[Linux-cluster] Block Size

Jeremy Eder jeder at invision.net
Thu Feb 19 20:42:05 UTC 2009

I think he's talking about the partition offset rather than filesystem block size...check this:

I agree, using gfs2 in a single node is just fine - when you create the partition, use lock_nolock.

Best Regards,

Jeremy Eder, RHCE, VCP

-----Original Message-----
From: linux-cluster-bounces at redhat.com [mailto:linux-cluster-bounces at redhat.com] On Behalf Of Stephen Nelson-Smith
Sent: Thursday, February 19, 2009 15:35 PM
To: linux clustering
Subject: [Linux-cluster] Block Size


I have a 1TB LUN on an EMC SAN, which I see in /proc/partitions as emcpowera.

Ultimately I will be sharing this between two, or three machines, but
at present only one is fibre attached.

I propose to create a GFS2 filesystem now, and use it with one machine
until it becomes necessary to share the LUN with the other machines.

The backups/storage engineer from my hostco recommends I use a
blocksize of 128kb to ensure there are no missed alignments on the SAN
that could degrade performance.  128kb sounds *huge* given that the
default is 4k.

Two questions:

Any problems with the recommended approach - ie mkfs.gfs2 first, worry
about clustering later?

Is mkfs.gfs2 -b 131072 really sane?



Stephen Nelson-Smith
Technical Director
Atalanta Systems Ltd

Linux-cluster mailing list
Linux-cluster at redhat.com

More information about the Linux-cluster mailing list