[Ocfs2-users] OCFS2 and Apache Problem

[Mark Fasheh]

On Tue, Nov 13, 2007 at 05:43:03PM -0800, Michael Moody wrote:
> This occurs:
>
> pba1 / # echo 'locate <6733855010560701275>' | debugfs.ocfs2 -n /dev/sdd5
> locate: Invalid block number - 6733855010560701275

Ok, I messed up - that log prints the block number as read off disk, which
is clearly wrong. Sorry.


> Also, if I unmount, and remount the nfs, should that fix it, or restart nfs 
> services on the nodes themselves?
>
> Any idea why I see this on both of the nodes, where only the first node is 
> exporting the filesystem via nfs?

Hmm, if it's happening on the non-exported node then it's it might not be
related to nfs.

How big is this particular file system, in terms of # of inodes. If it's not
a high number of files, you could just poke around from the shell and see if
'ls -l' within a directory reliably gives you that message.

Also, you could run fsck.ocfs2 in offline, *readonly* mode and send a log...
The problem with this is that it might give us tons of false positives :(

fsck.ocfs2 -fn /dev/sdd5

Be really really careful with that. Double check my command line via the man
page to make sure that fsck won't try to modify your file system.
	--Mark

--
Mark Fasheh
Senior Software Developer, Oracle
mark.fasheh at oracle.com