0

Good morning everyone -

Running a 6 node Pacemaker/Corosync cluster with lots of productive qemu/kvm VMs on it, we were wondering how to best do an "emergency shutdown" of the whole lot, ie. when the UPS signals "AC fail" and "low battery". After reading docs and forums, we are still not sure about the following:

  1. is there a way of addressing all running resources at once we did not see? Neither pcs resource disable --all nor pcs resource disable \* works...
  2. hence: how to safely get a (parse-able) list of resources running? Am I supposed to really parse the output of pcs resource ...?
  3. will pcs cluster stop --all shut down (or at least try to) all resources beforehand (that is to mean: runs pcs resource disable ... on all resources)? Or is it rather a kill -9 to all resources?

Thanks for all enlightenment!

Christian
  • 295
  • 1
  • 7
  • ... forgot to say that we've got foisted with the cluster which was set up by someone else, and have no test environment... – Christian Jan 17 '22 at 13:39

1 Answers1

1

i'm currently facing the same problem. We have a two node cluster with about 20 qemu/kvm vm's. A power outtage a short while ago showed me that our cluster shutdown concept does not work. I'm asking the same question as you in the pacemaker mailing list. Here is the archive: https://lists.clusterlabs.org/pipermail/users/2022-February/thread.html , it seems to be up t odate.

Bernd

eferrari
  • 11
  • 1