[Gluster-devel] [Errno 77] File descriptor in bad state

[Антон Халиков]

Hi guys

I've got the following python error from rdiff-backup tool:

IOError: [Errno 77] File descriptor in bad state
close failed: [Errno 77] File descriptor in bad state

Glusterfs log contains:

2008-07-17 15:23:14 C [rr.c:216:rr_schedule] rr: free space not
available on any server
2008-07-17 15:23:34 E [client-protocol.c:1158:client_writev] bbrick2:
: returning EBADFD
2008-07-17 15:23:34 E [fuse-bridge.c:1645:fuse_writev_cbk]
glusterfs-fuse: 726354: WRITE => -1 (77)
2008-07-17 15:23:34 E [client-protocol.c:1158:client_writev] bbrick1:
: returning EBADFD
2008-07-17 15:23:34 E [fuse-bridge.c:1645:fuse_writev_cbk]
glusterfs-fuse: 726356: WRITE => -1 (77)
2008-07-17 15:23:34 E [client-protocol.c:1238:client_flush] bbrick1: :
returning EBADFD
2008-07-17 15:23:34 W [client-protocol.c:1296:client_close] bbrick1:
no valid fd found, returning
2008-07-17 15:23:34 W [client-protocol.c:1296:client_close] bbrick-ns:
no valid fd found, returning
2008-07-17 15:23:34 E [client-protocol.c:1238:client_flush] bbrick2: :
returning EBADFD
2008-07-17 15:23:34 W [client-protocol.c:1296:client_close] bbrick2:
no valid fd found, returning
2008-07-17 15:23:34 W [client-protocol.c:1296:client_close] bbrick-ns:
no valid fd found, returning
2008-07-17 15:23:34 E [client-protocol.c:1238:client_flush] bbrick2: :
returning EBADFD
2008-07-17 15:23:34 W [client-protocol.c:1296:client_close] bbrick2:
no valid fd found, returning
2008-07-17 15:23:34 W [client-protocol.c:1296:client_close] bbrick-ns:
no valid fd found, returning
2008-07-17 15:23:34 E [client-protocol.c:1238:client_flush] bbrick2: :
returning EBADFD
2008-07-17 15:23:34 W [client-protocol.c:1296:client_close] bbrick2:
no valid fd found, returning
2008-07-17 15:23:34 W [client-protocol.c:1296:client_close] bbrick-ns:
no valid fd found, returning
2008-07-17 15:23:34 E [client-protocol.c:1238:client_flush] bbrick1: :
returning EBADFD
2008-07-17 15:23:35 E [client-protocol.c:1238:client_flush] bbrick1: :
returning EBADFD
2008-07-17 15:23:35 E [client-protocol.c:1238:client_flush] bbrick1: :
returning EBADFD

I also not sure what means the first record in my log because there is
a lot of space left.

My setup is quite simple: 2 bricks with unify where unify is not on
afr but on brick1 only instead so I have something like distributed
RAID-0 over glusterfs.

glusterfs 1.3.9tla787 built on Jun 30 2008 14:43:53
Repository revision: glusterfs--mainline--2.5--patch-787

-- 
Best regards,
Anton Khalikov