So I was changing some network related settings on 5 node ceph cluster (still in configuration/testing stage). To apply settings I rebooted nodes (3 of them monitors/managers, all five are osd hosts, sll nodes pve-v5.3-11) one by one while waiting till previous one comes back up and and then rebooting the next one after previous one stopped startup checks in . And after last monitor reboot I got no ceph control now. Menus in web interface are stuck on loading. I see global configuration, but nothing other. in OSD menu noout was set to on while I did this.
Did I run into some rookie mistake while doing this?
So far I haven't found anything one this while searching web. Probably will try to reconfigure nodes tomorrow I won't find a way to get them back up.
Any suggestions?
Few more notes:
* ceph status just hanges there...
* All nodes are reachable on all networks
CAUSE: jumbo frames 9000 broke ceph https://forum.proxmox.com/threads/l...got-timeout-in-proxmox-gui.50118/#post-233472 Interestingly this was it. Will check on what mtu works for this setup and comment for reference. Most likely somewhere in middle lacp or switches need a bit larger mtu than 9000
Did I run into some rookie mistake while doing this?
So far I haven't found anything one this while searching web. Probably will try to reconfigure nodes tomorrow I won't find a way to get them back up.
Any suggestions?
Few more notes:
* ceph status just hanges there...
* All nodes are reachable on all networks
CAUSE: jumbo frames 9000 broke ceph https://forum.proxmox.com/threads/l...got-timeout-in-proxmox-gui.50118/#post-233472 Interestingly this was it. Will check on what mtu works for this setup and comment for reference. Most likely somewhere in middle lacp or switches need a bit larger mtu than 9000
Last edited: