[Linux-cluster] stability gfs2

[David Teigland]

On Tue, Jun 12, 2007 at 11:07:27AM +0200, Frederik Wagner wrote:
> Hi group,
> 
> I was wondering about the stability of the gfs2 filesystem.
> After the release of Scientific Linux 5 (SL5) I tried the gfs2
> filesystem on our FC SAN Storage with a two node cluster frontend (this
> are supposed to be three in the future).
> 
> The system was running well on SL44 and is now running well with SL5 and
> gfs, but when I tried the gfs2 filesystem I had problems syncing (rsync)
> back the storage data from the backup. I got reproducable write errors,
> after which the whole filsystem was stuck and broken.
> 
> Unluckily I had to install the system in to production with gfs now, so
> I cannot post the error messages. (if the error messages are needed for
> further debugging, i could set up a testsystem) So up to now I'm wondering:
> 1. is gfs2 supposed to be completely stable?
> 2. are there any known serious bugs i missed?
> 3. why should i use gfs2 instead of gfs, except for a faster df command? :-)
> 4. are there any documentation speaking about the gfs2 fs system in
> detail? To me the documentation looks quite 'small'...

gfs2 isn't ready to be used, stick with gfs1.

Dave