rpms/resource-agents
7
0
Fork 0
Code Issues Pull Requests Releases Activity

import resource-agents-4.10.0-9.el9_0.5

Browse Source
This commit is contained in:
CentOS Sources 2022-11-02 10:00:17 -04:00 committed by Stepan Oksanichenko
parent b661c418e3
commit 456162fc4d
3 changed files with 1065 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

903
SOURCES/bz2130990-azure-events-az-new-ra.patch Normal file
View File

@ -0,0 +1,903 @@
From 5dcd5153f0318e4766f7f4d3e61dfdb4b352c39c Mon Sep 17 00:00:00 2001
From: MSSedusch <sedusch@microsoft.com>
Date: Mon, 30 May 2022 15:08:10 +0200
Subject: [PATCH 1/2] add new Azure Events AZ resource agent
---
.gitignore | 1 +
configure.ac | 8 +
doc/man/Makefile.am | 4 +
heartbeat/Makefile.am | 4 +
heartbeat/azure-events-az.in | 782 +++++++++++++++++++++++++++++++++++
5 files changed, 799 insertions(+)
create mode 100644 heartbeat/azure-events-az.in
diff --git a/.gitignore b/.gitignore
index 0c259b5cf..e2b7c039c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -54,6 +54,7 @@ heartbeat/Squid
heartbeat/SysInfo
heartbeat/aws-vpc-route53
heartbeat/azure-events
+heartbeat/azure-events-az
heartbeat/clvm
heartbeat/conntrackd
heartbeat/dnsupdate
diff --git a/configure.ac b/configure.ac
index eeecfad0e..5716a2be2 100644
--- a/configure.ac
+++ b/configure.ac
@@ -523,6 +523,13 @@ if test -z "$PYTHON" || test $BUILD_OCF_PY -eq 0; then
fi
AM_CONDITIONAL(BUILD_AZURE_EVENTS, test $BUILD_AZURE_EVENTS -eq 1)
+BUILD_AZURE_EVENTS_AZ=1
+if test -z "$PYTHON" || test $BUILD_OCF_PY -eq 0; then
+ BUILD_AZURE_EVENTS_AZ=0
+ AC_MSG_WARN("Not building azure-events-az")
+fi
+AM_CONDITIONAL(BUILD_AZURE_EVENTS_AZ, test $BUILD_AZURE_EVENTS_AZ -eq 1)
+
BUILD_GCP_PD_MOVE=1
if test -z "$PYTHON" || test "x${HAVE_PYMOD_GOOGLEAPICLIENT}" != xyes || test $BUILD_OCF_PY -eq 0; then
BUILD_GCP_PD_MOVE=0
@@ -976,6 +983,7 @@ rgmanager/Makefile \
dnl Files we output that need to be executable
AC_CONFIG_FILES([heartbeat/azure-events], [chmod +x heartbeat/azure-events])
+AC_CONFIG_FILES([heartbeat/azure-events-az], [chmod +x heartbeat/azure-events-az])
AC_CONFIG_FILES([heartbeat/AoEtarget], [chmod +x heartbeat/AoEtarget])
AC_CONFIG_FILES([heartbeat/ManageRAID], [chmod +x heartbeat/ManageRAID])
AC_CONFIG_FILES([heartbeat/ManageVE], [chmod +x heartbeat/ManageVE])
diff --git a/doc/man/Makefile.am b/doc/man/Makefile.am
index cd8fd16bf..658c700ac 100644
--- a/doc/man/Makefile.am
+++ b/doc/man/Makefile.am
@@ -219,6 +219,10 @@ if BUILD_AZURE_EVENTS
man_MANS += ocf_heartbeat_azure-events.7
endif
+if BUILD_AZURE_EVENTS_AZ
+man_MANS += ocf_heartbeat_azure-events-az.7
+endif
+
if BUILD_GCP_PD_MOVE
man_MANS += ocf_heartbeat_gcp-pd-move.7
endif
diff --git a/heartbeat/Makefile.am b/heartbeat/Makefile.am
index 20d41e36a..1133dc13e 100644
--- a/heartbeat/Makefile.am
+++ b/heartbeat/Makefile.am
@@ -188,6 +188,10 @@ if BUILD_AZURE_EVENTS
ocf_SCRIPTS += azure-events
endif
+if BUILD_AZURE_EVENTS_AZ
+ocf_SCRIPTS += azure-events-az
+endif
+
if BUILD_GCP_PD_MOVE
ocf_SCRIPTS += gcp-pd-move
endif
diff --git a/heartbeat/azure-events-az.in b/heartbeat/azure-events-az.in
new file mode 100644
index 000000000..616fc8d9e
--- /dev/null
+++ b/heartbeat/azure-events-az.in
@@ -0,0 +1,782 @@
+#!@PYTHON@ -tt
+#
+# Resource agent for monitoring Azure Scheduled Events
+#
+# License: GNU General Public License (GPL)
+# (c) 2018 Tobias Niekamp, Microsoft Corp.
+# and Linux-HA contributors
+
+import os
+import sys
+import time
+import subprocess
+import json
+try:
+ import urllib2
+ from urllib2 import URLError
+except ImportError:
+ import urllib.request as urllib2
+ from urllib.error import URLError
+import socket
+from collections import defaultdict
+
+OCF_FUNCTIONS_DIR = os.environ.get("OCF_FUNCTIONS_DIR", "%s/lib/heartbeat" % os.environ.get("OCF_ROOT"))
+sys.path.append(OCF_FUNCTIONS_DIR)
+import ocf
+
+##############################################################################
+
+
+VERSION = "0.10"
+USER_AGENT = "Pacemaker-ResourceAgent/%s %s" % (VERSION, ocf.distro())
+
+attr_globalPullState = "azure-events-az_globalPullState"
+attr_lastDocVersion = "azure-events-az_lastDocVersion"
+attr_curNodeState = "azure-events-az_curNodeState"
+attr_pendingEventIDs = "azure-events-az_pendingEventIDs"
+attr_healthstate = "#health-azure"
+
+default_loglevel = ocf.logging.INFO
+default_relevantEventTypes = set(["Reboot", "Redeploy"])
+
+global_pullMaxAttempts = 3
+global_pullDelaySecs = 1
+
+##############################################################################
+
+class attrDict(defaultdict):
+ """
+ A wrapper for accessing dict keys like an attribute
+ """
+ def __init__(self, data):
+ super(attrDict, self).__init__(attrDict)
+ for d in data.keys():
+ self.__setattr__(d, data[d])
+
+ def __getattr__(self, key):
+ try:
+ return self[key]
+ except KeyError:
+ raise AttributeError(key)
+
+ def __setattr__(self, key, value):
+ self[key] = value
+
+##############################################################################
+
+class azHelper:
+ """
+ Helper class for Azure's metadata API (including Scheduled Events)
+ """
+ metadata_host = "http://169.254.169.254/metadata"
+ instance_api = "instance"
+ events_api = "scheduledevents"
+ api_version = "2019-08-01"
+
+ @staticmethod
+ def _sendMetadataRequest(endpoint, postData=None):
+ """
+ Send a request to Azure's Azure Metadata Service API
+ """
+ url = "%s/%s?api-version=%s" % (azHelper.metadata_host, endpoint, azHelper.api_version)
+ data = ""
+ ocf.logger.debug("_sendMetadataRequest: begin; endpoint = %s, postData = %s" % (endpoint, postData))
+ ocf.logger.debug("_sendMetadataRequest: url = %s" % url)
+
+ if postData and type(postData) != bytes:
+ postData = postData.encode()
+
+ req = urllib2.Request(url, postData)
+ req.add_header("Metadata", "true")
+ req.add_header("User-Agent", USER_AGENT)
+ try:
+ resp = urllib2.urlopen(req)
+ except URLError as e:
+ if hasattr(e, 'reason'):
+ ocf.logger.warning("Failed to reach the server: %s" % e.reason)
+ clusterHelper.setAttr(attr_globalPullState, "IDLE")
+ elif hasattr(e, 'code'):
+ ocf.logger.warning("The server couldn\'t fulfill the request. Error code: %s" % e.code)
+ clusterHelper.setAttr(attr_globalPullState, "IDLE")
+ else:
+ data = resp.read()
+ ocf.logger.debug("_sendMetadataRequest: response = %s" % data)
+
+ if data:
+ data = json.loads(data)
+
+ ocf.logger.debug("_sendMetadataRequest: finished")
+ return data
+
+ @staticmethod
+ def getInstanceInfo():
+ """
+ Fetch details about the current VM from Azure's Azure Metadata Service API
+ """
+ ocf.logger.debug("getInstanceInfo: begin")
+
+ jsondata = azHelper._sendMetadataRequest(azHelper.instance_api)
+ ocf.logger.debug("getInstanceInfo: json = %s" % jsondata)
+
+ if jsondata:
+ ocf.logger.debug("getInstanceInfo: finished, returning {}".format(jsondata["compute"]))
+ return attrDict(jsondata["compute"])
+ else:
+ ocf.ocf_exit_reason("getInstanceInfo: Unable to get instance info")
+ sys.exit(ocf.OCF_ERR_GENERIC)
+
+ @staticmethod
+ def pullScheduledEvents():
+ """
+ Retrieve all currently scheduled events via Azure Metadata Service API
+ """
+ ocf.logger.debug("pullScheduledEvents: begin")
+
+ jsondata = azHelper._sendMetadataRequest(azHelper.events_api)
+ ocf.logger.debug("pullScheduledEvents: json = %s" % jsondata)
+
+ ocf.logger.debug("pullScheduledEvents: finished")
+ return attrDict(jsondata)
+
+ @staticmethod
+ def forceEvents(eventIDs):
+ """
+ Force a set of events to start immediately
+ """
+ ocf.logger.debug("forceEvents: begin")
+
+ events = []
+ for e in eventIDs:
+ events.append({
+ "EventId": e,
+ })
+ postData = {
+ "StartRequests" : events
+ }
+ ocf.logger.info("forceEvents: postData = %s" % postData)
+ resp = azHelper._sendMetadataRequest(azHelper.events_api, postData=json.dumps(postData))
+
+ ocf.logger.debug("forceEvents: finished")
+ return
+
+##############################################################################
+
+class clusterHelper:
+ """
+ Helper functions for Pacemaker control via crm
+ """
+ @staticmethod
+ def _getLocation(node):
+ """
+ Helper function to retrieve local/global attributes
+ """
+ if node:
+ return ["--node", node]
+ else:
+ return ["--type", "crm_config"]
+
+ @staticmethod
+ def _exec(command, *args):
+ """
+ Helper function to execute a UNIX command
+ """
+ args = list(args)
+ ocf.logger.debug("_exec: begin; command = %s, args = %s" % (command, str(args)))
+
+ def flatten(*n):
+ return (str(e) for a in n
+ for e in (flatten(*a) if isinstance(a, (tuple, list)) else (str(a),)))
+ command = list(flatten([command] + args))
+ ocf.logger.debug("_exec: cmd = %s" % " ".join(command))
+ try:
+ ret = subprocess.check_output(command)
+ if type(ret) != str:
+ ret = ret.decode()
+ ocf.logger.debug("_exec: return = %s" % ret)
+ return ret.rstrip()
+ except Exception as err:
+ ocf.logger.exception(err)
+ return None
+
+ @staticmethod
+ def setAttr(key, value, node=None):
+ """
+ Set the value of a specific global/local attribute in the Pacemaker cluster
+ """
+ ocf.logger.debug("setAttr: begin; key = %s, value = %s, node = %s" % (key, value, node))
+
+ if value:
+ ret = clusterHelper._exec("crm_attribute",
+ "--name", key,
+ "--update", value,
+ clusterHelper._getLocation(node))
+ else:
+ ret = clusterHelper._exec("crm_attribute",
+ "--name", key,
+ "--delete",
+ clusterHelper._getLocation(node))
+
+ ocf.logger.debug("setAttr: finished")
+ return len(ret) == 0
+
+ @staticmethod
+ def getAttr(key, node=None):
+ """
+ Retrieve a global/local attribute from the Pacemaker cluster
+ """
+ ocf.logger.debug("getAttr: begin; key = %s, node = %s" % (key, node))
+
+ val = clusterHelper._exec("crm_attribute",
+ "--name", key,
+ "--query", "--quiet",
+ "--default", "",
+ clusterHelper._getLocation(node))
+ ocf.logger.debug("getAttr: finished")
+ if not val:
+ return None
+ return val if not val.isdigit() else int(val)
+
+ @staticmethod
+ def getAllNodes():
+ """
+ Get a list of hostnames for all nodes in the Pacemaker cluster
+ """
+ ocf.logger.debug("getAllNodes: begin")
+
+ nodes = []
+ nodeList = clusterHelper._exec("crm_node", "--list")
+ for n in nodeList.split("\n"):
+ nodes.append(n.split()[1])
+ ocf.logger.debug("getAllNodes: finished; return %s" % str(nodes))
+
+ return nodes
+
+ @staticmethod
+ def getHostNameFromAzName(azName):
+ """
+ Helper function to get the actual host name from an Azure node name
+ """
+ return clusterHelper.getAttr("hostName_%s" % azName)
+
+ @staticmethod
+ def removeHoldFromNodes():
+ """
+ Remove the ON_HOLD state from all nodes in the Pacemaker cluster
+ """
+ ocf.logger.debug("removeHoldFromNodes: begin")
+
+ for n in clusterHelper.getAllNodes():
+ if clusterHelper.getAttr(attr_curNodeState, node=n) == "ON_HOLD":
+ clusterHelper.setAttr(attr_curNodeState, "AVAILABLE", node=n)
+ ocf.logger.info("removeHoldFromNodes: removed ON_HOLD from node %s" % n)
+
+ ocf.logger.debug("removeHoldFromNodes: finished")
+ return False
+
+ @staticmethod
+ def otherNodesAvailable(exceptNode):
+ """
+ Check if there are any nodes (except a given node) in the Pacemaker cluster that have state AVAILABLE
+ """
+ ocf.logger.debug("otherNodesAvailable: begin; exceptNode = %s" % exceptNode)
+
+ for n in clusterHelper.getAllNodes():
+ state = clusterHelper.getAttr(attr_curNodeState, node=n)
+ state = stringToNodeState(state) if state else AVAILABLE
+ if state == AVAILABLE and n != exceptNode.hostName:
+ ocf.logger.info("otherNodesAvailable: at least %s is available" % n)
+ ocf.logger.debug("otherNodesAvailable: finished")
+ return True
+ ocf.logger.info("otherNodesAvailable: no other nodes are available")
+ ocf.logger.debug("otherNodesAvailable: finished")
+
+ return False
+
+ @staticmethod
+ def transitionSummary():
+ """
+ Get the current Pacemaker transition summary (used to check if all resources are stopped when putting a node standby)
+ """
+ # <tniek> Is a global crm_simulate "too much"? Or would it be sufficient it there are no planned transitions for a particular node?
+ # # crm_simulate -Ls
+ # Transition Summary:
+ # * Promote rsc_SAPHana_HN1_HDB03:0 (Slave -> Master hsr3-db1)
+ # * Stop rsc_SAPHana_HN1_HDB03:1 (hsr3-db0)
+ # * Move rsc_ip_HN1_HDB03 (Started hsr3-db0 -> hsr3-db1)
+ # * Start rsc_nc_HN1_HDB03 (hsr3-db1)
+ # # Excepted result when there are no pending actions:
+ # Transition Summary:
+ ocf.logger.debug("transitionSummary: begin")
+
+ summary = clusterHelper._exec("crm_simulate", "-Ls")
+ if not summary:
+ ocf.logger.warning("transitionSummary: could not load transition summary")
+ return False
+ if summary.find("Transition Summary:") < 0:
+ ocf.logger.warning("transitionSummary: received unexpected transition summary: %s" % summary)
+ return False
+ summary = summary.split("Transition Summary:")[1]
+ ret = summary.split("\n").pop(0)
+
+ ocf.logger.debug("transitionSummary: finished; return = %s" % str(ret))
+ return ret
+
+ @staticmethod
+ def listOperationsOnNode(node):
+ """
+ Get a list of all current operations for a given node (used to check if any resources are pending)
+ """
+ # hsr3-db1:/home/tniek # crm_resource --list-operations -N hsr3-db0
+ # rsc_azure-events-az (ocf::heartbeat:azure-events-az): Started: rsc_azure-events-az_start_0 (node=hsr3-db0, call=91, rc=0, last-rc-change=Fri Jun 8 22:37:46 2018, exec=115ms): complete
+ # rsc_azure-events-az (ocf::heartbeat:azure-events-az): Started: rsc_azure-events-az_monitor_10000 (node=hsr3-db0, call=93, rc=0, last-rc-change=Fri Jun 8 22:37:47 2018, exec=197ms): complete
+ # rsc_SAPHana_HN1_HDB03 (ocf::suse:SAPHana): Master: rsc_SAPHana_HN1_HDB03_start_0 (node=hsr3-db0, call=-1, rc=193, last-rc-change=Fri Jun 8 22:37:46 2018, exec=0ms): pending
+ # rsc_SAPHanaTopology_HN1_HDB03 (ocf::suse:SAPHanaTopology): Started: rsc_SAPHanaTopology_HN1_HDB03_start_0 (node=hsr3-db0, call=90, rc=0, last-rc-change=Fri Jun 8 22:37:46 2018, exec=3214ms): complete
+ ocf.logger.debug("listOperationsOnNode: begin; node = %s" % node)
+
+ resources = clusterHelper._exec("crm_resource", "--list-operations", "-N", node)
+ if len(resources) == 0:
+ ret = []
+ else:
+ ret = resources.split("\n")
+
+ ocf.logger.debug("listOperationsOnNode: finished; return = %s" % str(ret))
+ return ret
+
+ @staticmethod
+ def noPendingResourcesOnNode(node):
+ """
+ Check that there are no pending resources on a given node
+ """
+ ocf.logger.debug("noPendingResourcesOnNode: begin; node = %s" % node)
+
+ for r in clusterHelper.listOperationsOnNode(node):
+ ocf.logger.debug("noPendingResourcesOnNode: * %s" % r)
+ resource = r.split()[-1]
+ if resource == "pending":
+ ocf.logger.info("noPendingResourcesOnNode: found resource %s that is still pending" % resource)
+ ocf.logger.debug("noPendingResourcesOnNode: finished; return = False")
+ return False
+ ocf.logger.info("noPendingResourcesOnNode: no pending resources on node %s" % node)
+ ocf.logger.debug("noPendingResourcesOnNode: finished; return = True")
+
+ return True
+
+ @staticmethod
+ def allResourcesStoppedOnNode(node):
+ """
+ Check that all resources on a given node are stopped
+ """
+ ocf.logger.debug("allResourcesStoppedOnNode: begin; node = %s" % node)
+
+ if clusterHelper.noPendingResourcesOnNode(node):
+ if len(clusterHelper.transitionSummary()) == 0:
+ ocf.logger.info("allResourcesStoppedOnNode: no pending resources on node %s and empty transition summary" % node)
+ ocf.logger.debug("allResourcesStoppedOnNode: finished; return = True")
+ return True
+ ocf.logger.info("allResourcesStoppedOnNode: transition summary is not empty")
+ ocf.logger.debug("allResourcesStoppedOnNode: finished; return = False")
+ return False
+
+ ocf.logger.info("allResourcesStoppedOnNode: still pending resources on node %s" % node)
+ ocf.logger.debug("allResourcesStoppedOnNode: finished; return = False")
+ return False
+
+##############################################################################
+
+AVAILABLE = 0 # Node is online and ready to handle events
+STOPPING = 1 # Standby has been triggered, but some resources are still running
+IN_EVENT = 2 # All resources are stopped, and event has been initiated via Azure Metadata Service
+ON_HOLD = 3 # Node has a pending event that cannot be started there are no other nodes available
+
+def stringToNodeState(name):
+ if type(name) == int: return name
+ if name == "STOPPING": return STOPPING
+ if name == "IN_EVENT": return IN_EVENT
+ if name == "ON_HOLD": return ON_HOLD
+ return AVAILABLE
+
+def nodeStateToString(state):
+ if state == STOPPING: return "STOPPING"
+ if state == IN_EVENT: return "IN_EVENT"
+ if state == ON_HOLD: return "ON_HOLD"
+ return "AVAILABLE"
+
+##############################################################################
+
+class Node:
+ """
+ Core class implementing logic for a cluster node
+ """
+ def __init__(self, ra):
+ self.raOwner = ra
+ self.azInfo = azHelper.getInstanceInfo()
+ self.azName = self.azInfo.name
+ self.hostName = socket.gethostname()
+ self.setAttr("azName", self.azName)
+ clusterHelper.setAttr("hostName_%s" % self.azName, self.hostName)
+
+ def getAttr(self, key):
+ """
+ Get a local attribute
+ """
+ return clusterHelper.getAttr(key, node=self.hostName)
+
+ def setAttr(self, key, value):
+ """
+ Set a local attribute
+ """
+ return clusterHelper.setAttr(key, value, node=self.hostName)
+
+ def selfOrOtherNode(self, node):
+ """
+ Helper function to distinguish self/other node
+ """
+ return node if node else self.hostName
+
+ def setState(self, state, node=None):
+ """
+ Set the state for a given node (or self)
+ """
+ node = self.selfOrOtherNode(node)
+ ocf.logger.debug("setState: begin; node = %s, state = %s" % (node, nodeStateToString(state)))
+
+ clusterHelper.setAttr(attr_curNodeState, nodeStateToString(state), node=node)
+
+ ocf.logger.debug("setState: finished")
+
+ def getState(self, node=None):
+ """
+ Get the state for a given node (or self)
+ """
+ node = self.selfOrOtherNode(node)
+ ocf.logger.debug("getState: begin; node = %s" % node)
+
+ state = clusterHelper.getAttr(attr_curNodeState, node=node)
+ ocf.logger.debug("getState: state = %s" % state)
+ ocf.logger.debug("getState: finished")
+ if not state:
+ return AVAILABLE
+ return stringToNodeState(state)
+
+ def setEventIDs(self, eventIDs, node=None):
+ """
+ Set pending EventIDs for a given node (or self)
+ """
+ node = self.selfOrOtherNode(node)
+ ocf.logger.debug("setEventIDs: begin; node = %s, eventIDs = %s" % (node, str(eventIDs)))
+
+ if eventIDs:
+ eventIDStr = ",".join(eventIDs)
+ else:
+ eventIDStr = None
+ clusterHelper.setAttr(attr_pendingEventIDs, eventIDStr, node=node)
+
+ ocf.logger.debug("setEventIDs: finished")
+ return
+
+ def getEventIDs(self, node=None):
+ """
+ Get pending EventIDs for a given node (or self)
+ """
+ node = self.selfOrOtherNode(node)
+ ocf.logger.debug("getEventIDs: begin; node = %s" % node)
+
+ eventIDStr = clusterHelper.getAttr(attr_pendingEventIDs, node=node)
+ if eventIDStr:
+ eventIDs = eventIDStr.split(",")
+ else:
+ eventIDs = None
+
+ ocf.logger.debug("getEventIDs: finished; eventIDs = %s" % str(eventIDs))
+ return eventIDs
+
+ def updateNodeStateAndEvents(self, state, eventIDs, node=None):
+ """
+ Set the state and pending EventIDs for a given node (or self)
+ """
+ ocf.logger.debug("updateNodeStateAndEvents: begin; node = %s, state = %s, eventIDs = %s" % (node, nodeStateToString(state), str(eventIDs)))
+
+ self.setState(state, node=node)
+ self.setEventIDs(eventIDs, node=node)
+
+ ocf.logger.debug("updateNodeStateAndEvents: finished")
+ return state
+
+ def putNodeStandby(self, node=None):
+ """
+ Put self to standby
+ """
+ node = self.selfOrOtherNode(node)
+ ocf.logger.debug("putNodeStandby: begin; node = %s" % node)
+
+ clusterHelper._exec("crm_attribute",
+ "--node", node,
+ "--name", attr_healthstate,
+ "--update", "-1000000",
+ "--lifetime=forever")
+
+ ocf.logger.debug("putNodeStandby: finished")
+
+ def isNodeInStandby(self, node=None):
+ """
+ check if node is in standby
+ """
+ node = self.selfOrOtherNode(node)
+ ocf.logger.debug("isNodeInStandby: begin; node = %s" % node)
+ isInStandy = False
+
+ healthAttributeStr = clusterHelper.getAttr(attr_healthstate, node)
+ if healthAttributeStr is not None:
+ try:
+ healthAttribute = int(healthAttributeStr)
+ isInStandy = healthAttribute < 0
+ except ValueError:
+ # Handle the exception
+ ocf.logger.warn("Health attribute %s on node %s cannot be converted to an integer value" % (healthAttributeStr, node))
+
+ ocf.logger.debug("isNodeInStandby: finished - result %s" % isInStandy)
+ return isInStandy
+
+ def putNodeOnline(self, node=None):
+ """
+ Put self back online
+ """
+ node = self.selfOrOtherNode(node)
+ ocf.logger.debug("putNodeOnline: begin; node = %s" % node)
+
+ clusterHelper._exec("crm_attribute",
+ "--node", node,
+ "--name", "#health-azure",
+ "--update", "0",
+ "--lifetime=forever")
+
+ ocf.logger.debug("putNodeOnline: finished")
+
+ def separateEvents(self, events):
+ """
+ Split own/other nodes' events
+ """
+ ocf.logger.debug("separateEvents: begin; events = %s" % str(events))
+
+ localEvents = []
+ remoteEvents = []
+ for e in events:
+ e = attrDict(e)
+ if e.EventType not in self.raOwner.relevantEventTypes:
+ continue
+ if self.azName in e.Resources:
+ localEvents.append(e)
+ else:
+ remoteEvents.append(e)
+ ocf.logger.debug("separateEvents: finished; localEvents = %s, remoteEvents = %s" % (str(localEvents), str(remoteEvents)))
+ return (localEvents, remoteEvents)
+
+##############################################################################
+
+class raAzEvents:
+ """
+ Main class for resource agent
+ """
+ def __init__(self, relevantEventTypes):
+ self.node = Node(self)
+ self.relevantEventTypes = relevantEventTypes
+
+ def monitor(self):
+ ocf.logger.debug("monitor: begin")
+
+ events = azHelper.pullScheduledEvents()
+
+ # get current document version
+ curDocVersion = events.DocumentIncarnation
+ lastDocVersion = self.node.getAttr(attr_lastDocVersion)
+ ocf.logger.debug("monitor: lastDocVersion = %s; curDocVersion = %s" % (lastDocVersion, curDocVersion))
+
+ # split events local/remote
+ (localEvents, remoteEvents) = self.node.separateEvents(events.Events)
+
+ # ensure local events are only executing once
+ if curDocVersion == lastDocVersion:
+ ocf.logger.info("monitor: already handled curDocVersion, skip")
+ return ocf.OCF_SUCCESS
+
+ localAzEventIDs = set()
+ for e in localEvents:
+ localAzEventIDs.add(e.EventId)
+
+ curState = self.node.getState()
+ clusterEventIDs = self.node.getEventIDs()
+
+ ocf.logger.debug("monitor: curDocVersion has not been handled yet")
+
+ if clusterEventIDs:
+ # there are pending events set, so our state must be STOPPING or IN_EVENT
+ i = 0; touchedEventIDs = False
+ while i < len(clusterEventIDs):
+ # clean up pending events that are already finished according to AZ
+ if clusterEventIDs[i] not in localAzEventIDs:
+ ocf.logger.info("monitor: remove finished local clusterEvent %s" % (clusterEventIDs[i]))
+ clusterEventIDs.pop(i)
+ touchedEventIDs = True
+ else:
+ i += 1
+ if len(clusterEventIDs) > 0:
+ # there are still pending events (either because we're still stopping, or because the event is still in place)
+ # either way, we need to wait
+ if touchedEventIDs:
+ ocf.logger.info("monitor: added new local clusterEvent %s" % str(clusterEventIDs))
+ self.node.setEventIDs(clusterEventIDs)
+ else:
+ ocf.logger.info("monitor: no local clusterEvents were updated")
+ else:
+ # there are no more pending events left after cleanup
+ if clusterHelper.noPendingResourcesOnNode(self.node.hostName):
+ # and no pending resources on the node -> set it back online
+ ocf.logger.info("monitor: all local events finished -> clean up, put node online and AVAILABLE")
+ curState = self.node.updateNodeStateAndEvents(AVAILABLE, None)
+ self.node.putNodeOnline()
+ clusterHelper.removeHoldFromNodes()
+ # If Azure Scheduled Events are not used for 24 hours (e.g. because the cluster was asleep), it will be disabled for a VM.
+ # When the cluster wakes up and starts using it again, the DocumentIncarnation is reset.
+ # We need to remove it during cleanup, otherwise azure-events-az will not process the event after wakeup
+ self.node.setAttr(attr_lastDocVersion, None)
+ else:
+ ocf.logger.info("monitor: all local events finished, but some resources have not completed startup yet -> wait")
+ else:
+ if curState == AVAILABLE:
+ if len(localAzEventIDs) > 0:
+ if clusterHelper.otherNodesAvailable(self.node):
+ ocf.logger.info("monitor: can handle local events %s -> set state STOPPING" % (str(localAzEventIDs)))
+ curState = self.node.updateNodeStateAndEvents(STOPPING, localAzEventIDs)
+ else:
+ ocf.logger.info("monitor: cannot handle azEvents %s (only node available) -> set state ON_HOLD" % str(localAzEventIDs))
+ self.node.setState(ON_HOLD)
+ else:
+ ocf.logger.debug("monitor: no local azEvents to handle")
+
+ if curState == STOPPING:
+ eventIDsForNode = {}
+ if clusterHelper.noPendingResourcesOnNode(self.node.hostName):
+ if not self.node.isNodeInStandby():
+ ocf.logger.info("monitor: all local resources are started properly -> put node standby and exit")
+ self.node.putNodeStandby()
+ return ocf.OCF_SUCCESS
+
+ for e in localEvents:
+ ocf.logger.info("monitor: handling remote event %s (%s; nodes = %s)" % (e.EventId, e.EventType, str(e.Resources)))
+ # before we can force an event to start, we need to ensure all nodes involved have stopped their resources
+ if e.EventStatus == "Scheduled":
+ allNodesStopped = True
+ for azName in e.Resources:
+ hostName = clusterHelper.getHostNameFromAzName(azName)
+ state = self.node.getState(node=hostName)
+ if state == STOPPING:
+ # the only way we can continue is when node state is STOPPING, but all resources have been stopped
+ if not clusterHelper.allResourcesStoppedOnNode(hostName):
+ ocf.logger.info("monitor: (at least) node %s has still resources running -> wait" % hostName)
+ allNodesStopped = False
+ break
+ elif state in (AVAILABLE, IN_EVENT, ON_HOLD):
+ ocf.logger.info("monitor: node %s is still %s -> remote event needs to be picked up locally" % (hostName, nodeStateToString(state)))
+ allNodesStopped = False
+ break
+ if allNodesStopped:
+ ocf.logger.info("monitor: nodes %s are stopped -> add remote event %s to force list" % (str(e.Resources), e.EventId))
+ for n in e.Resources:
+ hostName = clusterHelper.getHostNameFromAzName(n)
+ if hostName in eventIDsForNode:
+ eventIDsForNode[hostName].append(e.EventId)
+ else:
+ eventIDsForNode[hostName] = [e.EventId]
+ elif e.EventStatus == "Started":
+ ocf.logger.info("monitor: remote event already started")
+
+ # force the start of all events whose nodes are ready (i.e. have no more resources running)
+ if len(eventIDsForNode.keys()) > 0:
+ eventIDsToForce = set([item for sublist in eventIDsForNode.values() for item in sublist])
+ ocf.logger.info("monitor: set nodes %s to IN_EVENT; force remote events %s" % (str(eventIDsForNode.keys()), str(eventIDsToForce)))
+ for node, eventId in eventIDsForNode.items():
+ self.node.updateNodeStateAndEvents(IN_EVENT, eventId, node=node)
+ azHelper.forceEvents(eventIDsToForce)
+ self.node.setAttr(attr_lastDocVersion, curDocVersion)
+ else:
+ ocf.logger.info("monitor: some local resources are not clean yet -> wait")
+
+ ocf.logger.debug("monitor: finished")
+ return ocf.OCF_SUCCESS
+
+##############################################################################
+
+def setLoglevel(verbose):
+ # set up writing into syslog
+ loglevel = default_loglevel
+ if verbose:
+ opener = urllib2.build_opener(urllib2.HTTPHandler(debuglevel=1))
+ urllib2.install_opener(opener)
+ loglevel = ocf.logging.DEBUG
+ ocf.log.setLevel(loglevel)
+
+description = (
+ "Microsoft Azure Scheduled Events monitoring agent",
+ """This resource agent implements a monitor for scheduled
+(maintenance) events for a Microsoft Azure VM.
+
+If any relevant events are found, it moves all Pacemaker resources
+away from the affected node to allow for a graceful shutdown.
+
+ Usage:
+ [OCF_RESKEY_eventTypes=VAL] [OCF_RESKEY_verbose=VAL] azure-events-az ACTION
+
+ action (required): Supported values: monitor, help, meta-data
+ eventTypes (optional): List of event types to be considered
+ relevant by the resource agent (comma-separated).
+ Supported values: Freeze,Reboot,Redeploy
+ Default = Reboot,Redeploy
+/ verbose (optional): If set to true, displays debug info.
+ Default = false
+
+ Deployment:
+ crm configure primitive rsc_azure-events-az ocf:heartbeat:azure-events-az \
+ op monitor interval=10s
+ crm configure clone cln_azure-events-az rsc_azure-events-az
+
+For further information on Microsoft Azure Scheduled Events, please
+refer to the following documentation:
+https://docs.microsoft.com/en-us/azure/virtual-machines/linux/scheduled-events
+""")
+
+def monitor_action(eventTypes):
+ relevantEventTypes = set(eventTypes.split(",") if eventTypes else [])
+ ra = raAzEvents(relevantEventTypes)
+ return ra.monitor()
+
+def validate_action(eventTypes):
+ if eventTypes:
+ for event in eventTypes.split(","):
+ if event not in ("Freeze", "Reboot", "Redeploy"):
+ ocf.ocf_exit_reason("Event type not one of Freeze, Reboot, Redeploy: " + eventTypes)
+ return ocf.OCF_ERR_CONFIGURED
+ return ocf.OCF_SUCCESS
+
+def main():
+ agent = ocf.Agent("azure-events-az", shortdesc=description[0], longdesc=description[1])
+ agent.add_parameter(
+ "eventTypes",
+ shortdesc="List of resources to be considered",
+ longdesc="A comma-separated list of event types that will be handled by this resource agent. (Possible values: Freeze,Reboot,Redeploy)",
+ content_type="string",
+ default="Reboot,Redeploy")
+ agent.add_parameter(
+ "verbose",
+ shortdesc="Enable verbose agent logging",
+ longdesc="Set to true to enable verbose logging",
+ content_type="boolean",
+ default="false")
+ agent.add_action("start", timeout=10, handler=lambda: ocf.OCF_SUCCESS)
+ agent.add_action("stop", timeout=10, handler=lambda: ocf.OCF_SUCCESS)
+ agent.add_action("validate-all", timeout=20, handler=validate_action)
+ agent.add_action("monitor", timeout=240, interval=10, handler=monitor_action)
+ setLoglevel(ocf.is_true(ocf.get_parameter("verbose", "false")))
+ agent.run()
+
+if __name__ == '__main__':
+ main()
\ No newline at end of file
From a95337d882c7cc69d604b050159ad50b679f18be Mon Sep 17 00:00:00 2001
From: MSSedusch <sedusch@microsoft.com>
Date: Thu, 2 Jun 2022 14:10:33 +0200
Subject: [PATCH 2/2] Remove developer documentation
---
heartbeat/azure-events-az.in | 11 -----------
1 file changed, 11 deletions(-)
diff --git a/heartbeat/azure-events-az.in b/heartbeat/azure-events-az.in
index 616fc8d9e..59d095306 100644
--- a/heartbeat/azure-events-az.in
+++ b/heartbeat/azure-events-az.in
@@ -723,17 +723,6 @@ description = (
If any relevant events are found, it moves all Pacemaker resources
away from the affected node to allow for a graceful shutdown.
- Usage:
- [OCF_RESKEY_eventTypes=VAL] [OCF_RESKEY_verbose=VAL] azure-events-az ACTION
-
- action (required): Supported values: monitor, help, meta-data
- eventTypes (optional): List of event types to be considered
- relevant by the resource agent (comma-separated).
- Supported values: Freeze,Reboot,Redeploy
- Default = Reboot,Redeploy
-/ verbose (optional): If set to true, displays debug info.
- Default = false
-
Deployment:
crm configure primitive rsc_azure-events-az ocf:heartbeat:azure-events-az \
op monitor interval=10s

147
SOURCES/bz2134542-IPsrcaddr-proto-metric-scope-default-route-fixes.patch Normal file
View File

@ -0,0 +1,147 @@
From 237d55120a7c8d761f839c96651e722b3bb3bc88 Mon Sep 17 00:00:00 2001
From: Oyvind Albrigtsen <oalbrigt@redhat.com>
Date: Wed, 12 Oct 2022 13:57:30 +0200
Subject: [PATCH 1/4] IPsrcaddr: fix PROTO regex
---
heartbeat/IPsrcaddr | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/heartbeat/IPsrcaddr b/heartbeat/IPsrcaddr
index 7dbf65ff5..24406d296 100755
--- a/heartbeat/IPsrcaddr
+++ b/heartbeat/IPsrcaddr
@@ -188,7 +188,7 @@ IPADDR="\($OCTET\.\)\{3\}$OCTET"
SRCCLAUSE="src$WS$WS*\($IPADDR\)"
MATCHROUTE="\(.*${WS}\)\($SRCCLAUSE\)\($WS.*\|$\)"
METRICCLAUSE=".*\(metric$WS[^ ]\+\)"
-PROTOCLAUSE=".*\(proto$WS[^ ]\+\)"
+PROTOCLAUSE=".*\(proto$WS[^ ]\+\).*"
FINDIF=findif
# findif needs that to be set
From c70ba457851a401cb201cb87d23bdbc5f4fcd2b3 Mon Sep 17 00:00:00 2001
From: Oyvind Albrigtsen <oalbrigt@redhat.com>
Date: Wed, 12 Oct 2022 14:00:30 +0200
Subject: [PATCH 2/4] IPsrcaddr: detect metric for main table only, and allow
specifying metric if necessary
---
heartbeat/IPsrcaddr | 18 +++++++++++++++++-
1 file changed, 17 insertions(+), 1 deletion(-)
diff --git a/heartbeat/IPsrcaddr b/heartbeat/IPsrcaddr
index 24406d296..4745eb8a7 100755
--- a/heartbeat/IPsrcaddr
+++ b/heartbeat/IPsrcaddr
@@ -59,12 +59,14 @@ OCF_RESKEY_ipaddress_default=""
OCF_RESKEY_cidr_netmask_default=""
OCF_RESKEY_destination_default="0.0.0.0/0"
OCF_RESKEY_proto_default=""
+OCF_RESKEY_metric_default=""
OCF_RESKEY_table_default=""
: ${OCF_RESKEY_ipaddress=${OCF_RESKEY_ipaddress_default}}
: ${OCF_RESKEY_cidr_netmask=${OCF_RESKEY_cidr_netmask_default}}
: ${OCF_RESKEY_destination=${OCF_RESKEY_destination_default}}
: ${OCF_RESKEY_proto=${OCF_RESKEY_proto_default}}
+: ${OCF_RESKEY_metric=${OCF_RESKEY_metric_default}}
: ${OCF_RESKEY_table=${OCF_RESKEY_table_default}}
#######################################################################
@@ -143,6 +145,14 @@ Proto to match when finding network. E.g. "kernel".
<content type="string" default="${OCF_RESKEY_proto_default}" />
</parameter>
+<parameter name="metric">
+<longdesc lang="en">
+Metric. Only needed if incorrect metric value is used.
+</longdesc>
+<shortdesc lang="en">Metric</shortdesc>
+<content type="string" default="${OCF_RESKEY_metric_default}" />
+</parameter>
+
<parameter name="table">
<longdesc lang="en">
Table to modify. E.g. "local".
@@ -548,8 +558,14 @@ rc=$?
INTERFACE=`echo $findif_out | awk '{print $1}'`
LISTROUTE=`$IP2UTIL route list dev $INTERFACE scope link $PROTO match $ipaddress`
-METRIC=`echo $LISTROUTE | sed -n "s/$METRICCLAUSE/\1/p"`
[ -z "$PROTO" ] && PROTO=`echo $LISTROUTE | sed -n "s/$PROTOCLAUSE/\1/p"`
+if [ -n "$OCF_RESKEY_metric" ]; then
+ METRIC="metric $OCF_RESKEY_metric"
+elif [ -z "$TABLE" ] || [ "${TABLE#table }" = "main" ]; then
+ METRIC=`echo $LISTROUTE | sed -n "s/$METRICCLAUSE/\1/p"`
+else
+ METRIC=""
+fi
if [ "$OCF_RESKEY_destination" = "0.0.0.0/0" ] ;then
NETWORK=`echo $LISTROUTE | grep -m 1 -o '^[^ ]*'`
From c514f12f7a19440f475938f2a4659e5e9667fa25 Mon Sep 17 00:00:00 2001
From: Oyvind Albrigtsen <oalbrigt@redhat.com>
Date: Wed, 12 Oct 2022 14:01:26 +0200
Subject: [PATCH 3/4] IPsrcaddr: use scope host when using non-main tables
---
heartbeat/IPsrcaddr | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/heartbeat/IPsrcaddr b/heartbeat/IPsrcaddr
index 4745eb8a7..926246008 100755
--- a/heartbeat/IPsrcaddr
+++ b/heartbeat/IPsrcaddr
@@ -279,8 +279,14 @@ srca_stop() {
[ $rc = 2 ] && errorexit "The address you specified to stop does not match the preferred source address"
+ if [ -z "$TABLE" ] || [ "${TABLE#table }" = "main" ]; then
+ SCOPE="link"
+ else
+ SCOPE="host"
+ fi
+
PRIMARY_IP="$($IP2UTIL -4 -o addr show dev $INTERFACE primary | awk '{split($4,a,"/");print a[1]}')"
- OPTS="proto kernel scope link src $PRIMARY_IP"
+ OPTS="proto kernel scope $SCOPE src $PRIMARY_IP"
$IP2UTIL route replace $TABLE $NETWORK dev $INTERFACE $OPTS $METRIC || \
errorexit "command 'ip route replace $TABLE $NETWORK dev $INTERFACE $OPTS $METRIC' failed"
From 1f387ac8017b3eee23b41eadafd58ce21a29eb21 Mon Sep 17 00:00:00 2001
From: Oyvind Albrigtsen <oalbrigt@redhat.com>
Date: Thu, 13 Oct 2022 13:11:28 +0200
Subject: [PATCH 4/4] IPsrcaddr: fix monitor/status for default route not being
equal to src IP before start, and change route src correctly in stop-action
---
heartbeat/IPsrcaddr | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/heartbeat/IPsrcaddr b/heartbeat/IPsrcaddr
index 926246008..1bd41a930 100755
--- a/heartbeat/IPsrcaddr
+++ b/heartbeat/IPsrcaddr
@@ -229,6 +229,7 @@ srca_read() {
[ -z "$SRCIP" ] && return 1
[ $SRCIP = $1 ] && return 0
+ [ "$__OCF_ACTION" = "monitor" ] || [ "$__OCF_ACTION" = "status" ] && [ "${ROUTE%% *}" = "default" ] && return 1
return 2
}
@@ -292,8 +293,8 @@ srca_stop() {
errorexit "command 'ip route replace $TABLE $NETWORK dev $INTERFACE $OPTS $METRIC' failed"
if [ "$OCF_RESKEY_destination" = "0.0.0.0/0" ] ;then
- $CMDCHANGE $ROUTE_WO_SRC || \
- errorexit "command '$CMDCHANGE $ROUTE_WO_SRC' failed"
+ $CMDCHANGE $ROUTE_WO_SRC src $PRIMARY_IP || \
+ errorexit "command '$CMDCHANGE $ROUTE_WO_SRC src $PRIMARY_IP' failed"
fi
return $?

16
SPECS/resource-agents.spec
View File

@ -45,7 +45,7 @@
Name: resource-agents
Summary: Open Source HA Reusable Cluster Resource Scripts
Version: 4.10.0
Release: 9%{?rcver:%{rcver}}%{?numcomm:.%{numcomm}}%{?alphatag:.%{alphatag}}%{?dirty:.%{dirty}}%{?dist}.3
Release: 9%{?rcver:%{rcver}}%{?numcomm:.%{numcomm}}%{?alphatag:.%{alphatag}}%{?dirty:.%{dirty}}%{?dist}.5
License: GPLv2+ and LGPLv2+
URL: https://github.com/ClusterLabs/resource-agents
Source0: %{upstream_prefix}-%{upstream_version}.tar.gz
@ -62,6 +62,8 @@ Patch9: bz2055016-2-IPsrcaddr-error-message-route-not-found.patch
Patch10: bz2055016-3-IPsrcaddr-fix-indentation.patch
Patch11: bz2055016-4-IPsrcaddr-fixes.patch
Patch12: bz2087930-NovaEvacuate-add-user_domain-project_domain.patch
Patch13: bz2130990-azure-events-az-new-ra.patch
Patch14: bz2134542-IPsrcaddr-proto-metric-scope-default-route-fixes.patch
# bundled ha-cloud-support libs
Patch500: ha-cloud-support-aws.patch
@ -198,6 +200,8 @@ exit 1
%patch10 -p1
%patch11 -p1
%patch12 -p1
%patch13 -p1
%patch14 -p1
# bundled ha-cloud-support libs
%patch500 -p1
@ -528,6 +532,16 @@ rm -rf %{buildroot}/usr/share/doc/resource-agents
%{_usr}/lib/ocf/lib/heartbeat/OCF_*.pm
%changelog
* Fri Oct 14 2022 Oyvind Albrigtsen <oalbrigt@redhat.com> - 4.10.0-9.5
- IPsrcaddr: proto, metric, scope and default route fixes
Resolves: rhbz#2134542
* Tue Oct 4 2022 Oyvind Albrigtsen <oalbrigt@redhat.com> - 4.10.0-9.4
- azure-events-az: new resource agent
Resolves: rhbz#2130990
* Thu Jun 9 2022 Oyvind Albrigtsen <oalbrigt@redhat.com> - 4.10.0-9.3
- NovaEvacuate: add user_domain and project_domain parameters