[ClusterLabs] pcs resource disable/enable is taking longer time in 32 node cluster setup for all components up and running.

[Tomas Jelinek]

Hi,

What method are you using to stop all resources?

If you run 'pcs resource disable <resource name>' in a loop, then it may 
take time for 300+ resources. By doing it this way, you run pacemaker 
scheduler for stopping each resource individually.

If you want to stop all resources, just run 'pcs property set 
stop-all-resources=true'.

If you want to stop all resources and control them individually, do it 
like this:
1) pcs cluster cib > original.xml
2) cp original.xml new.xml
3) repeat for all resources: pcs -f new.xml resource disable <resource name>
4) pcs cluster cib-push new.xml diff-against=original.xml
This way, pacemaker scheduler is run only once.

Regards,
Tomas


Dne 05. 03. 21 v 8:19 S Sathish S napsal(a):
> Hi Team,
> 
> We are setup 32 nodes pacemaker cluster setup each node has 10 resource 
> so total [around 300+ components] are up and running. Whenever we 
> perform pcs resource disable/enable command for all resource in a script 
> it take 5-8 mins in our test lab.
> 
> If we execute same in Single node, 3 nodes and 5 nodes, it is taking 
> less than 2 minutes. We never experienced delay up to 9 node cluster setup.
> 
> Currently, we don’t have 32 node cluster lab setup to reproduce the 
> issue and collect the logs, we would like to confirm i.e., Is there any 
> known issues faced or similar challenges when pcs is managing the 32 
> node cluster set up.
> 
> Thanks and Regards,
> 
> S Sathish S
> 
> 
> _______________________________________________
> Manage your subscription:
> https://lists.clusterlabs.org/mailman/listinfo/users
> 
> ClusterLabs home: https://www.clusterlabs.org/
>