I got a handy reply off list[1], which confirms a bit the direction I
was looking for. I am just wondering if there are some handy ceph
specific things that can be used to optimize such a process. Eg I am
using rbd images with minimum features[2] because I am still using the
3.10 kernel, I have no idea how object-map or journaling would be able
to assist, if at all. Now I have to issue 8 commands to access data on a
snapshot. Could be nice if I could mount a snapshot with just one. Same
would be for archiving and restoring such sparse images. I have
configured most vm's with discard='unmap' to keep the images small in
ceph.
[1]
snapshot, export | yourfavoritecompressor | remote system?
[2]
[@c01 ~]# rbd info rbd/xxxxx
rbd image xxxxx':
size 50 GiB in 12800 objects
order 22 (4 MiB objects)
snapshot_count: 7
id: 3621b16b8b4567
block_name_prefix: rbd_data.3621b16b8b4567
format: 2
features: layering
op_features:
flags:
create_timestamp: Mon Apr 15 22:24:05 2019
access_timestamp: Mon Nov 30 16:23:39 2020
modify_timestamp: Mon Nov 30 16:25:45 2020
-----Original Message-----
From: Janne Johansson [mailto:icepic.dz@xxxxxxxxx]
Sent: Monday, November 30, 2020 12:43 PM
To: Marc Roos
Cc: ceph-users
Subject: *****SPAM***** Re: rbd image backup best practice
Den fre 27 nov. 2020 kl 23:21 skrev Marc Roos
<M.Roos@xxxxxxxxxxxxxxxxx>:
Is there a best practice or guide for backuping rbd images?
One would think that most things that apply to an iscsi mounted device
would be equally valid for an RBD mount, so you might look into that for
hints and tips on how to backup remote network data devices if you are
seeing this from the mounting client point of view.
If now, it probably matters what you are aiming for since "backup" is
quite a wide concept apart from "copy of my data".
Is it best practice for storing sparse images for conserving backup
space, quick restores, valid images, legal archiving demands, or "want
to try this weird update and be able to move backwards 30 minutes" ?
The solution will be quite different depending on what the problem is,
more than perhaps "what the mountpoint type is".
--
May the most significant bit of your life be positive.
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
