Dmytro,

There are two PRs we've been working on to improve Ceph snapshots. Firstly, a 
patch was merged last week that copies the image to secondary as a QCOW2 image, 
rather than raw. This obviously creates a much smaller image if the volume is 
sparse. Secondly, we've been working on enabling a feature that prevents 
copying to secondary from occurring at all. This one is a work in progress, but 
it looks like we've almost got most of the kinks worked out.


You can look at that PR here: https://github.com/apache/cloudstack/pull/1644


- Si

________________________________
From: Dmytro Shevchenko <[email protected]>
Sent: Thursday, September 29, 2016 9:47 AM
To: [email protected]
Subject: VM not responding on commands until snapshot running

Hi,

I noticed that while volume snapshot running, VM not executing other
commands as Reboot or Stop. They are waiting in queue until previous
command completed. Can anyone explain this behaviour, is this normal and
why?

I'm using Ceph as primary storage and creating snapshot takes several
seconds, but as next step CS copy this snap to secondary storage via
'qemu-img convert ...' and this operation is quite slow.

Tested on 4.5.1 and 4.8.0.1 versions. Ceph as primary storage and NFS as
secondary.


--
Best regards
Dmytro

Reply via email to