Hello, I changed timers for HA settings to minimize the number of attempts to recover the hosts and maximize the SLA for VMs.
For now if I run the echo c > /proc/sysrq-trigger the host failed and the HA state show Degraded and after a while changed to recovered state, however the VMs is not accessible. How can I change settings to disable the number of attempts for recover or set it to 1 time attempt for recover and if host not recovered (OOBM ON and server loading failed) trigger HA event and restart VMs on another host (like VMware VCenter perform the HA event) Appreciate your support Regards, Jafar Aghabalayev -----Original Message----- From: Boris Stoyanov <[email protected]> Sent: Saturday, February 11, 2023 2:37 PM To: [email protected]; Wei ZHOU <[email protected]> Subject: Re: Cluster HA - KVM Hosts CAUTION: This email originated from outside the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. Please report all suspicious emails to [email protected]. How did you power off the host? If you shut it down gracefully HA will not kick in, hence considered an intended action. The goal of this feature is to guard against unexpected events, like for example host has crashed. If you're looking to test it out you can try to provoke a kernel panic in the kvm host, something like: echo c > /proc/sysrq-trigger Then soon HA investigator will determin the host as unhealthy, (pings failing), soon after that there will be an activity checks on the VMs disks running on the host. It will check if there's any changes and disk activity, if that also fails it will determine the host for recovery. It will issue a restart with the IPMI tool, if number of restarts (depending on configuration) also fails it will fence the host and will start the VMs on another host in the same cluster. I hope this is insightful for you. Let me know how it goes at your end. Bobby. From: Jafar Aghabalayev <[email protected]> Date: Saturday, 11 February 2023, 11:21 To: Wei ZHOU <[email protected]>, [email protected] <[email protected]> Subject: Re: Cluster HA - KVM Hosts Hello, I created new offering, with HA enabled and used it Sent from Outlook for Android<https://aka.ms/AAb9ysg> ________________________________ From: Wei ZHOU <[email protected]> Sent: Saturday, February 11, 2023 12:44:45 PM To: [email protected] <[email protected]> Cc: Jafar Aghabalayev <[email protected]> Subject: Re: Cluster HA - KVM Hosts CAUTION: This email originated from outside the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. Please report all suspicious emails to [email protected]. Hi, Is HA enabled on the VM? You can check the vm details. HA is not enabled for built-in service offerings (Small/Medium). -Wei On Saturday, 11 February 2023, [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> wrote: Hi Jafar, FYI. VM HA does not work when server is stopped gracefully * Issue #4211 * apache/cloudstack | | | | | | | | | | | VM HA does not work when server is stopped gracefully * Issue #4211 * apache/cloudstack When we stop a server (hypervisor) gracefully, vms on it will NOT be started on other servers even HA is enabled. ISSUE TYPE Bug Report COMPONENT NAME VM HA CLOUDSTACK VERSION 4.14/4.15 CONFIGURATI... | | | | Thanks Pradeep Sent from Yahoo Mail on Android On Sat, 11 Feb 2023 at 12:02 am, Jafar Aghabalayev<[email protected]<mailto:[email protected]>> wrote: Hello Community, I have configured cluster with 2 KVM Hosts. HA enabled for hosts, ha.tag attribute on global configuration is the same as for ha.tag attribute for hosts. VMs are running with HA enabled offering. NFS used as primary storage. I tried to shutdown one of the hosts, however HA not performed. State and power state indicated as down, but resource state show as up and VM located at the failed host show in running state (in real it is unaccessible). I tried to stop libvirtd service and the result same. Can anyone help me with this issue?
