Support for GFS with 100TB filesystem

[Werner Hoehn]

We are trying to offer RHEL AP together with GFS and support to a very important customer. The configuration includes 2 dl380 as head nodes connected to 4 msa2312fc with a total of 9 additional JBODs. Under 4+1 raid5 the useable filesystem space is 100TB.
The support for GFS filesystems is limited to 25TB by RedHat. For larger sizes one should 'ask the local support representative'.
There are larger GFS filesystems installed already, however, how do we get support from HP and RedHat for our config ?

Regards, thanks,
Werner

[Steven E. Protter]

Shalom,

You should try and segment your data a little bit and make smaller filesystems.

This is a management hassle one single file system so big.

If you must go this route, you need to create a software call with Red Hat. If HP is providing your Linux support, you can call the response center or use ITRC.

SEP

[Alzhy]

Werner.. did you manage to config this?

What version of RHEL did you use?

FYI: I am running RHEL 5.5, RHCS/GFS latest and I have a GFS2 Filesystem from an EVA array sized 500GB shared between 2 nodes. I use it as a KVM repository for my KVM Highly Available Virtualization environment

# >df -k /vmpool
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/vg_vmpool-vmpool
524227840 436174060 88053780 84% /vmpool


# >mount |grep gfs
none on /sys/kernel/config type configfs (rw)
/dev/mapper/vg_vmpool-vmpool on /vmpool type gfs2 (rw,noatime,nodiratime,hostdata=jid=0:id=196610:first=1)

[Werner Hoehn]

We did not receive support for GFS with that large filesystems. Not from HP and not from RedHat.
The customer selected a vendor with a lustre based solution.

[Alzhy]

That sucks.. maybe it was a limitation with older RHEL/AP and GFS? I have no problems at all wit RHEL 5.5 and GFS2, I even tried up to 1TB and still good.

But Lustre is nonethless good as well sir.

Cheers.