OSD mounting to tmpfs : Is that expected?

[RobFantini]

Hello

after upgrading ceph to ceph-nautilus in pve6 , osd's created mount to tmpfs .

Is that a bug?

If so is there anything i could try to fix the issue?

PS : for detail please see https://forum.proxmox.com/threads/ceph-nautilus-move-osd-to-another-node.56862/

 

[dcsapak]

after upgrading ceph to ceph-nautilus in pve6 , osd's created mount to tmpfs .

yes this is expected as with ceph-volume the metadata is not in a partition anymore but in the lvm metadata (see lvs -o +lv_tags) from which
ceph builds a temporary filesystem

 

[RobFantini]

can you check my other thread?

during the process of moving osd's away from a failing motherboard - i had to recreate the osds.

1- the pool has a mix of new and old type osd's.

2- pve shows that the pool lost approx 2T worth of storage capacity. see the attached pic.

I am concerned that the pool is unstable.

Please look at the pic. It shows we have lost 2T as a result of moving osd's from one node to others.