[Gluster-devel] glusterfsd crash, fd_ref() after fd_destroy().

[Emmanuel Dreyfus]

Emmanuel Dreyfus <manu at netbsd.org> wrote:

> As I understand this is caused by an fd being used after fd_destroy().
> It may be related to that change I submited in git, since I run with it:
> Change-Id: I74c87bd9a53c0deeef79b93b4a0066e751b17dca
> BUG: 2923

More on this: My logs are full of messages about faulty gfid. My patch
makes them non fatal, but I suspect they can still be harmful in some
way:

[2011-08-22 06:26:00.103796] I
[dht-common.c:838:dht_lookup_everywhere_cbk] 0-gfs-dht: deleting stale
linkfile /manu/netbsd/usr/src/lib/libcompat/obj/stty.cat3 on
gfs-replicate-0[2011-08-22 06:26:00.104894] W
[dht-common.c:781:dht_lookup_everywhere_cbk] 0-gfs-dht:
/manu/netbsd/usr/src/lib/libcompat/obj/stty.cat3: gfid differs on
subvolume gfs-replicate-1

While it seems to work, I very rarely have odd failures that I cannot
reproduce, and I suspect it maybe caused by this error. Any hint on how
to investigate? How can gfid mismatch so often? 

I also have a huge amount of this, which seems to be have no real
consequence. Opinion?

[2011-08-22 06:26:00.583922] E [fuse-bridge.c:2840:fuse_setlk_resume]
0-glusterfs-fuse: xl is NULL

-- 
Emmanuel Dreyfus
http://hcpnet.free.fr/pubz
manu at netbsd.org