Page MenuHomeClusterLabs Projects
Files F3151762

No OneTemporary
Actions

View Options

diff --git a/agents/gce/fence_gce.py b/agents/gce/fence_gce.py
index 3abb5207..3eca0139 100644
--- a/agents/gce/fence_gce.py
+++ b/agents/gce/fence_gce.py
@@ -1,114 +1,254 @@
#!@PYTHON@ -tt
import atexit
+import logging
+import os
import sys
+import time
+if sys.version_info >= (3, 0):
+ # Python 3 imports.
+ import urllib.parse as urlparse
+ import urllib.request as urlrequest
+else:
+ # Python 2 imports.
+ import urllib as urlparse
+ import urllib2 as urlrequest
sys.path.append("@FENCEAGENTSLIBDIR@")
import googleapiclient.discovery
from fencing import fail_usage, run_delay, all_opt, atexit_handler, check_input, process_input, show_docs, fence_action
+
+METADATA_SERVER = 'http://metadata.google.internal/computeMetadata/v1/'
+METADATA_HEADERS = {'Metadata-Flavor': 'Google'}
+
+
def translate_status(instance_status):
"Returns on | off | unknown."
if instance_status == "RUNNING":
return "on"
elif instance_status == "TERMINATED":
return "off"
return "unknown"
def get_nodes_list(conn, options):
result = {}
try:
instanceList = conn.instances().list(project=options["--project"], zone=options["--zone"]).execute()
for instance in instanceList["items"]:
result[instance["id"]] = (instance["name"], translate_status(instance["status"]))
except Exception as err:
fail_usage("Failed: get_nodes_list: {}".format(str(err)))
return result
+
def get_power_status(conn, options):
try:
instance = conn.instances().get(
project=options["--project"],
zone=options["--zone"],
instance=options["--plug"]).execute()
return translate_status(instance["status"])
except Exception as err:
fail_usage("Failed: get_power_status: {}".format(str(err)))
+def wait_for_operation(conn, project, zone, operation):
+ while True:
+ result = conn.zoneOperations().get(
+ project=project,
+ zone=zone,
+ operation=operation['name']).execute()
+ if result['status'] == 'DONE':
+ if 'error' in result:
+ raise Exception(result['error'])
+ return
+ time.sleep(1)
+
+
def set_power_status(conn, options):
try:
if options["--action"] == "off":
- conn.instances().stop(
+ logging.info("Issuing poweroff of %s in zone %s" % (options["--plug"], options["--zone"]))
+ operation = conn.instances().stop(
project=options["--project"],
zone=options["--zone"],
instance=options["--plug"]).execute()
+ wait_for_operation(conn, options["--project"], options["--zone"], operation)
+ logging.info("Poweroff of %s in zone %s complete" % (options["--plug"], options["--zone"]))
elif options["--action"] == "on":
- conn.instances().start(
+ logging.info("Issuing poweron of %s in zone %s" % (options["--plug"], options["--zone"]))
+ operation = conn.instances().start(
project=options["--project"],
zone=options["--zone"],
instance=options["--plug"]).execute()
+ wait_for_operation(conn, options["--project"], options["--zone"], operation)
+ logging.info("Poweron of %s in zone %s complete" % (options["--plug"], options["--zone"]))
except Exception as err:
fail_usage("Failed: set_power_status: {}".format(str(err)))
+def power_cycle(conn, options):
+ try:
+ logging.info('Issuing reset of %s in zone %s' % (options["--plug"], options["--zone"]))
+ operation = conn.instances().reset(
+ project=options["--project"],
+ zone=options["--zone"],
+ instance=options["--plug"]).execute()
+ wait_for_operation(conn, options["--project"], options["--zone"], operation)
+ logging.info('Reset of %s in zone %s complete' % (options["--plug"], options["--zone"]))
+ return True
+ except Exception as err:
+ logging.error("Failed: power_cycle: {}".format(str(err)))
+ return False
+
+
+def get_instance(conn, project, zone, instance):
+ request = conn.instances().get(
+ project=project, zone=zone, instance=instance)
+ return request.execute()
+
+
+def get_zone(conn, project, instance):
+ request = conn.instances().aggregatedList(project=project)
+ while request is not None:
+ response = request.execute()
+ zones = response.get('items', {})
+ for zone in zones.values():
+ for inst in zone.get('instances', []):
+ if inst['name'] == instance:
+ return inst['zone'].split("/")[-1]
+ request = conn.instances().aggregatedList_next(
+ previous_request=request, previous_response=response)
+ raise Exception("Unable to find instance %s" % (instance))
+
+
+def get_metadata(metadata_key, params=None, timeout=None):
+ """Performs a GET request with the metadata headers.
+
+ Args:
+ metadata_key: string, the metadata to perform a GET request on.
+ params: dictionary, the query parameters in the GET request.
+ timeout: int, timeout in seconds for metadata requests.
+
+ Returns:
+ HTTP response from the GET request.
+
+ Raises:
+ urlerror.HTTPError: raises when the GET request fails.
+ """
+ timeout = timeout or 60
+ metadata_url = os.path.join(METADATA_SERVER, metadata_key)
+ params = urlparse.urlencode(params or {})
+ url = '%s?%s' % (metadata_url, params)
+ request = urlrequest.Request(url, headers=METADATA_HEADERS)
+ request_opener = urlrequest.build_opener(urlrequest.ProxyHandler({}))
+ return request_opener.open(request, timeout=timeout * 1.1).read()
+
+
def define_new_opts():
all_opt["zone"] = {
"getopt" : ":",
"longopt" : "zone",
"help" : "--zone=[name] Zone, e.g. us-central1-b",
"shortdesc" : "Zone.",
- "required" : "1",
+ "required" : "0",
"order" : 2
}
all_opt["project"] = {
"getopt" : ":",
"longopt" : "project",
"help" : "--project=[name] Project ID",
"shortdesc" : "Project ID.",
- "required" : "1",
+ "required" : "0",
"order" : 3
}
+ all_opt["stackdriver-logging"] = {
+ "getopt" : "",
+ "longopt" : "stackdriver-logging",
+ "help" : "--stackdriver-logging Enable Logging to Stackdriver",
+ "shortdesc" : "Stackdriver-logging support.",
+ "longdesc" : "If enabled IP failover logs will be posted to stackdriver logging.",
+ "required" : "0",
+ "order" : 4
+ }
+
def main():
conn = None
- device_opt = ["port", "no_password", "zone", "project"]
+ device_opt = ["port", "no_password", "zone", "project", "stackdriver-logging", "method"]
atexit.register(atexit_handler)
define_new_opts()
all_opt["power_timeout"]["default"] = "60"
+ all_opt["method"]["default"] = "cycle"
+ all_opt["method"]["help"] = "-m, --method=[method] Method to fence (onoff|cycle) (Default: cycle)"
options = check_input(device_opt, process_input(device_opt))
docs = {}
docs["shortdesc"] = "Fence agent for GCE (Google Cloud Engine)"
docs["longdesc"] = "fence_gce is an I/O Fencing agent for GCE (Google Cloud " \
"Engine). It uses the googleapiclient library to connect to GCE.\n" \
"googleapiclient can be configured with Google SDK CLI or by " \
"executing 'gcloud auth application-default login'.\n" \
"For instructions see: https://cloud.google.com/compute/docs/tutorials/python-guide"
docs["vendorurl"] = "http://cloud.google.com"
show_docs(options, docs)
run_delay(options)
+ # Prepare logging
+ if options.get('--verbose') is None:
+ logging.getLogger('googleapiclient').setLevel(logging.ERROR)
+ logging.getLogger('oauth2client').setLevel(logging.ERROR)
+ if options.get('--stackdriver-logging') is not None and options.get('--plug'):
+ try:
+ import google.cloud.logging.handlers
+ client = google.cloud.logging.Client()
+ handler = google.cloud.logging.handlers.CloudLoggingHandler(client, name=options['--plug'])
+ handler.setLevel(logging.INFO)
+ formatter = logging.Formatter('gcp:stonith "%(message)s"')
+ handler.setFormatter(formatter)
+ root_logger = logging.getLogger()
+ if options.get('--verbose') is None:
+ root_logger.setLevel(logging.INFO)
+ root_logger.addHandler(handler)
+ except ImportError:
+ logging.error('Couldn\'t import google.cloud.logging, '
+ 'disabling Stackdriver-logging support')
+
+ # Prepare cli
try:
credentials = None
if tuple(googleapiclient.__version__) < tuple("1.6.0"):
import oauth2client.client
credentials = oauth2client.client.GoogleCredentials.get_application_default()
conn = googleapiclient.discovery.build('compute', 'v1', credentials=credentials)
except Exception as err:
fail_usage("Failed: Create GCE compute v1 connection: {}".format(str(err)))
+ # Get project and zone
+ if not options.get("--project"):
+ try:
+ options["--project"] = get_metadata('project/project-id')
+ except Exception as err:
+ fail_usage("Failed retrieving GCE project. Please provide --project option: {}".format(str(err)))
+
+ if not options.get("--zone"):
+ try:
+ options["--zone"] = get_zone(conn, options['--project'], options['--plug'])
+ except Exception as err:
+ fail_usage("Failed retrieving GCE zone. Please provide --zone option: {}".format(str(err)))
+
# Operate the fencing device
- result = fence_action(conn, options, set_power_status, get_power_status, get_nodes_list)
+ result = fence_action(conn, options, set_power_status, get_power_status, get_nodes_list, power_cycle)
sys.exit(result)
if __name__ == "__main__":
main()
diff --git a/tests/data/metadata/fence_gce.xml b/tests/data/metadata/fence_gce.xml
index 2a147f21..79b82ebb 100644
--- a/tests/data/metadata/fence_gce.xml
+++ b/tests/data/metadata/fence_gce.xml
@@ -1,111 +1,129 @@
<?xml version="1.0" ?>
<resource-agent name="fence_gce" shortdesc="Fence agent for GCE (Google Cloud Engine)" >
<longdesc>fence_gce is an I/O Fencing agent for GCE (Google Cloud Engine). It uses the googleapiclient library to connect to GCE.
googleapiclient can be configured with Google SDK CLI or by executing 'gcloud auth application-default login'.
For instructions see: https://cloud.google.com/compute/docs/tutorials/python-guide</longdesc>
<vendor-url>http://cloud.google.com</vendor-url>
<parameters>
<parameter name="action" unique="0" required="1">
<getopt mixed="-o, --action=[action]" />
<content type="string" default="reboot" />
<shortdesc lang="en">Fencing action</shortdesc>
</parameter>
+ <parameter name="method" unique="0" required="0">
+ <getopt mixed="-m, --method=[method]" />
+ <content type="select" default="cycle" >
+ <option value="onoff" />
+ <option value="cycle" />
+ </content>
+ <shortdesc lang="en">Method to fence</shortdesc>
+ </parameter>
<parameter name="plug" unique="0" required="1" obsoletes="port">
<getopt mixed="-n, --plug=[id]" />
<content type="string" />
<shortdesc lang="en">Physical plug number on device, UUID or identification of machine</shortdesc>
</parameter>
<parameter name="port" unique="0" required="1" deprecated="1">
<getopt mixed="-n, --plug=[id]" />
<content type="string" />
<shortdesc lang="en">Physical plug number on device, UUID or identification of machine</shortdesc>
</parameter>
- <parameter name="zone" unique="0" required="1">
+ <parameter name="zone" unique="0" required="0">
<getopt mixed="--zone=[name]" />
<content type="string" />
<shortdesc lang="en">Zone.</shortdesc>
</parameter>
- <parameter name="project" unique="0" required="1">
+ <parameter name="project" unique="0" required="0">
<getopt mixed="--project=[name]" />
<content type="string" />
<shortdesc lang="en">Project ID.</shortdesc>
</parameter>
+ <parameter name="stackdriver-logging" unique="0" required="0" deprecated="1">
+ <getopt mixed="--stackdriver-logging" />
+ <content type="boolean" />
+ <shortdesc lang="en">Stackdriver-logging support.</shortdesc>
+ </parameter>
+ <parameter name="stackdriver_logging" unique="0" required="0" obsoletes="stackdriver-logging">
+ <getopt mixed="--stackdriver-logging" />
+ <content type="boolean" />
+ <shortdesc lang="en">Stackdriver-logging support.</shortdesc>
+ </parameter>
<parameter name="quiet" unique="0" required="0">
<getopt mixed="-q, --quiet" />
<content type="boolean" />
<shortdesc lang="en">Disable logging to stderr. Does not affect --verbose or --debug-file or logging to syslog.</shortdesc>
</parameter>
<parameter name="verbose" unique="0" required="0">
<getopt mixed="-v, --verbose" />
<content type="boolean" />
<shortdesc lang="en">Verbose mode</shortdesc>
</parameter>
<parameter name="debug" unique="0" required="0" deprecated="1">
<getopt mixed="-D, --debug-file=[debugfile]" />
<content type="string" />
<shortdesc lang="en">Write debug information to given file</shortdesc>
</parameter>
<parameter name="debug_file" unique="0" required="0" obsoletes="debug">
<getopt mixed="-D, --debug-file=[debugfile]" />
<content type="string" />
<shortdesc lang="en">Write debug information to given file</shortdesc>
</parameter>
<parameter name="version" unique="0" required="0">
<getopt mixed="-V, --version" />
<content type="boolean" />
<shortdesc lang="en">Display version information and exit</shortdesc>
</parameter>
<parameter name="help" unique="0" required="0">
<getopt mixed="-h, --help" />
<content type="boolean" />
<shortdesc lang="en">Display help and exit</shortdesc>
</parameter>
<parameter name="separator" unique="0" required="0">
<getopt mixed="-C, --separator=[char]" />
<content type="string" default="," />
<shortdesc lang="en">Separator for CSV created by 'list' operation</shortdesc>
</parameter>
<parameter name="delay" unique="0" required="0">
<getopt mixed="--delay=[seconds]" />
<content type="second" default="0" />
<shortdesc lang="en">Wait X seconds before fencing is started</shortdesc>
</parameter>
<parameter name="login_timeout" unique="0" required="0">
<getopt mixed="--login-timeout=[seconds]" />
<content type="second" default="5" />
<shortdesc lang="en">Wait X seconds for cmd prompt after login</shortdesc>
</parameter>
<parameter name="power_timeout" unique="0" required="0">
<getopt mixed="--power-timeout=[seconds]" />
<content type="second" default="60" />
<shortdesc lang="en">Test X seconds for status change after ON/OFF</shortdesc>
</parameter>
<parameter name="power_wait" unique="0" required="0">
<getopt mixed="--power-wait=[seconds]" />
<content type="second" default="0" />
<shortdesc lang="en">Wait X seconds after issuing ON/OFF</shortdesc>
</parameter>
<parameter name="shell_timeout" unique="0" required="0">
<getopt mixed="--shell-timeout=[seconds]" />
<content type="second" default="3" />
<shortdesc lang="en">Wait X seconds for cmd prompt after issuing command</shortdesc>
</parameter>
<parameter name="retry_on" unique="0" required="0">
<getopt mixed="--retry-on=[attempts]" />
<content type="integer" default="1" />
<shortdesc lang="en">Count of attempts to retry power on</shortdesc>
</parameter>
</parameters>
<actions>
<action name="on" automatic="0"/>
<action name="off" />
<action name="reboot" />
<action name="status" />
<action name="list" />
<action name="list-status" />
<action name="monitor" />
<action name="metadata" />
<action name="manpage" />
<action name="validate-all" />
</actions>
</resource-agent>

File Metadata

Mime Type
text/x-diff
Expires
Mon, Feb 24, 10:27 AM (20 h, 16 m ago)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
1464136
Default Alt Text
(14 KB)

Event Timeline