KVM cluster with NFS primary storage – VM HA not working when host is powered down

[akoskuczi-bw]

problem

In a KVM cluster with NFS primary storage, VM HA does not work when a host is powered down.

• The host status transitions to Down, HA state shows Fenced.
• VMs from the powered-down host are not restarted on other available hosts in the cluster.
• Both Host HA and VM HA are enabled.
• OOB driver: IPMI.

Expected behavior

VMs from the failed host should be restarted on other available hosts in the cluster.

Actual behavior

• Host goes to Down and HA state Fenced.
• VMs are not started elsewhere.
• Management server logs show a NoTransitionException.

Relevant log snippet

WARN [o.a.c.h.HAManagerImpl] (BackgroundTaskPollManager-4:[ctx-c2bf501d]) (logid:96e12771) Unable to find next HA state for current HA state=[Fenced] for event=[Ineligible] for host Host {"id":4,"name":"csh-1-2.clab.run","type":"Routing","uuid":"f8f86177-f0e3-4994-8609-dd55e0e35a3e"} with id 4. com.cloud.utils.fsm.NoTransitionException: Unable to transition to a new state from Fenced via Ineligible
at com.cloud.utils.fsm.StateMachine2.getTransition(StateMachine2.java:108)
at com.cloud.utils.fsm.StateMachine2.getNextState(StateMachine2.java:94)
at org.apache.cloudstack.ha.HAManagerImpl.transitionHAState(HAManagerImpl.java:153)
at org.apache.cloudstack.ha.HAManagerImpl.validateAndFindHAProvider(HAManagerImpl.java:233)
at org.apache.cloudstack.ha.HAManagerImpl$HAManagerBgPollTask.runInContext(HAManagerImpl.java:665)
at org.apache.cloudstack.managed.context.ManagedContextRunnable$1.run(ManagedContextRunnable.java:49)
at org.apache.cloudstack.managed.context.impl.DefaultManagedContext$1.call(DefaultManagedContext.java:56)
at org.apache.cloudstack.managed.context.impl.DefaultManagedContext.callWithContext(DefaultManagedContext.java:103)
at org.apache.cloudstack.managed.context.impl.DefaultManagedContext.runWithContext(DefaultManagedContext.java:53)
at org.apache.cloudstack.managed.context.ManagedContextRunnable.run(ManagedContextRunnable.java:46)
at java.base/java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:539)
at java.base/java.util.concurrent.FutureTask.runAndReset(FutureTask.java:305)
at java.base/java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:305)
at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1136)
at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:635)
at java.base/java.lang.Thread.run(Thread.java:840)

versions

Environment

• CloudStack version: 4.20.1.0
• Hypervisor: KVM
• Primary storage: NFS
• HA settings: Host HA enabled, VM HA enabled, OOB driver = IPMI

The steps to reproduce the bug

1.1. Enable Host HA and VM HA in a KVM cluster (NFS primary storage).
2. Power off a host that runs VMs.
3. Observe host and VM states in the management server.

What to do about it?

No response

[boring-cyborg[bot]]

Thanks for opening your first issue here! Be sure to follow the issue template!

[kiranchavala]

@akoskuczi-bw

Could you please try the steps mentioned in this link

#10477 (comment)

cc @rajujith

[akoskuczi-bw]

VM-HA functions properly, but only when HOST-HA is disabled. When HOST-HA is also enabled on the hosts, the log contains the entries mentioned above, and the VMs fail to start on the healthy hosts even after several hours of waiting.

[DaanHoogland]

It would be nice if we can define however these should work nicely together. This has never been a focus of anybody’s attention.

[DaanHoogland]

VM-HA functions properly, but only when HOST-HA is disabled. When HOST-HA is also enabled on the hosts, the log contains the entries mentioned above, and the VMs fail to start on the healthy hosts even after several hours of waiting.

This right, @akoskuczi-bw , VM-HA and host-HA do not play nice together. This is a known issue.