Cluster restart (best practice)

[Gerhard W. Recher]

Hi Folks,

I wonder not seeing any posts about this.

Scenario: You update/dist-upgrade each node and need a restart because of kernel enhancements...

I stop all vm's and containers on all nodes, and restart all nodes with reboot on cli in timely short sequenze ...

this always cause some heavy ceph activities ... and i'm not really sure if this is the rigth way to accomplish a clean cluster restart of e.g. 4 nodes ....

regards

Gerhard

 

[dcsapak]

i would restart one node, wait until everything is settled, and the restart the next node and so on..

 

[dfenwick]

i would restart one node, wait until everything is settled, and the restart the next node and so on..

That's exactly how I do it. It presumes there's ample capacity within the other nodes in your cluster for the VM migration, but it works.