Ceph.conf is empty after editing it on a running Ceph cluster

[ChrisE]

We wanted to edit the /etc/pve/ceph.conf to decrease the I/O when remove/adding a SSD to the cluster. I found on the proxmox forum that we can use:

osd max backfills = 1
osd recovery max active = 1

To reduce the I/O when its rebuilding.
So I picked one node and directly edit the /etc/pve/ceph.conf

But after editing /etc/pve/ceph.conf we were unable to view the OSDs in the web interface. It only showed: rados_connect failed - No such file or directory (500). And all the monitors were gone in the tab Monitor.

Then I saw that the ceph.conf was completely empty on all the nodes, Fortunately ceph is still running.
I made a screenshot of the ceph.conf before I started, but can I just copy that into the empty ceph.conf or not?

So the question is how can I fix the problem with the ceph.conf?
And what is the best way to edit the ceph.conf on a running cluster?

Any and all help would be greatly appreciated.

 

[dietmar]

So the question is how can I fix the problem with the ceph.conf?

just copy the backup int this file.

And what is the best way to edit the ceph.conf on a running cluster?

Any reasonable editor works (vi, nano, emacs, ...).