Ah do you test cluster fencing ?
Personally too and without any problem.
But molly-guard is only efficient and enabled in SSH interactive mode.
Thus it shouldn't interfere with internal Proxmox mechanisms (non interactive mode).
BUT ...
Awesome news! We will get it into our official testing pipeline ASAP!
Great work!
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
Indeed. But this situation should not happen. If a node is fenced, having CEPH out of order is OK for me. The problem on the point you mention is that CEPH OSDs didn't start after cluster rejoin.
And here the case with same cluster partition on the two DCs:
========================================================================
========================================================================...
Hello,
With Damien BROCHARD (Ozerim), we made some extensive CEPH tests in lab. Our running and functional lab : we configured a simulation of 2 DC cluster, with CEPH stretched cluster mode and a 3rd DC of only one PVE as tie breaker (CEPH) /...
Hello,
With Damien BROCHARD (Ozerim), we made some extensive CEPH tests in lab.
As I thought we cannot add a new CEPH monitor in proxmox after stretched cluster mode has been enabled :
root@training2-vsrv2:~# pveceph mon destroy
400 not enough...
I missed the reply, but.. I was indeed pinpointing the issue that could happen without the stretch mode enabled, as it was suggested as a workaround.. But yeah.