HA During Manager Outage

[dya]

Does HA function while the manager is down?

The specific cases are as follows:
1. Will virtual machines already running during the manager outage start on another virtual machine if the host fails?
2. Will virtual machines newly started during the manager outage start on another virtual machine if the host fails?

*Apologies if the machine translation is unclear.

[dya]

How about this matter?
Do you understand?

[PeterTzvetanov]

Hello dya,

the pacemaker/corosync will have to care about the take over of the VMs that needs to failover to the host distributed across the hosts left ok in the cluster.

With regards  

[dya]

Does HA function while the manager is down?

The specific cases are as follows:
1. Will virtual machines already running during the manager outage start on another virtual machine if the host fails?
2. Will virtual machines newly started during the manager outage start on another virtual machine if the host fails?

Is this what it means to be “be able to do”?

[mnavada]

Hello dya,

Hereis my response.

The specific cases are as follows:
1. Will virtual machines already running during the manager outage start on another virtual machine if the host fails?  
Ans: yes.
2. Will virtual machines newly started during the manager outage start on another virtual machine if the host fails?
Ans: yes.

The "pacemaker and corosyc" cluster services will be running on all KVM nodes. If any hosts fails, these cluster services will manage the movement of the VMs to other nodes in the cluster without any intervention of "VME manager".  

[dya]

1. Will virtual machines already running during the manager outage start on another virtual machine if the host fails?  
Ans: yes.
2. Will virtual machines newly started during the manager outage start on another virtual machine if the host fails?
Ans: yes.

Thank you. I understand.

The "pacemaker and corosyc" cluster services will be running on all KVM nodes. If any hosts fails, these cluster services will manage the movement of the VMs to other nodes in the cluster without any intervention of "VME manager".  

This is a new discovery.

I understood that virtual machines are not resources managed by Pacemaker, so I did not expect “Pacemaker and Corosyc” to manage the migration of virtual machines.

[dya]

Regarding the following, I still believe “pacemaker and corosyc” are unrelated, correct?
It's likely that the Morpheus agent (morphd) running on the host detected a host failure based on the heartbeat file and restarted the virtual machine on another host, isn't it?

The "pacemaker and corosyc" cluster services will be running on all KVM nodes. If any hosts fails, these cluster services will manage the movement of the VMs to other nodes in the cluster without any intervention of "VME manager".  

■"pcs status" and morphd status
Pacemaker manages only GFS2-related items and fencing.

■Heartbeat File
The following heartbeat file does not seem to update when morphd is stopped.