[Gluster-users] Adding a brick confusion

[james.bellinger at icecube.wisc.edu]

During the drain failure (described elsewhere, may not be relevant) one of
the other disk arrays unexpectedly died.  This system has no replication.

I stopped gluster on the host, rebuilt a new array (goodbye data) and
restarted gluster.  I then removed the brick.  I had to rename the mount
point to add it back (even shutting gluster down completely on all hosts
didn't help; and yes I was using the setfattr procedure), but OK, it is
back.  Empty, but that's life.

However, when I remount the filesystem on one of the clients, I get a mix
of success and failure, as shown below.

Is this because the old defunct brick's state was saved somehow?  (The
fact that I had to change the mount point is suggestive.)

Is there anything I can do about this?

thanks,
James Bellinger


# ls -l /data/uwa/jvansanten/sim
ls: cannot access /data/uwa/jvansanten/sim/2011: Invalid argument
total 265
drwxrwxr-x 4 jvansanten jvansanten 45056 Jul 19 11:51 2010
?????????? ? ?          ?              ?            ? 2011
drwxrwxr-x 4 jvansanten jvansanten 45056 Sep 15 03:10 dags
drwxrwxr-x 2 jvansanten jvansanten 45056 Sep 15 04:13 dags.cobaltgpu
drwxrwxr-x 2 jvansanten jvansanten 45056 Sep 15 05:06 dags.muongun
drwxrwxr-x 3 jvansanten jvansanten 45056 Jul 18 22:22 generated
drwxrwxr-x 5 jvansanten jvansanten 45056 Jul 19 11:51 logs
-rw-rw-r-- 1 jvansanten jvansanten   153 Jul 26 21:13 sweep.sh