[SSI-devel] SSI-1.2.0-FC2 samba client IO error

[Roger T. <op...@bl...>]

This rarely happens but sometimes my smbfs mount on the initnode would
just break on its own.  I'd have to umount/mount the smbfs mount to fix
it.

The SSI cluster is a samba client samba-client-3.0.10-1.fc2.  The samba
WINS server samba-3.0.5-2 is outside the cluster.

Feb 10 05:34:56 node1 kernel: smb_get_length: recv error = 512
Feb 10 05:34:56 node1 kernel: smb_request: result -512, setting invalid
Feb 10 05:34:56 node1 kernel: smb_retry: signal failed, error=-1
Feb 10 05:34:56 node1 kernel: smb_writepage_sync: failed write,
wsize=4096, result=-5
Feb 10 05:34:56 node1 kernel: smb_lookup: find eightyterabytesample file
failed, error=-5
... message repeats

When I go into /mnt/smb the mountpoint on the cluster, I get an IO error.

Feb 10 05:34:56 server smbd[8681]: [2005/02/10 05:34:56, 0]
lib/util_sock.c:read_socket_data(367)
Feb 10 05:34:56 server smbd[8681]:   read_socket_data: recv failure for 4.
Error = Connection reset by peer
Feb 10 05:34:57 server samba(pam_unix)[8681]: session closed for user bbbb

Maybe this has to do with SSI as it seems to be happening on the samba
client on the cluster and not happening (not that I've noticed) on Windows
clients for example.  We'll have to see.

I was about to post to SSI-users.


-Roger