VM server reboot

[solaris123]

Why does a VM server reboot after being migrated to another host when one host fails specifically my Linux VM server? I  am running esxi5.5

[vThinkBeyondVM]

I suspect, you are using vSphere HA in your environment. Please confirm.

In case of host failure, HA restart the VMs on another available host in the HA cluster. (vSphere HA cluster is cluster of ESXi hosts). Note that it does not vMotion, as there is shared datastore, it just un-register from failed host and re-register on new host and re-started.

[solaris123]

Yes we are using HA in our enviroment, and thanks for your response. However I'm not seeing what you are describing across all VM's on the failed host, out of 15 VM's running on that failed host, i only see about 1 or 2 that are actually rebooted, what could be causing the reboot to take place. It is my understanding that when a host fails, the VM's are vmotioned over to another host without a reboot of the VM, am I missing something here.

[bspagna89]

Hi,

Can you confirm that the VM in question that reboots upon migration has VMware tools installed? Also, are you using HA VM monitoring? If so, it would seem that if the VM fails to heartbeat during a vMotion that HA is restarting it thinking it crashed.

[Sateesh_vCloud]

It is my understanding that when a host fails, the VM's are vmotioned over to another host without a reboot of the VM, am I missing something here. (YES Reboot will happen when HA invoked)

HA never use V-Motion

Please confIrm that migrations are happening via DRS but not by HA

[vThinkBeyondVM]

I can confirm you that vSphere HA restarts the VMs (no vMotion). vMotion is used by DRS for load balancing. In case of vMotion, you will not have any downtime but in case of any HA solution, you will have minimal downtime. In vSphere HA's case, restart time is downtime.

Refer to understand the same in detail: VMware documentation on vSphere HA: vSphere Documentation Center

In you case, we need to dig more in order to understand what exactly is happening. are you using DRS as well? Please give more details on your cluster.

[solaris123]

Thanks to all for your input, very helpful. I believe i got my answer.

[antonmajor]

When a host fails, ha will restart(reboot) the failed virtual machines on another functioning host in the cluster.

If you need further resiliency on some critical production servers, then you may consider scaling out your app design to support cluster/failover OR simply enable fault tolerance if it meets FT requirements and wins an argument against additional resource consumption across your cluster.