Hi,
We have Proxmox clusters on OVH servers, and currently the vRacks (private networks between the servers) are very unstable (I mean more than usual).
We are seeing a lot of 2 - 3 minutes cut in the network, I'm pretty sure they are changing a bunch of switchs every day. I increased the glusterfs timeout to 5 minutes, but I also need to increase the proxmox timeout to avoid every node self-fencing.
How would I do that ?
Yes, I know, it's bad but I have no choice. Currently all the VMs are crashing, switching read only or just being turned off by self fencing every day, we can't go on like this. We need to increase the timeout for a little while, until OVH finally stops doing whatever they are doing.
Since glusterfs has it's own split-brain prevention mechanism, and it works pretty well, I'm not worried about split brains. There shouldn't be any difference between 1 minute and 5 minutes of timeout.
Thanks
We have Proxmox clusters on OVH servers, and currently the vRacks (private networks between the servers) are very unstable (I mean more than usual).
We are seeing a lot of 2 - 3 minutes cut in the network, I'm pretty sure they are changing a bunch of switchs every day. I increased the glusterfs timeout to 5 minutes, but I also need to increase the proxmox timeout to avoid every node self-fencing.
How would I do that ?
Yes, I know, it's bad but I have no choice. Currently all the VMs are crashing, switching read only or just being turned off by self fencing every day, we can't go on like this. We need to increase the timeout for a little while, until OVH finally stops doing whatever they are doing.
Since glusterfs has it's own split-brain prevention mechanism, and it works pretty well, I'm not worried about split brains. There shouldn't be any difference between 1 minute and 5 minutes of timeout.
Thanks