[Gluster-users] Gluster weridnes: File descriptor in bad state errors

[Nocturn]

Hi

I moved a replicated gluster cluster with two backend nodes into our staging
environment today.
Initial tests where Ok, but the client nodes began showing errors quickly.

On doing an ls of newly created directories, the nodes give File descriptor
in bad state errors
The parent directory shows up as:  drwxr-xr-x 1 root root 16 Jul 29 10:32 ..
?????????? ? ?    ?     ?            ? dev

Gluster verion is 3.1.1 from EPEL repos on Scientific Linux 6.0 (a RHEL6
clone).

This is the log file for that volume:
[2011-07-28 15:24:33.167689] I
[client-handshake.c:993:select_server_supported_programs] remote2: Using
Program GlusterFS-3.1.0, Num (1298437), Version (310)
[2011-07-28 15:24:33.168086] I [client-handshake.c:829:client_setvolume_cbk]
remote1: Connected to 10.9.144.11:24007, attached to remote volume 'brick'.
[2011-07-28 15:24:33.168139] I [afr-common.c:2571:afr_notify] replicate:
Subvolume 'remote1' came back up; going online.
[2011-07-28 15:24:33.171751] I [client-handshake.c:829:client_setvolume_cbk]
remote2: Connected to 10.9.144.12:24007, attached to remote volume 'brick'.
[2011-07-28 15:24:33.171775] I [fuse-bridge.c:2817:fuse_init]
glusterfs-fuse: FUSE inited with protocol versions: glusterfs 7.13 kernel
7.13
[2011-07-28 15:24:33.172460] I [afr-common.c:819:afr_fresh_lookup_cbk]
replicate: added root inode
[2011-07-29 15:51:52.716659] I
[afr-dir-read.c:171:afr_examine_dir_readdir_cbk] replicate:  entry self-heal
triggered. path: /data/tst/mysite/csa, reason: checksums of directory
differ, forced merge option set
[2011-07-29 15:51:52.718539] E [afr-common.c:110:afr_set_split_brain]
replicate: invalid argument: inode
[2011-07-29 15:51:52.718569] I
[afr-self-heal-common.c:1526:afr_self_heal_completion_cbk] replicate:
background  entry self-heal completed on /data/tst/mysite/csa
[2011-07-29 15:51:52.718682] W [fuse-bridge.c:2021:fuse_readdir_cbk]
glusterfs-fuse: 383: READDIR => -1 (File descriptor in bad state)

What is going on?

Guy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20110729/141809c8/attachment.html>