Hey Bernhard,
That was just enough to jog my memory :) Gluster 3.4 changed the port selection
they used. They went from ports 24007-24012 in gluster 3.3 to 49153 and up in
3.4 The problem is that libvirt and gluster aren't aware of each other and
and when qemu doesn't get its port it thinks the world is ending and gives
up. I found trying the migration a second time can help. You can also track the
full problem over at https://bugzilla.redhat.com/show_bug.cgi?id=987555 which
has both teams in there discussing how to handle the port stuff. If you
absolutely need live migration you can do it by hand using qemu directly and
pick your ports as to not collide until the port range problem is solved.
Also you should consider moving your guests images to using gfapi instead of the
gluster FUSE mount you're using if you need more performance. It requires a
couple hacks and a recompile on Ubuntu but we use it over here and couldn't
live without it. I'll be posting a writeup soon if you're interested.
Best,
Josh
----- Original Message -----
From: "Bernhard Glomm" <bernhard.glomm at ecologic.eu>
To: gluster at joshboon.com
Sent: Sunday, December 1, 2013 11:25:07 AM
Subject: Re: [Gluster-users] glusterfs 3.4
Hey Josh,
thnx for reply
well to be honest... nothing special at all.
I didn't set any special gluster options.
I created the mirror like before in 3.3
gluster volume create <vol_name> replica 2 <ip>:/<path>
<ip>:/<path>
EXACT the same command both times (thnx copyndpaste ;-)
I mounted the volume exactly the same on both machines
mount -t glusterfs <my_own_ip>:/<vol-name> <mountpoint>
(asuming this is fuse mount?)
I (tried to) migrate/ed the VM with the same command
migrate --verbose --live --unsafe --p2p --domain <dom_name> --desturi
qemu+ssh://<ip>/system
with version 3.3 no problem
with version 3.4.1 I got
"-incoming tcp:0.0.0.0:49153: Failed to bind socket: Address already in
use"
any hints on this?
I can do with 3.3 for now, works like a charm
but would be happy to feel more comfortable about possible upgrades ;-)
best
Bernhard
Am 30.11.2013 23:27:14, schrieb Josh Boon:
Hey Bernhard,
I'm running a very similar setup to you. What are your gluster options?
What's the migration command you're using? Are you using gfapi or a fuse
mount? I think I've hit this error before but I'll need some info to jog
brain on how I fixed it.
Best,
Josh
From: "Bernhard Glomm" <bernhard.glomm at ecologic.eu>
To: gluster-users at gluster.org
Sent: Saturday, November 30, 2013 8:13:26 AM
Subject: [Gluster-users] glusterfs 3.4
Hi all, I just stumbled over a possible bug in glusterfs-3.4.1 Since I used the
package from ppa:semiosis/ubuntu-glusterfs-3.4 for ubuntu (running 13.04 up to
date) I like to report here first / ask for confirmation. I had glusterfs 3.3
installed and it worked fine. I used it as the storage backend for the image
files of my kvm virtualized instances (two sided mirror/two kvm hosts running
four VM as a test environment) I upgraded to 3.4 and all seemed okay on the
first glance, but live-migration failed with error: "-incoming
tcp:0.0.0.0:49153: Failed to bind socket: Address already in use" reverting
back to glusterfs 3.3 and the problem was gone anybody knows about that prob?
couldn't find anything on the net about it yet. best regards Bernhard --
sysadmin www.ecologic.eu
_______________________________________________
Gluster-users mailing list
Gluster-users at gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users
--
Bernhard Glomm
IT Administration
Phone: +49 (30) 86880 134
Fax: +49 (30) 86880 100
Skype: bernhard.glomm.ecologic
Ecologic Institut gemeinn?tzige GmbH | Pfalzburger Str. 43/44 | 10717 Berlin |
Germany
GF: R. Andreas Kraemer | AG: Charlottenburg HRB 57947 | USt/VAT-IdNr.:
DE811963464
Ecologic? is a Trade Mark (TM) of Ecologic Institut gemeinn?tzige GmbH
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://supercolony.gluster.org/pipermail/gluster-users/attachments/20131201/c849c562/attachment.html>