[Linux-cluster] gfs1 vrs gfs2 best practices in CentOS 5.1 (RHEL5.1)
linux-cluster at merctech.com
linux-cluster at merctech.com
Wed Jan 30 17:26:16 UTC 2008
I've seen suggestions that GFS2 is not yet ready for production use...however,
in CentOS 5.1 (and, I assume, the upstream provider), GFS2 is the default:
lrwxrwxrwx 1 root root 10 Jan 29 17:44 /sbin/mount.gfs -> mount.gfs2
-rwxr-xr-x 1 root root 42000 Nov 12 14:04 /sbin/mount.gfs2
lrwxrwxrwx 1 root root 11 Jan 29 17:44 /sbin/umount.gfs -> umount.gfs2
-rwxr-xr-x 1 root root 40840 Nov 12 14:04 /sbin/umount.gfs2
This is on a CentOS 5.1 system, fully up-to-date with "yum".
In other words, installing gfs1 (via the gfs-utils package through yum) includes
gfs2 as a dependency, and the gfs2 installation replaces mount.gfs and
umount.gfs with links to the gfs2 version.
The umount.gfs2 binary is not backward compatible--I was not able to mount a
gfs1 filesystem with mount.gfs2, and there is no mount.gfs (version1) binary
installed.
My questions are:
Does this mean that GFS2 under CentOS5.1 (RHEL5.1) is now
production-ready?
If not, what's the recommended way of installing mount.gfs (version1)
under CentOS 5.1?
[Yes, I can certainly compile mount.gfs from source, or remove gfs-utils and
gfs2-utils and the force the installation of gfs-utils without it's
dependencies. However, for long-term system maintenance (and my sanity), I
strongly prefer not to administer servers with those kind of "exceptions".]
Thanks,
Mark
-----
Mark Bergman
http://wwwkeys.pgp.net:11371/pks/lookup?op=get&search=bergman%40merctech.com
More information about the Linux-cluster
mailing list