Roland Fischer wrote:
Hi,
i think the algorithms doesnt work yet. I read the pdf ->
http://ftp.gluster.com/pub/gluster/glusterfs/3.0/LATEST/GlusterFS-3.0.0-Release-Notes.pdf
and this article is very interesting:
The “Diff” algorithm is especially beneficial for situations such as
running VM images,
where self-heal of a recovering replicated copy of the image will
occur much faster because
only the changed blocks need to be synchronized.
i use 12 GB (to 30GB) Xen Images and client side replication. If i
simulate an Hardwareerror on glusterfsserver 1 and after 5 minutes the
glusterfsserver 1 came back online, glusterfsserver2 starts sync but i
think not with "diff" algotihm ether with full!! i see the sync if i
do an iftop on both glusterfsserver.
while syncing you cant do nothing in domU. no "df -h", no "ls" nothing
- you can do nothing. if the iftop on both glusterfsserver has reached
e.g. 12GB the domU works further.
And thats the problem. in case of sync the domU must run further. This
is our destination.
Roland,
Thanks for all your input. We've been simulating the same situation
ourselves and we do
see that self-heal appears to be slow. We'll post an update on the bug
(bugzilla: 457) once we resolve the issue.
Vikas
