[Cluster-devel] Problems mounting GFS2/GFS devices and possible ipv6 dlm issue.

[Fabio Massimo Di Nitto <fabbione@ubuntu.com>]

Just a short update after fixing a bunhc of bugs

Fabio Massimo Di Nitto wrote:
> David Teigland wrote:
> 
> GFS1 lock_nolock
> .18-rc2: ok
> .17-5-server: ok
> 
> GFS1 lock_dlm
> 18-rc2: nok
> .17-5-server: nok
> 
> the mount operation seems to succeed. in dmesg:
> 
> [  148.044734] GFS: fsid=edgy:gfs1.4294967295: can't mount journal #4294967295
> [  148.044800] GFS: fsid=edgy:gfs1.4294967295: there are only 4 journals (0 - 3)
> 
> (note that there is no mount.gfs and it's not in CVS either. There is man page
> for it tho)

This is now ok after symlinking mount/umount.gfs to mount/jmount.gfs2 and
applying the patch i posted a few minutes ago to handle /mnt or /mnt/.

> 
> umount generate a OOPS.

not anymore.

It is still NOT possible to mount the shared device on two nodes when using IPv6
 address. It seems that the dlm kernel module isn't configured properly and we
get messages like this one in dmesg:

[  403.530413] Joined cluster. Now mounting FS...
[  403.537627] dlm: reject connect from unknown addr
[  403.538067] dlm: Error sending to node 2 -32
[  403.538119] dlm: Initiating association with node 2

also when trying to running other services, so it's not GFS/GFS2.

gfs2 status isn't changed.

Thanks
Fabio

-- 
I'm going to make him an offer he can't refuse.