[Linux-cluster] GFS vs Ext3/4

jeremymiller at ups.com jeremymiller at ups.com
Mon Nov 22 21:16:20 UTC 2010

> Forgive me, but I am not entirely sure how switching the
> underlying FS alone will reduce the overhead in cluster.conf.
> Can you explain that bit? Perhaps a snippet from your
> cluster.conf would help.
It reduces the overhead for rgmanager - less resources for the daemon to manage.  Our cluster.conf has 2,303 lines - each of the 80+ database services typically has 3-5 filesystems, for which you have associated fs devices, lv_names and ip addresses.  Not to mention, 2000+ lines breaks Luci/conga so we edit the file manually - a rather tedious and error-prone process.  We've seen the management of the resources in this file become an issue for rgmanager having to scan each service, check status and stop/start all of them.  By using GFS2 the cluster.conf file is _greatly_ simplified since all the lv's and filesystem data is moved out to fstab - and rgmanager has much less to do.

> In general though, you should be able to mount a GFS2
> partition on just one node without trouble. Personally, I
> like to use shared storage and then put an LVM on it, enable
> Cluster LVM and then create a VG->LV which I then put the
> GFS2 partition on. For two node clusters, a DRBD device would
> suit just fine and would incur no extra hardware costs.
> So short answer; It should be fine to just use GFS2.

Came across this document today:


Quote:  "Although a GFS2 file system can be implemented in a standalone system or as part of a cluster configuration, for the Red Hat Enterprise Linux 6 release Red Hat does not support the use of GFS2 as a single-node file system. Red Hat does support a number of high-performance single node file systems which are optimized for single node and thus have generally lower overhead than a cluster file system. Red Hat recommends using these file systems in preference to GFS2 in cases where only a single node needs to mount the file system."

I guess that pretty much answers my question, unless others on the list would seem to think this not a problematic issue and/or vouch for continuing the use of GFS2 in spite of the application only residing on a single node of the cluster.

More information about the Linux-cluster mailing list