FlusterFS and HA, Stroage 'glus0' is not online

[Ming]

I tested as below
1. Setup 2 nodes Proxmox and GlusterFS for HA , named P0, P1
2. unplug P0 power
3. Failover success, P1 started the VM
4. reinstall all necessary including glusterfs on new node called P2
5. make all setting like original, just with new IP and new node name P2
6. gluster peer probe, replace-brick all good, restart gluster service
7. pvecm added into cluster
8. pvecm delete the old node
9. all gluster, pvecm status are good, gluster volume status show online
10. tried live migration to new P2 OK, tried live migration back to P1 also OK,
11. PROBLEM here, unplug P1 power, and P2 trying to take care the VM
12 P2 trying to start the VM, but show Stroage 'glus0' is not online

Help......any one know the solution? thx

 

[dcsapak]

first: ha with 2 nodes is not possible/supported

second, you have probably entered server p0 and p1 in the glusterfs storage config, and when neither of them is online, the storage is also not online

 

[Ming]

i have some reason to make it works with 2 nodes, i set the corosync and it is able to failover at the first time with 1 node down

and i checked the gluster config, the bricks pointed with IP, and proxmox shows P0 and P2 on web console,
i wonder why proxmox can not load the gluster storage??
do the bricks need to use same TCP port?? i can not find any command to make it use same tcp port

 

[dcsapak]

can you post your /etc/pve/storage.cfg

 

[Ming]

Oh....thank you.....very good advice.......the IP is wrong in storage.cfg.......now the problem is solved....thankq