Thx again, Slavka!
I am going to open an issue regarding this. I would expect another behavior as 
an administrator.

Regards,
Swen


-----Ursprüngliche Nachricht-----
Von: Slavka Peleva <slav...@storpool.com.INVALID> 
Gesendet: Dienstag, 20. Februar 2024 09:22
An: users@cloudstack.apache.org
Betreff: Re: ha checks on kvm hosts running even with ha disabled

Unfortunately, there is no check if the HA is enabled/disabled in the agent's 
code. The `reboot.host.and.alert.management.on.heartbeat.timeout`
property is enabled by default and only manually you can disable it.

Best regards,
Slavka

On Tue, Feb 20, 2024 at 10:14 AM <m...@swen.io> wrote:

> Hi Slavka,
>
> thx for your reply!
>
> Do you know the reason why KVMHAMonitor will be initialized even when 
> host ha has been disabled by UI?
> From an admin point of view
> reboot.host.and.alert.management.on.heartbeat.timeout=false should be 
> set automatically when you disable host ha. Is there a reason for not doing 
> it?
>
> Regards,
> Swen
>
> -----Ursprüngliche Nachricht-----
> Von: Slavka Peleva <slav...@storpool.com.INVALID>
> Gesendet: Dienstag, 20. Februar 2024 08:50
> An: users@cloudstack.apache.org
> Betreff: Re: ha checks on kvm hosts running even with ha disabled
>
> Hi Swen,
>
> The KVMHAMonitor is initialised here
> <
> https://github.com/apache/cloudstack/blob/8f6721ed4c4e1b31081a951c62ff
> be5331cf16d4/plugins/hypervisors/kvm/src/main/java/com/cloud/hyperviso
> r/kvm/resource/LibvirtComputingResource.java#L1202
> >
> even
> if you didn't enable the HA.
> If I may advise you to disable in the `agent.properties` file the 
> property 
> `reboot.host.and.alert.management.on.heartbeat.timeout=false`. 
> CloudStack will still execute the heartbeat check but won't restart the agent 
> host on failure.
>
> Best regards,
> Slavka
>
> On Tue, Feb 20, 2024 at 12:41 AM Swen <m...@swen.io> wrote:
>
> > Hi all,
> >
> > we encountered a strange issue today in our lab installation. We are 
> > running CS 4.19.0 upgraded from CS 4.18.1 with linstor as primary 
> > storage. We are using a linstor jar file provided by Linbit which is 
> > not the default one part of default 4.19.0!
> >
> > This updated plugin already includes a new feature to use linstor 
> > storage for host ha. I provide this only for information reasons.
> >
> > The issue we encountered was that even we do not have HA enabled on 
> > cluster and host level, cloudstack agent on our KVM hosts triggered 
> > HA actions and rebooted our hosts. We found this on our agent.log:
> >
> > Feb 19 11:53:05 pc-kvm-2 java[6617]: WARN  
> > [kvm.resource.KVMHAMonitor]
> > (Thread-1:) (logid:) Write heartbeat for pool 
> > [71c272d3-b180-4b18-a0fc-cfc1dc5b86c9] failed: Down; try: 2 of 5.
> >
> > Feb 19 11:58:58 pc-kvm-2 java[9465]: WARN  
> > [kvm.resource.KVMHAMonitor]
> > (Thread-1:) (logid:) Write heartbeat for pool 
> > [71c272d3-b180-4b18-a0fc-cfc1dc5b86c9] failed: Down; try: 3 of 5.
> >
> > Feb 19 12:00:08 pc-kvm-2 java[9465]: WARN  
> > [kvm.resource.KVMHAMonitor]
> > (Thread-1:) (logid:) Write heartbeat for pool 
> > [71c272d3-b180-4b18-a0fc-cfc1dc5b86c9] failed: Down; try: 4 of 5.
> >
> > Feb 19 12:01:08 pc-kvm-2 java[9465]: WARN  
> > [kvm.resource.KVMHAMonitor]
> > (Thread-1:) (logid:) Write heartbeat for pool 
> > [71c272d3-b180-4b18-a0fc-cfc1dc5b86c9] failed: Down; try: 5 of 5.
> >
> > Feb 19 12:01:08 pc-kvm-2 java[9465]: WARN  
> > [kvm.resource.KVMHAMonitor]
> > (Thread-1:) (logid:) Write heartbeat for pool 
> > [71c272d3-b180-4b18-a0fc-cfc1dc5b86c9] failed: Down; stopping 
> > cloudstack-agent.
> >
> > Feb 19 12:02:08 pc-kvm-2 heartbeat: kvmspheartbeat.sh will reboot 
> > system because it was unable to write the heartbeat to the storage.
> >
> > We and Linbit did some debugging. We tried to understand why the 
> > cloudstack agent is running those checks in the first place. We were 
> > unable to find any code which checks if host HA is enabled or not 
> > and will not perform HA tasks if HA is disabled. Can somebody please 
> > double-check this?
> >
> > Thank you very much!
> >
> > Regards,
> > Swen
> >
> >
> >
>
>
>


Reply via email to