[Linux-cluster] rhel5 gfs1 or gfs2, how to prove that i have gfs1

Kevin Anderson kanderso at redhat.com
Mon Aug 18 14:38:54 UTC 2008

On Mon, 2008-08-18 at 19:55 +0530, Anuj Singh (अनुज) wrote:
> Hello everyone, 
> I created a cluster on rhel5,
> To make the gfs1 file system I used following command.
> ' gfs_mkfs -t new_cluster:GFS -p lock_dlm -j2 /dev/MyVol0/MyLV0'
> Mounted /dev/MyVol0/MyLV0  on  /image.
> 'mount -t gfs /dev/MyVol0/MyLV0 /image '
> No error I got.
> Mount command gives me. 
> /dev/mapper/MyVol0-MyLV0 on /image type gfs
> (rw,hostdata=jid=0:id=131073:first=1)

Looks good to me.

> My questions:
> 1) How to prove that I have gfs1 file system in use. 
> 2) Ismod |grep gfs gives me.
> [root at pr0031 CL]# lsmod |grep gfs
> gfs                   252740  1 
> gfs2                  341965  2 gfs,lock_dlm
> configfs               28753  2 dlm
> 3) rmmod gfs2
> ERROR: Module gfs2 is in use by gfs,lock_dlm
> ^^ why gfs2 module is in use? 
> as lock_dlm is associated with gfs2, does it mean it's not production
> ready?

There lock modules are including in the upstream kernel as part of the
gfs2 tree.  There are three symbols that gfs2 and gfs share, with gfs2
being the owner of the symbols that provides access to the lock modules.
This requires gfs2 module to be loaded in order for gfs to access the
lock modules.  That is the only part of gfs2 being leveraged for gfs.
We are working to remove that dependency so as not to confuse people.
Net result is that if you use mkfs -t gfs or gfs_mkfs  - you get a gfs

So, you are good to go with a supported gfs configuration on RHEL5.


More information about the Linux-cluster mailing list