Re: Cluster NFS causes kernel bug

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, 2007-09-13 at 17:34 -0400, Ward, Timothy - SSD wrote:
From: Kevin Anderson <kanderso@xxxxxxxxxx>

On Thu, 2007-09-13 at 10:32 -0400, Ward, Timothy - SSD wrote:
> I am trying to setup nfs as a clustered service.  The filesystem nfs 
> is serving is a GFS2 fs.  So I am trying to find out is the problem 
> with GFS2, NFS, the SMP processors or something else?
> 
We have fixed a number of bugs with GFS2 over the last few weeks, so
would point at GFS2 as being the issue.  Pick up the latest kernel from
Steve Whitehouses GIT tree.  It has a number of fixes, with a number of
additional ones queued up to be pushed shortly.

If you need a stable environment at this time, use GFS1 instead of gfs2.

Kevin


------------------------------
Kevin,

Thank you for the valuable insight!

I used the updated gfs2 RPM that was updated for FC6.  Would you suggest
using FC5 with  gfs1 or is there a gfs1 RPM someplace for FC6?  I tried
to compile the code myself with limited success.

We are working on getting gfs1 to work with FC7.  It is a bit complicated due to gfs1 needing access to the dlm harness code that is part of the gfs2 kernel module.  The symbols are not currently exported in the gfs2 code, so a patch needs to be generated.  The patch isn't something we believe would be accepted upstream since the symbols aren't used by any other in kernel module.   Here is the exports you need:
--- linux-2.6.17.noarch/fs/gfs2/locking.c~      2006-08-10 13:33:09.000000000 -0
400
+++ linux-2.6.17.noarch/fs/gfs2/locking.c       2006-08-10 13:33:23.000000000 -0
400
@@ -188,4 +188,6 @@ void __init gfs2_init_lmh(void)

EXPORT_SYMBOL_GPL(gfs2_register_lockproto);
EXPORT_SYMBOL_GPL(gfs2_unregister_lockproto);
-
+EXPORT_SYMBOL_GPL(gfs2_withdraw_lockproto);
+EXPORT_SYMBOL_GPL(gfs2_mount_lockproto);
+EXPORT_SYMBOL_GPL(gfs2_unmount_lockproto);


This is from our RHEL5 tree, so not sure it will apply directly, but you can see what gets added.  Am sure there are probably other issues with building against the FC6 environment, but this is a start. 

We have been primarily focused on getting GFS2 stable, but one of the guys is going to spend sometime over the next couple of weeks working on getting FC7 rpms built for gfs.

Kevin
--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster

[Index of Archives]     [Corosync Cluster Engine]     [GFS]     [Linux Virtualization]     [Centos Virtualization]     [Centos]     [Linux RAID]     [Fedora Users]     [Fedora SELinux]     [Big List of Linux Books]     [Yosemite Camping]

  Powered by Linux