_exitcode = 0
for cmd in self.cmds:
res = self.run_cmd(cmd)
if res != cmd['expected_exitcode']:
print cmd['cmd_output']
print "Step %d FAILED - command returned %d, expected %d" % (i, res, cmd['expected_exitcode'])
self.result_txt = "FAILURE - '%s' failed at step %d. Command: lrmd_test %s" % (self.name, i, cmd['args'])
self.result_exitcode = -1
break
else:
if self.verbose:
print cmd['cmd_output'].strip()
print "Step %d SUCCESS" % (i)
i = i + 1
self.clean_environment()
print self.result_txt
if self.verbose:
print "--- END TEST - %s\n" % self.name
self.executed = 1
return res
class Tests:
def __init__(self, verbose = 0, tls = 0):
self.tests = []
self.verbose = verbose
self.tls = tls;
self.rsc_classes = output_from_command("crm_resource --list-standards")
self.rsc_classes = self.rsc_classes[:-1] # Strip trailing empty line
self.need_authkey = 0
if self.tls:
self.rsc_classes.remove("stonith")
print "Testing "+repr(self.rsc_classes)
self.common_cmds = {
"ocf_reg_line" : "-c register_rsc -r ocf_test_rsc -t 3000 -C ocf -P pacemaker -T Dummy",
"ocf_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:ocf_test_rsc action:none rc:ok op_status:complete\"",
"ocf_unreg_line" : "-c unregister_rsc -r \"ocf_test_rsc\" -t 3000",
"ocf_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:ocf_test_rsc action:none rc:ok op_status:complete\"",
"ocf_start_line" : "-c exec -r \"ocf_test_rsc\" -a \"start\" -t 3000 ",
"ocf_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:ocf_test_rsc action:start rc:ok op_status:complete\" ",
"ocf_stop_line" : "-c exec -r \"ocf_test_rsc\" -a \"stop\" -t 3000 ",
"ocf_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:ocf_test_rsc action:stop rc:ok op_status:complete\" ",
"ocf_monitor_line" : "-c exec -r \"ocf_test_rsc\" -a \"monitor\" -i \"2000\" -t 3000",
"ocf_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:ocf_test_rsc action:monitor rc:ok op_status:complete\" -t 3000",
"ocf_cancel_line" : "-c cancel -r \"ocf_test_rsc\" -a \"monitor\" -i \"2000\" -t \"3000\" ",
"ocf_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:ocf_test_rsc action:monitor rc:ok op_status:Cancelled\" ",
"systemd_reg_line" : "-c register_rsc -r systemd_test_rsc -t 3000 -C systemd -T lrmd_dummy_daemon",
"systemd_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:systemd_test_rsc action:none rc:ok op_status:complete\"",
"systemd_unreg_line" : "-c unregister_rsc -r \"systemd_test_rsc\" -t 3000",
"systemd_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:systemd_test_rsc action:none rc:ok op_status:complete\"",
"systemd_start_line" : "-c exec -r \"systemd_test_rsc\" -a \"start\" -t 3000 ",
"systemd_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:systemd_test_rsc action:start rc:ok op_status:complete\" ",
"systemd_stop_line" : "-c exec -r \"systemd_test_rsc\" -a \"stop\" -t 3000 ",
"systemd_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:systemd_test_rsc action:stop rc:ok op_status:complete\" ",
"systemd_monitor_line" : "-c exec -r \"systemd_test_rsc\" -a \"monitor\" -i \"2000\" -t 3000",
"systemd_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:systemd_test_rsc action:monitor rc:ok op_status:complete\" -t 3000",
"systemd_cancel_line" : "-c cancel -r \"systemd_test_rsc\" -a \"monitor\" -i \"2000\" -t \"3000\" ",
"systemd_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:systemd_test_rsc action:monitor rc:ok op_status:Cancelled\" ",
"upstart_reg_line" : "-c register_rsc -r upstart_test_rsc -t 3000 -C upstart -T lrmd_dummy_daemon",
"upstart_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:upstart_test_rsc action:none rc:ok op_status:complete\"",
"upstart_unreg_line" : "-c unregister_rsc -r \"upstart_test_rsc\" -t 3000",
"upstart_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:upstart_test_rsc action:none rc:ok op_status:complete\"",
"upstart_start_line" : "-c exec -r \"upstart_test_rsc\" -a \"start\" -t 3000 ",
"upstart_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:upstart_test_rsc action:start rc:ok op_status:complete\" ",
"upstart_stop_line" : "-c exec -r \"upstart_test_rsc\" -a \"stop\" -t 3000 ",
"upstart_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:upstart_test_rsc action:stop rc:ok op_status:complete\" ",
"upstart_monitor_line" : "-c exec -r \"upstart_test_rsc\" -a \"monitor\" -i \"2000\" -t 3000",
"upstart_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:upstart_test_rsc action:monitor rc:ok op_status:complete\" -t 3000",
"upstart_cancel_line" : "-c cancel -r \"upstart_test_rsc\" -a \"monitor\" -i \"2000\" -t \"3000\" ",
"upstart_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:upstart_test_rsc action:monitor rc:ok op_status:Cancelled\" ",
"service_reg_line" : "-c register_rsc -r service_test_rsc -t 3000 -C service -T lrmd_dummy_daemon",
"service_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:service_test_rsc action:none rc:ok op_status:complete\"",
"service_unreg_line" : "-c unregister_rsc -r \"service_test_rsc\" -t 3000",
"service_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:service_test_rsc action:none rc:ok op_status:complete\"",
"service_start_line" : "-c exec -r \"service_test_rsc\" -a \"start\" -t 3000 ",
"service_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:service_test_rsc action:start rc:ok op_status:complete\" ",
"service_stop_line" : "-c exec -r \"service_test_rsc\" -a \"stop\" -t 3000 ",
"service_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:service_test_rsc action:stop rc:ok op_status:complete\" ",
"service_monitor_line" : "-c exec -r \"service_test_rsc\" -a \"monitor\" -i \"2000\" -t 3000",
"service_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:service_test_rsc action:monitor rc:ok op_status:complete\" -t 3000",
"service_cancel_line" : "-c cancel -r \"service_test_rsc\" -a \"monitor\" -i \"2000\" -t \"3000\" ",
"service_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:service_test_rsc action:monitor rc:ok op_status:Cancelled\" ",
"lsb_reg_line" : "-c register_rsc -r lsb_test_rsc -t 3000 -C lsb -T lrmd_dummy_daemon",
"lsb_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:lsb_test_rsc action:none rc:ok op_status:complete\" ",
"lsb_unreg_line" : "-c unregister_rsc -r \"lsb_test_rsc\" -t 3000",
"lsb_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:lsb_test_rsc action:none rc:ok op_status:complete\"",
"lsb_start_line" : "-c exec -r \"lsb_test_rsc\" -a \"start\" -t 3000 ",
"lsb_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:lsb_test_rsc action:start rc:ok op_status:complete\" ",
"lsb_stop_line" : "-c exec -r \"lsb_test_rsc\" -a \"stop\" -t 3000 ",
"lsb_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:lsb_test_rsc action:stop rc:ok op_status:complete\" ",
"lsb_monitor_line" : "-c exec -r \"lsb_test_rsc\" -a status -i \"2000\" -t 3000",
"lsb_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:lsb_test_rsc action:status rc:ok op_status:complete\" -t 3000",
"lsb_cancel_line" : "-c cancel -r \"lsb_test_rsc\" -a \"status\" -i \"2000\" -t \"3000\" ",
"lsb_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:lsb_test_rsc action:status rc:ok op_status:Cancelled\" ",
"stonith_reg_line" : "-c register_rsc -r stonith_test_rsc -t 3000 -C stonith -P pacemaker -T fence_dummy_monitor",
"stonith_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:stonith_test_rsc action:none rc:ok op_status:complete\" ",
"stonith_unreg_line" : "-c unregister_rsc -r \"stonith_test_rsc\" -t 3000",
"stonith_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:stonith_test_rsc action:none rc:ok op_status:complete\"",
"stonith_start_line" : "-c exec -r \"stonith_test_rsc\" -a \"start\" -t 8000 ",
"stonith_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:start rc:ok op_status:complete\" ",
"stonith_stop_line" : "-c exec -r \"stonith_test_rsc\" -a \"stop\" -t 3000 ",
"stonith_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:stop rc:ok op_status:complete\" ",
"stonith_monitor_line" : "-c exec -r \"stonith_test_rsc\" -a \"monitor\" -i \"2000\" -t 3000",
"stonith_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:monitor rc:ok op_status:complete\" -t 3000",
"stonith_cancel_line" : "-c cancel -r \"stonith_test_rsc\" -a \"monitor\" -i \"2000\" -t \"3000\" ",
"stonith_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:monitor rc:ok op_status:Cancelled\" ",
}
def new_test(self, name, description):
test = Test(name, description, self.verbose, self.tls)
self.tests.append(test)
return test
def setup_test_environment(self):
os.system("service pacemaker_remote stop")
self.cleanup_test_environment()
if self.tls and not os.path.isfile("/etc/pacemaker/authkey"):
self.need_authkey = 1
os.system("mkdir -p /etc/pacemaker")
os.system("dd if=/dev/urandom of=/etc/pacemaker/authkey bs=4096 count=1")
### Make fake systemd daemon and unit file ###
dummy_daemon = "#!/bin/bash\nwhile true\ndo\nsleep 5\ndone"
dummy_service_file = ("[Unit]\n"
"Description=Dummy Resource\n"
"[Service]\n"
"Type=simple\n"
"ExecStart=/usr/sbin/lrmd_dummy_daemon\n")
dummy_upstart_job = ("""
description "Dummy service for regression tests"
exec dd if=/dev/random of=/dev/null
""")
dummy_fence_sleep_agent = ("""#!/usr/bin/python
import sys
import time
def main():
for line in sys.stdin.readlines():
if line.count("monitor") > 0:
time.sleep(30000)
sys.exit(0)
sys.exit(-1)
if __name__ == "__main__":
main()
""")
dummy_fence_agent = ("""#!/usr/bin/python
import sys
def main():
for line in sys.stdin.readlines():
if line.count("monitor") > 0:
sys.exit(0)
if line.count("metadata") > 0:
print ''
print ' dummy description.'
print ' http://www.example.com'
print ' '
print ' '
print ' '
print ' '
print ' Fencing Action'
print ' '
print ' '
print ' '
print ' '
print ' Physical plug number or name of virtual machine'
print ' '
print ' '
print ' '
print ' '
print ' '
print ' '
print ' '
print ' '
print ''
sys.exit(0)
sys.exit(-1)
if __name__ == "__main__":
main()
""")
os.system("cat <<-END >>/etc/init/lrmd_dummy_daemon.conf\n%s\nEND" % (dummy_upstart_job))
os.system("cat <<-END >>/usr/sbin/lrmd_dummy_daemon\n%s\nEND" % (dummy_daemon))
os.system("cat <<-END >>/lib/systemd/system/lrmd_dummy_daemon.service\n%s\nEND" % (dummy_service_file))
os.system("chmod u+x /usr/sbin/lrmd_dummy_daemon")
os.system("cat <<-END >>/usr/sbin/fence_dummy_sleep\n%s\nEND" % (dummy_fence_sleep_agent))
os.system("chmod 711 /usr/sbin/fence_dummy_sleep")
os.system("cat <<-END >>/usr/sbin/fence_dummy_monitor\n%s\nEND" % (dummy_fence_agent))
os.system("chmod 711 /usr/sbin/fence_dummy_monitor")
os.system("cp /usr/share/pacemaker/tests/cts/LSBDummy /etc/init.d/lrmd_dummy_daemon")
os.system("mkdir -p @CRM_CORE_DIR@/root")
os.system("systemctl daemon-reload")
def cleanup_test_environment(self):
if self.need_authkey:
os.system("rm -f /etc/pacemaker/authkey")
os.system("rm -f /lib/systemd/system/lrmd_dummy_daemon.service")
os.system("rm -f /etc/init.d/lrmd_dummy_daemon")
os.system("rm -f /usr/sbin/lrmd_dummy_daemon")
os.system("rm -f /usr/sbin/fence_dummy_monitor")
os.system("rm -f /usr/sbin/fence_dummy_sleep")
os.system("systemctl daemon-reload")
### These are tests that should apply to all resource classes ###
def build_generic_tests(self):
common_cmds = self.common_cmds
### register/unregister tests ###
for rsc in self.rsc_classes:
test = self.new_test("generic_registration_%s" % (rsc), "Simple resource registration test for %s standard" % (rsc))
test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)])
test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)])
### start/stop tests ###
for rsc in self.rsc_classes:
test = self.new_test("generic_start_stop_%s" % (rsc), "Simple start and stop test for %s standard" % (rsc))
test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)])
test.add_cmd(common_cmds["%s_start_line" % (rsc)] + " " + common_cmds["%s_start_event" % (rsc)])
test.add_cmd(common_cmds["%s_stop_line" % (rsc)] + " " + common_cmds["%s_stop_event" % (rsc)])
test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)])
### monitor cancel test ###
for rsc in self.rsc_classes:
test = self.new_test("generic_monitor_cancel_%s" % (rsc), "Simple monitor cancel test for %s standard" % (rsc))
test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)])
test.add_cmd(common_cmds["%s_start_line" % (rsc)] + " " + common_cmds["%s_start_event" % (rsc)])
test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)])
test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled ####
test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled ####
test.add_cmd(common_cmds["%s_cancel_line" % (rsc)] + " " + common_cmds["%s_cancel_event" % (rsc)])
test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this happens the monitor did not actually cancel correctly. ###
test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this happens the monitor did not actually cancel correctly. ###
test.add_cmd(common_cmds["%s_stop_line" % (rsc)] + " " + common_cmds["%s_stop_event" % (rsc)])
test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)])
### stop implies cancel test ###
for rsc in self.rsc_classes:
test = self.new_test("generic_stop_implies_cancel_%s" % (rsc), "Verify stopping a resource implies cancel of recurring ops for %s standard" % (rsc))
test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)])
test.add_cmd(common_cmds["%s_start_line" % (rsc)] + " " + common_cmds["%s_start_event" % (rsc)])
test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)])
test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled ####
test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled ####
test.add_cmd(common_cmds["%s_stop_line" % (rsc)] + " " + common_cmds["%s_stop_event" % (rsc)])
test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this happens the monitor did not actually cancel correctly. ###
test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this happens the monitor did not actually cancel correctly. ###
test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)])
### These are complex tests that involve managing multiple resouces of different types ###
def build_multi_rsc_tests(self):
common_cmds = self.common_cmds
# do not use service and systemd at the same time, it is the same resource.
### register start monitor stop unregister resources of each type at the same time. ###
test = self.new_test("multi_rsc_start_stop_all", "Start, monitor, and stop resources of multiple types and classes")
for rsc in self.rsc_classes:
test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)])
for rsc in self.rsc_classes:
test.add_cmd(common_cmds["%s_start_line" % (rsc)] + " " + common_cmds["%s_start_event" % (rsc)])
for rsc in self.rsc_classes:
test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)])
for rsc in self.rsc_classes:
test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor is not being rescheduled ####
for rsc in self.rsc_classes:
test.add_cmd(common_cmds["%s_cancel_line" % (rsc)] + " " + common_cmds["%s_cancel_event" % (rsc)])
for rsc in self.rsc_classes:
test.add_cmd(common_cmds["%s_stop_line" % (rsc)] + " " + common_cmds["%s_stop_event" % (rsc)])
for rsc in self.rsc_classes:
test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)])
### These are tests related to how the lrmd handles failures. ###
def build_negative_tests(self):
### ocf start timeout test ###
test = self.new_test("ocf_start_timeout", "Force start timeout to occur, verify start failure.")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -k \"op_sleep\" -v \"3\" -t 3000 -w")
test.add_cmd("-l "
"\"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:unknown error op_status:Timed Out\" -t 3000")
test.add_cmd("-c exec -r test_rsc -a stop -t 3000"
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:stop rc:ok op_status:complete\" ")
test.add_cmd("-c unregister_rsc -r test_rsc -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### stonith start timeout test ###
test = self.new_test("stonith_start_timeout", "Force start timeout to occur, verify start failure.")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"stonith\" -P \"pacemaker\" -T \"fence_dummy_sleep\" -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 -w")
test.add_cmd("-l "
"\"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:unknown error op_status:Timed Out\" -t 4000")
test.add_cmd("-c exec -r test_rsc -a stop -t 3000"
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:stop rc:ok op_status:complete\" ")
test.add_cmd("-c unregister_rsc -r test_rsc -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### stonith component fail ###
common_cmds = self.common_cmds
test = self.new_test("stonith_component_fail", "Kill stonith component after lrmd connects")
test.add_cmd(common_cmds["stonith_reg_line"] + " " + common_cmds["stonith_reg_event"])
test.add_cmd(common_cmds["stonith_start_line"] + " " + common_cmds["stonith_start_event"])
test.add_cmd("-c exec -r \"stonith_test_rsc\" -a \"monitor\" -i \"600000\" "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd_and_kill("killall -9 -q stonithd" ,"-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:monitor rc:unknown error op_status:error\" -t 15000")
test.add_cmd(common_cmds["stonith_unreg_line"] + " " + common_cmds["stonith_unreg_event"])
### monitor fail for ocf resources ###
test = self.new_test("monitor_fail_ocf", "Force ocf monitor to fail, verify failure is reported.")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"monitor\" -i \"100\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ")
test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
- test.add_cmd_and_kill("rm -f /var/run/Dummy-test_rsc.state", "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 6000")
+ test.add_cmd_and_kill("rm -f @localstatedir@/run/Dummy-test_rsc.state", "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 6000")
test.add_cmd("-c cancel -r \"test_rsc\" -a \"monitor\" -i \"100\" -t \"3000\" "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ")
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 3000")
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### verify notify changes only for monitor operation. ###
test = self.new_test("monitor_changes_only", "Verify when flag is set, only monitor changes are notified.")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 -o "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"monitor\" -i \"100\" -t 3000 -o "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ")
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
- test.add_cmd_and_kill("rm -f /var/run/Dummy-test_rsc.state", "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 6000")
+ test.add_cmd_and_kill("rm -f @localstatedir@/run/Dummy-test_rsc.state", "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 6000")
test.add_cmd("-c cancel -r \"test_rsc\" -a \"monitor\" -i \"100\" -t \"3000\" "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ")
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 3000")
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### monitor fail for systemd resource ###
if "systemd" in self.rsc_classes:
test = self.new_test("monitor_fail_systemd", "Force systemd monitor to fail, verify failure is reported..")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C systemd -T lrmd_dummy_daemon -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"monitor\" -i \"100\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ")
test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd_and_kill("killall -9 -q lrmd_dummy_daemon", "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 8000")
test.add_cmd("-c cancel -r \"test_rsc\" -a \"monitor\" -i \"100\" -t \"3000\" "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ")
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 3000")
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### monitor fail for upstart resource ###
if "upstart" in self.rsc_classes:
test = self.new_test("monitor_fail_upstart", "Force upstart monitor to fail, verify failure is reported..")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C upstart -T lrmd_dummy_daemon -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"monitor\" -i \"100\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ")
test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd_and_kill("killall -9 -q dd", "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 8000")
test.add_cmd("-c cancel -r \"test_rsc\" -a \"monitor\" -i \"100\" -t \"3000\" "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ")
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" -t 3000")
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### Cancel non-existent operation on a resource ###
test = self.new_test("cancel_non_existent_op", "Attempt to cancel the wrong monitor operation, verify expected failure")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"monitor\" -i \"100\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ")
test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_expected_fail_cmd("-c cancel -r test_rsc -a \"monitor\" -i 1234 -t \"3000\" " ### interval is wrong, should fail
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ")
test.add_expected_fail_cmd("-c cancel -r test_rsc -a stop -i 100 -t \"3000\" " ### action name is wrong, should fail
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### Attempt to invoke non-existent rsc id ###
test = self.new_test("invoke_non_existent_rsc", "Attempt to perform operations on a non-existent rsc id.")
test.add_expected_fail_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:unknown error op_status:complete\" ")
test.add_expected_fail_cmd("-c exec -r test_rsc -a stop -t 3000"
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:stop rc:ok op_status:complete\" ")
test.add_expected_fail_cmd("-c exec -r test_rsc -a monitor -i 3000 -t 3000"
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ")
test.add_expected_fail_cmd("-c cancel -r test_rsc -a start -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:Cancelled\" ")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### Register and start a resource that doesn't exist, systemd ###
if "systemd" in self.rsc_classes:
test = self.new_test("start_uninstalled_systemd", "Register uninstalled systemd agent, try to start, verify expected failure")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C systemd -T this_is_fake1234 -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:not installed op_status:complete\" ")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
if "upstart" in self.rsc_classes:
test = self.new_test("start_uninstalled_upstart", "Register uninstalled upstart agent, try to start, verify expected failure")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C upstart -T this_is_fake1234 -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:not installed op_status:complete\" ")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### Register and start a resource that doesn't exist, ocf ###
test = self.new_test("start_uninstalled_ocf", "Register uninstalled ocf agent, try to start, verify expected failure.")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C ocf -P pacemaker -T this_is_fake1234 -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:not installed op_status:complete\" ")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### Register ocf with non-existent provider ###
test = self.new_test("start_ocf_bad_provider", "Register ocf agent with a non-existent provider, verify expected failure.")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C ocf -P pancakes -T Dummy -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:not installed op_status:complete\" ")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### Register ocf with empty provider field ###
test = self.new_test("start_ocf_no_provider", "Register ocf agent with a no provider, verify expected failure.")
test.add_expected_fail_cmd("-c register_rsc -r \"test_rsc\" -C ocf -T Dummy -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_expected_fail_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:Error\" ")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### These are tests that target specific cases ###
def build_custom_tests(self):
### start delay then stop test ###
test = self.new_test("start_delay", "Verify start delay works as expected.")
test.add_cmd("-c register_rsc -r test_rsc -P pacemaker -C ocf -T Dummy "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" -t 3000")
test.add_cmd("-c exec -r test_rsc -s 6000 -a start -w -t 6000")
test.add_expected_fail_cmd("-l "
"\"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" -t 2000")
test.add_cmd("-l "
"\"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" -t 6000")
test.add_cmd("-c exec -r test_rsc -a stop -t 3000"
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:stop rc:ok op_status:complete\" ")
test.add_cmd("-c unregister_rsc -r test_rsc -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### start delay, but cancel before it gets a chance to start. ###
test = self.new_test("start_delay_cancel", "Using start_delay, start a rsc, but cancel the start op before execution.")
test.add_cmd("-c register_rsc -r test_rsc -P pacemaker -C ocf -T Dummy "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" -t 3000")
test.add_cmd("-c exec -r test_rsc -s 5000 -a start -w -t 4000")
test.add_cmd("-c cancel -r test_rsc -a start -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:Cancelled\" ")
test.add_expected_fail_cmd("-l "
"\"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" -t 5000")
test.add_cmd("-c unregister_rsc -r test_rsc -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### Register a bunch of resources, verify we can get info on them ###
test = self.new_test("verify_get_rsc_info", "Register multiple resources, verify retrieval of rsc info.")
if "systemd" in self.rsc_classes:
test.add_cmd("-c register_rsc -r rsc1 -C systemd -T lrmd_dummy_daemon -t 3000 ")
test.add_cmd("-c get_rsc_info -r rsc1 ")
test.add_cmd("-c unregister_rsc -r rsc1 -t 3000 ")
test.add_expected_fail_cmd("-c get_rsc_info -r rsc1 ")
if "upstart" in self.rsc_classes:
test.add_cmd("-c register_rsc -r rsc1 -C upstart -T lrmd_dummy_daemon -t 3000 ")
test.add_cmd("-c get_rsc_info -r rsc1 ")
test.add_cmd("-c unregister_rsc -r rsc1 -t 3000 ")
test.add_expected_fail_cmd("-c get_rsc_info -r rsc1 ")
test.add_cmd("-c register_rsc -r rsc2 -C ocf -T Dummy -P pacemaker -t 3000 ")
test.add_cmd("-c get_rsc_info -r rsc2 ")
test.add_cmd("-c unregister_rsc -r rsc2 -t 3000 ")
test.add_expected_fail_cmd("-c get_rsc_info -r rsc2 ")
### Register duplicate, verify only one entry exists and can still be removed.
test = self.new_test("duplicate_registration", "Register resource multiple times, verify only one entry exists and can be removed.")
test.add_cmd("-c register_rsc -r rsc2 -C ocf -T Dummy -P pacemaker -t 3000 ")
test.add_cmd_check_stdout("-c get_rsc_info -r rsc2 ", "id:rsc2 class:ocf provider:pacemaker type:Dummy")
test.add_cmd("-c register_rsc -r rsc2 -C ocf -T Dummy -P pacemaker -t 3000 ")
test.add_cmd_check_stdout("-c get_rsc_info -r rsc2 ", "id:rsc2 class:ocf provider:pacemaker type:Dummy")
test.add_cmd("-c register_rsc -r rsc2 -C ocf -T Stateful -P pacemaker -t 3000 ")
test.add_cmd_check_stdout("-c get_rsc_info -r rsc2 ", "id:rsc2 class:ocf provider:pacemaker type:Stateful")
test.add_cmd("-c unregister_rsc -r rsc2 -t 3000 ")
test.add_expected_fail_cmd("-c get_rsc_info -r rsc2 ")
### verify the option to only send notification to the original client. ###
test = self.new_test("notify_orig_client_only", "Verify option to only send notifications to the client originating the action.")
test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" -t 3000 "
"-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -t 3000 "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ")
test.add_cmd("-c exec -r \"test_rsc\" -a \"monitor\" -i \"100\" -t 3000 -n "
"-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ")
# this will fail because the monitor notifications should only go to the original caller, which no longer exists.
test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" -t 3000")
test.add_cmd("-c cancel -r \"test_rsc\" -a \"monitor\" -i \"100\" -t \"3000\" ")
test.add_cmd("-c unregister_rsc -r \"test_rsc\" -t 3000 "
"-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ")
### get metadata ###
test = self.new_test("get_ocf_metadata", "Retrieve metadata for a resource")
test.add_cmd_check_stdout("-c metadata -C \"ocf\" -P \"pacemaker\" -T \"Dummy\""
,"resource-agent name=\"Dummy\"")
test.add_cmd("-c metadata -C \"ocf\" -P \"pacemaker\" -T \"Stateful\"")
test.add_expected_fail_cmd("-c metadata -P \"pacemaker\" -T \"Stateful\"")
test.add_expected_fail_cmd("-c metadata -C \"ocf\" -P \"pacemaker\" -T \"fake_agent\"")
### get metadata ###
test = self.new_test("get_lsb_metadata", "Retrieve metadata for a resource")
test.add_cmd_check_stdout("-c metadata -C \"lsb\" -T \"lrmd_dummy_daemon\""
,"resource-agent name='lrmd_dummy_daemon'")
### get stonith metadata ###
test = self.new_test("get_stonith_metadata", "Retrieve stonith metadata for a resource")
test.add_cmd_check_stdout("-c metadata -C \"stonith\" -P \"pacemaker\" -T \"fence_dummy_monitor\"",
"resource-agent name=\"fence_dummy_monitor\"")
### get metadata ###
if "systemd" in self.rsc_classes:
test = self.new_test("get_systemd_metadata", "Retrieve metadata for a resource")
test.add_cmd_check_stdout("-c metadata -C \"systemd\" -T \"lrmd_dummy_daemon\""
,"resource-agent name=\"lrmd_dummy_daemon\"")
### get metadata ###
if "upstart" in self.rsc_classes:
test = self.new_test("get_systemd_metadata", "Retrieve metadata for a resource")
test.add_cmd_check_stdout("-c metadata -C \"systemd\" -T \"lrmd_dummy_daemon\""
,"resource-agent name=\"lrmd_dummy_daemon\"")
### get ocf providers ###
test = self.new_test("list_ocf_providers", "Retrieve list of available resource providers, verifies pacemaker is a provider.")
test.add_cmd_check_stdout("-c list_ocf_providers ", "pacemaker")
test.add_cmd_check_stdout("-c list_ocf_providers -T ping", "pacemaker")
### Verify agents only exist in their lists ###
test = self.new_test("verify_agent_lists", "Verify the agent lists contain the right data.")
test.add_cmd_check_stdout("-c list_agents ", "Stateful") ### ocf ###
test.add_cmd_check_stdout("-c list_agents -C ocf", "Stateful")
test.add_cmd_check_stdout("-c list_agents -C lsb", "", "Stateful") ### should not exist
test.add_cmd_check_stdout("-c list_agents -C service", "", "Stateful") ### should not exist
test.add_cmd_check_stdout("-c list_agents ", "lrmd_dummy_daemon") ### init.d ###
test.add_cmd_check_stdout("-c list_agents -C lsb", "lrmd_dummy_daemon")
test.add_cmd_check_stdout("-c list_agents -C service", "lrmd_dummy_daemon")
test.add_cmd_check_stdout("-c list_agents -C ocf", "", "lrmd_dummy_daemon") ### should not exist
test.add_cmd_check_stdout("-c list_agents -C ocf", "", "lrmd_dummy_daemon") ### should not exist
test.add_cmd_check_stdout("-c list_agents -C lsb", "", "fence_dummy_monitor") ### should not exist
test.add_cmd_check_stdout("-c list_agents -C service", "", "fence_dummy_monitor") ### should not exist
test.add_cmd_check_stdout("-c list_agents -C ocf", "", "fence_dummy_monitor") ### should not exist
if "systemd" in self.rsc_classes:
test.add_cmd_check_stdout("-c list_agents ", "lrmd_dummy_daemon") ### systemd ###
test.add_cmd_check_stdout("-c list_agents -C service", "lrmd_dummy_daemon")
test.add_cmd_check_stdout("-c list_agents -C systemd", "", "Stateful") ### should not exist
test.add_cmd_check_stdout("-c list_agents -C systemd", "lrmd_dummy_daemon")
test.add_cmd_check_stdout("-c list_agents -C systemd", "", "fence_dummy_monitor") ### should not exist
if "upstart" in self.rsc_classes:
test.add_cmd_check_stdout("-c list_agents ", "lrmd_dummy_daemon") ### upstart ###
test.add_cmd_check_stdout("-c list_agents -C service", "lrmd_dummy_daemon")
test.add_cmd_check_stdout("-c list_agents -C upstart", "", "Stateful") ### should not exist
test.add_cmd_check_stdout("-c list_agents -C upstart", "lrmd_dummy_daemon")
test.add_cmd_check_stdout("-c list_agents -C upstart", "", "fence_dummy_monitor") ### should not exist
if "stonith" in self.rsc_classes:
test.add_cmd_check_stdout("-c list_agents -C stonith", "fence_dummy_monitor") ### stonith ###
test.add_cmd_check_stdout("-c list_agents -C stonith", "", "lrmd_dummy_daemon") ### should not exist
test.add_cmd_check_stdout("-c list_agents -C stonith", "", "Stateful") ### should not exist
test.add_cmd_check_stdout("-c list_agents ", "fence_dummy_monitor")
def print_list(self):
print "\n==== %d TESTS FOUND ====" % (len(self.tests))
print "%35s - %s" % ("TEST NAME", "TEST DESCRIPTION")
print "%35s - %s" % ("--------------------", "--------------------")
for test in self.tests:
print "%35s - %s" % (test.name, test.description)
print "==== END OF LIST ====\n"
def run_single(self, name):
for test in self.tests:
if test.name == name:
test.run()
break;
def run_tests_matching(self, pattern):
for test in self.tests:
if test.name.count(pattern) != 0:
test.run()
def run_tests(self):
for test in self.tests:
test.run()
def exit(self):
for test in self.tests:
if test.executed == 0:
continue
if test.get_exitcode() != 0:
sys.exit(-1)
sys.exit(0);
def print_results(self):
failures = 0;
success = 0;
print "\n\n======= FINAL RESULTS =========="
print "\n--- FAILURE RESULTS:"
for test in self.tests:
if test.executed == 0:
continue
if test.get_exitcode() != 0:
failures = failures + 1
test.print_result(" ")
else:
success = success + 1
if failures == 0:
print " None"
print "\n--- TOTALS\n Pass:%d\n Fail:%d\n" % (success, failures)
class TestOptions:
def __init__(self):
self.options = {}
self.options['list-tests'] = 0
self.options['run-all'] = 1
self.options['run-only'] = ""
self.options['run-only-pattern'] = ""
self.options['verbose'] = 0
self.options['invalid-arg'] = ""
self.options['show-usage'] = 0
self.options['pacemaker-remote'] = 0
def build_options(self, argv):
args = argv[1:]
skip = 0
for i in range(0, len(args)):
if skip:
skip = 0
continue
elif args[i] == "-h" or args[i] == "--help":
self.options['show-usage'] = 1
elif args[i] == "-l" or args[i] == "--list-tests":
self.options['list-tests'] = 1
elif args[i] == "-V" or args[i] == "--verbose":
self.options['verbose'] = 1
elif args[i] == "-R" or args[i] == "--pacemaker-remote":
self.options['pacemaker-remote'] = 1
elif args[i] == "-r" or args[i] == "--run-only":
self.options['run-only'] = args[i+1]
skip = 1
elif args[i] == "-p" or args[i] == "--run-only-pattern":
self.options['run-only-pattern'] = args[i+1]
skip = 1
def show_usage(self):
print "usage: " + sys.argv[0] + " [options]"
print "If no options are provided, all tests will run"
print "Options:"
print "\t [--help | -h] Show usage"
print "\t [--list-tests | -l] Print out all registered tests."
print "\t [--run-only | -r 'testname'] Run a specific test"
print "\t [--verbose | -V] Verbose output"
print "\t [--pacemaker-remote | -R Test pacemaker-remote binary instead of lrmd."
print "\t [--run-only-pattern | -p 'string'] Run only tests containing the string value"
print "\n\tExample: Run only the test 'start_top'"
print "\t\t python ./regression.py --run-only start_stop"
print "\n\tExample: Run only the tests with the string 'systemd' present in them"
print "\t\t python ./regression.py --run-only-pattern systemd"
def main(argv):
o = TestOptions()
o.build_options(argv)
tests = Tests(o.options['verbose'], o.options['pacemaker-remote'])
tests.build_generic_tests()
tests.build_multi_rsc_tests()
tests.build_negative_tests()
tests.build_custom_tests()
tests.setup_test_environment()
print "Starting ..."
if o.options['list-tests']:
tests.print_list()
elif o.options['show-usage']:
o.show_usage()
elif o.options['run-only-pattern'] != "":
tests.run_tests_matching(o.options['run-only-pattern'])
tests.print_results()
elif o.options['run-only'] != "":
tests.run_single(o.options['run-only'])
tests.print_results()
else:
tests.run_tests()
tests.print_results()
tests.cleanup_test_environment()
tests.exit()
if __name__=="__main__":
main(sys.argv)