VE on GFS2 [message #30965] |
Wed, 11 June 2008 11:03 |
scixpp
Messages: 4 Registered: June 2008
|
Junior Member |
|
|
Hi,
we have two nodes with a clustered disk on san, with GFS2. The OS is RedHat 5 and we use the kernel 2.6.18-53.1.19.el5.028stab053.14ent.
When we start a VE, created on this GFS2 filesystem, the init boot process eats 100% of CPU and no more process are started on the VE. If we try to go in with vzctl enter command, give the error:
"Unable to open pty: Invalid argument"
Seems the same problem reported time ago in this thread
http://forum.openvz.org/index.php?t=msg&goto=12876&& amp;srch=gfs2#msg_12876
But, if we try to create the same VE on a ext3 local filesystem, the VE goes perfect, without any problem.
There are some special configuration or drawback to run VE on GFS2 filesystems? any solution?
Thanks
|
|
|
Re: VE on GFS2 [message #31019 is a reply to message #30965] |
Fri, 13 June 2008 09:35 |
jonasb
Messages: 20 Registered: July 2007
|
Junior Member |
|
|
I have the same problem. You dont need a SAN to test it.
Just format a partition with GFS2 and create a VE on it.
However, I can mount a GFS2 volume somewhere inside the VE without problems.
[Updated on: Fri, 13 June 2008 09:36] Report message to a moderator
|
|
|
Re: VE on GFS2 [message #31055 is a reply to message #31019] |
Mon, 16 June 2008 12:21 |
scixpp
Messages: 4 Registered: June 2008
|
Junior Member |
|
|
Hi,
I have found a workaround for this problem until a clean solution come.
If I mount the /dev directory on a ext3 filesystem, all problems gone!. Seems it's something related to the devices access on a GFS2 partition during the VE startup.
You could create a vps.mount in /etc/vz/conf with p.e.
#!/bin/bash
. /etc/vz/vz.conf
. $VE_CONFFILE
set -x
mount -n --bind /vzdev/$VEID/dev $VE_ROOT/dev
and copy a /dev directory from the original VE on gfs2 to /vzdev/$VE_ROOT on a ext3 filesystem.
Xavier
[Updated on: Mon, 16 June 2008 12:24] Report message to a moderator
|
|
|