Ceph overwritten - some help please?

Dragon19

Active Member
Jan 4, 2020
49
9
28
34
While removing a node from cluster and adding a new one I stupidly created a new ceph which somehow affected all nodes.

rbd and ceph.conf were wiped out in /etc/ceph. I have the FSID. There are encrypted OSD (OSD are intact).

/var/lib/ceph is intact but a lot of it was overwritten.

I have the original admin key. I have the original keyring for monitors and the kv_backend. I have the mds original keyring. The OSD lockbox keys seem to be gone or overwritten. I have the original radosgw keyring.

Is there anyway to recover from this? I have backups but not of recent data. If not, no big loss just a huge waste of my time to manually add the data again.

It will not let me create new monitors and when running ceph -s I get the following:

unable to get monitor info from DNS SRV with service name: ceph-mon
[errno 2] error connecting to the cluster

It's likely because the newly generated ceph wasn't finished before I hard restarted the node to prevent further data loss. I did try adding the monitor address to the new ceph.conf though.

Code:
global]
         auth client required = cephx
         auth cluster required = cephx
         auth service required = cephx
         cluster network = 172.16.3.1/24
         fsid = lolnothere
         mon allow pool delete = true
         osd pool default min size = 1
         osd pool default size = 1
         public network = 172.16.2.1/24

[client]
         keyring = /etc/pve/priv/$cluster.$name.keyring

[mon]
        host = cl-01
        mon addr = 172.16.2.1:6789
[ceph-mon]
        host = cl-01
        mon addr = 172.16.2.1:6789
 
The config needs to have all the MONs listed, so they are found by the ceph daemons. Besides that, the rest of the config should have been stored on the MONs and therefore still be available.