Map rbd image on multiple nodes...

[Vassilis Kipouros]

Good day,

I have a small three node proxmox cluster with ceph.
Running CTs that all need access to a folder inside an rbd image.

Currently I use a fileserver VMfileserver,
that mounts the rbd image as second hd, exports nfs shares,
and all nodes mount the shared directory under /mnt/storage.

The CTs use mp: to access it. And since the folder is available to all nodes,
I can migrate my containers to any node.

I was thinking that maybe I could get rid of the fileserver VM and map rbd localy using:

# rbd -p [poolname] map [imagename]
# rbd showmapped
id pool image snap device
0 [poolname] [imagename] – /dev/rbd0
# mount /dev/rbd0 /mnt/storage

Since I want to migrate my containers between nodes, my question is:

- Can the rbd image be mapped and mounted simultaneously on all nodes?
- And if so, what would be the way to go about making the map&mount
of the rbd image automatic after reboot?

Any ideas/suggestions welcome.

Regards

 

[dietmar]

- Can the rbd image be mapped and mounted simultaneously on all nodes?

This is a quite bad idea if you use normal file systems like ext4 or xfs and will result in data loss.

 

[Vassilis Kipouros]

I use btrfs or ext4...

So what do you propose I do?
Keep the fileserver vm?
Is there another way to give access on the image filesystem on multiple CTs?

 

[grin]

Use CephFS.