Quick ceph question - unable to parse addrs in ceph.conf

I’m running Ceph as backend storage for my OpenStack cluster. I thought I’d try it out on LXD, so I create an RBD pool called lxd, ran RBD init on the pool then tried the command

lxc storage create ceph1 ceph source=lxd

but unfortunately get the following back:

Now this ceph.conf works fine for OpenStack so not sure whats wrong with its format?

host = p67
keyring = /var/lib/ceph/radosgw/ceph-rgw.p67.rgw0/keyring
log file = /var/log/ceph/ceph-rgw-p67.rgw0.log
rgw frontends = beast endpoint=
rgw thread pool size = 512

host = p70
keyring = /var/lib/ceph/radosgw/ceph-rgw.p70.rgw0/keyring
log file = /var/log/ceph/ceph-rgw-p70.rgw0.log
rgw frontends = beast endpoint=
rgw thread pool size = 512

# Please do not change this file directly since it is managed by Ansible and will be overwritten
cluster network =
fsid = 9bd31b02-ee3c-4dc7-8334-c12bc9596ca0
mon host = [v2:,v1:],[v2:,v1:]
mon initial members = p67,p70
osd pool default crush rule = -1
public network =

osd memory target = 23535812608

What version of the snap is that?

I thought for a second it may have been the old version and I hadn’t migrated!
But it looks like it is the latest snap v ersion 4.0.0

Ok, what version of ceph do you normally run with this config?

It’s running Ceph Octopus which is I think the latest stable version of ceph.
This is the first time I’ve ever tested ceph with LXD

The LXD snap ships its own version of the ceph client at version 12.2.13.
It’s usually not an issue talking to more recent servers but in this case there may be something that this version of ceph doesn’t like.

Note that you could feed it a much smaller ceph config just for LXD by doing:

  • snap set lxd ceph.builtin=true
  • mkdir /var/snap/lxd/common/ceph
  • systemctl reload snap.lxd.daemon

At which point /var/snap/lxd/common/ceph will be used rather than /etc/ceph

ok I’ll give that a try thanks!