[Linux-cluster] GFS inodes

[David Teigland]

On Mon, Jul 23, 2007 at 01:12:22PM +0200, Steffan Henke wrote:
> gfs_tool lockdump returns an error message on all cluster nodes:
> gfs_tool: unknown mountpoint /gfs

Maybe gfs is on an odd device that gfs_tool can't match up with the
mountpoint.  Try this instead:

# gfs_tool list
18446675904036380672 sdb foo:x.0

# gfs_tool lockdump 18446675904036380672 > lockdump.txt

> >state the iopen locks are in for the unlinked files.  Also send the output
> >of 'gfs_tool gettune'.
> 
> gfs_tool gettune /gfs
> ilimit1 = 10
> ilimit1_tries = 3
> ilimit1_min = 10
> ilimit2 = 10
> ilimit2_tries = 10
> ilimit2_min = 3

I can't really say what effect these non-default values would have.  Could
you try the default ilimits?

> A gfs_tool reclaim returns *some* metadata, but no inode at all, while a
> service gfs restart shows

reclaim converts deallocated metadata blocks into normal free blocks.
Your problem is that inodes are never being deallocated in the first place
after being unlinked.

Dave