[pve-devel] corosync bug: cluster break after 1 node clean shutdown

[Alexandre DERUMIER]

What I'm not sure, is what was the running libknet version.

package was upgraded everywhere to 1.16,  but I have notice that corosync process is not restarted when libknet is upgraded.

so it's quite possible that corosync was still running 1.13 libknet. (I need to find a way to  find last corosync or node  restart).

I think we should force corosync restart  on libknet upgrade. (or maybe bump corosync package version at the same time)