[Linux-cluster] GFS equivalent of ext3's block group size

Steven Whitehouse swhiteho at redhat.com
Thu Dec 11 11:17:39 UTC 2008


On Thu, 2008-12-11 at 11:10 +0000, Gordan Bobic wrote:
> I'm trying to apply some of the ext2/3 based logic to optimizing GFS on a RAID array. The idea is to try to make sure block groups all start on different disks, since beginning of each block group sees most access. If they are all on one disk, this disk becomes the bottleneck.
> So my question is, what is the equivalent in GFS? I'm guessing at resource group, but this only seems adjustable in 1MB increments, which is a bit large. Is there a way to adjust this with finer granularity?
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster

A resource group is almost the direct equivalent of a block group as you
suggest. I don't see any reason why we'd need to restrict them to 1M
intervals though, so I suspect that is entirely down to a decision
made by the author of mkfs and there is no reason why that couldn't
be relaxed in the future,


More information about the Linux-cluster mailing list