btzq commented on issue #10479:
URL: https://github.com/apache/cloudstack/issues/10479#issuecomment-2692728611

   FYI, if you are going to test a Hypervisor Failure, to trigger the VM HA to 
auto start on another healthy hypervisor, i suggest to simulate a force shut 
down or Power Off instead. Just like how you simulate a real scenario is going 
to happen. 
   
   There have been reports that restarting a hypervisor doesnt trigger the VM 
HA cause it isnt considered an outage. 
   
   And there are some parameters u need to tune to reduce the time Cloudstack 
Management server needs to realise the hypervisor is down. If not, by default, 
it takes a loooong time and multiple people have reported this.
   
   But on the other than, you want to make sure not to tune these parameters 
too low, because you might cause a ‘split brain scenario’ (depending on your 
underlying storage). What if the hypervisor is still online? But Cloudstack 
Management cant reach it and thinks its down? So it started the VM in another 
hypervisor causing 2 VM to exist and talk to the same storage volume? 
Becareful… dependin on the storage u use (NFS, Linstor etc), there will be 
different recommendations of avoiding split brain.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to