[Gluster-users] Problem with gluster 3.1

[Amar Tumballi]

Hi Jan,

This means, the 'trusted.gfid' xattr (which is very much required by the
glusterfs to work fine and is sort of inode number of a file in regular fs)
is missing for that particular folder. One way to fix it is by below steps

on other nodes where you don't have any logs:

sh# getfattr -d -m trusted.gfid <some folder>
<output>

on the node which has issue of logs:

sh# setfattr -n trusted.gfid -v <value got in other node> <some folder>

This should solve the issue.

This state is mostly reached by not having this one machine ready before
having a client machine mounting the volume and accessing these folders.

Regards,
Amar

On Sat, May 21, 2011 at 4:23 AM, Jan Wynholds <jwynholds at 510systems.com>wrote:

> Hello:
>
> We have a 8 node gluster setup here.  Some of the nodes were running
> 3.0.7 and 3.0.8.  One of the nodes has an xfs filesystem (which is the
> only difference between the rest which are ext4).
>
> On the xfs (that was also previously running 3.0.7 and 8), grows logs
> at an alarming rate (gigs a day).
>
> The logs are filled with errors such as:
>
> [2011-05-20 15:10:46.630875] E [posix.c:438:posix_lookup]
> 0-store1-posix: lstat on [some folder] failed: No data available
>
> After looking through the documentation, and traffic on the mailing
> lists, I have come up empty on an explanation.
>
> Any insight would be most appreciated.
>
> --cheers
>
> jan
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20110524/2d1ec421/attachment.html>