[Linux-cluster] gfs2_tool unfreeze hang

[Steven Whitehouse]

Hi,


On Wed, 2012-10-31 at 14:07 -0500, james pedia wrote:
> Noticed this thread for the same issue at:
> 
> 
> https://www.redhat.com/archives/linux-cluster/2012-September/msg00084.html:
> 
> 
> I think I hit the same issue:
> 
> 
> (CentOS6.3)
> # uname -r
> 2.6.32-279.el6.x86_64
> 
> 
> gfs2-utils-3.0.12.1-32.el6_3.1.x86_64 is in use here.
> 
> 
> 
> 
> # gfs2_tool freeze /var/www/html
> # ls -l /var/www/html/
> total 8
> -rw-r--r-- 1 root root 10 Oct 30 23:47 a
> -rw-r--r-- 1 root root 41 Oct 30 20:44 index.html
> # cp /var/www/html/a /var/www/html/b
> (HANG HERE)
> 
> 
> Then try this:
> # gfs2_tool unfreeze /var/www/html
> (HANG AS WELL)
> 
> 
> The whole cluster has to be reset to recover from this. 
> 
> 
> 'dmsetup suspend' and 'dmsetup resume' are working fine. 
> 
> 
> Are these commands basically doing the same thing ('dmsetup suspend'
> vs 'gfs2_tool freeze')? 
> 
> 
> Is there a way to see if GFS2 file system is currently being suspended
> or frozen?
> 
Yes they do the same thing. I'd always recommend dmsetup suspend over
the gfs2_tool method though, since the latter is going away in due
course. There is, unfortunately, no way to check the suspend status of a
GFS2 filesystem currently,

Steve.

> 
> Thanks,
> -- 
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster