HA without reboot VM

[Zriouil]

Hi All.

during automatic failover, is what the rebooting of virtuals machines is obligatory ?
we can't avoid it ? I meant in case of failure, the virtual machines fail over to another server (hot migration) without service interruption.

thank you so much in advance.

best regards.

 

[wolfgang]

Hi,
how should this work?
you never no when a failure occurs!
to make a live mig you must start the mig before the node fail.

 

[Zriouil]

Hi,
Thanks for reply.

I have already done live migration manually, all is well.
But, i wonder about the automatic migration that occurs after a crach (HA is configured), is it always cold ?

 

[dietmar]

But, i wonder about the automatic migration that occurs after a crach (HA is configured), is it always cold ?

Yes (because it is after crash).

 

[Zriouil]

I see, Thank you very much.

This is not good, huuh.

Best regards.

 

[dietmar]

This is not good, huuh.

This is how most HA stacks work (rgmanager, pacemaker, ...). The purpose is to get high availability, so short service interruption is acceptable.

 

[m.ardito]

No expert here, but I guess the only way to have what you wish is having a cluster of specific services, not HA VMs.
eg: if your VMs are serving web pages, you need a cluster of web servers in load balancing/failsave configuration, every server on a different host/hypervisor.
the same could go for mail services, etc. while is more difficult in other cases, though, imho (eg: vitrual desktops).

Marco