[linux-cifs-client] Infrequent CIFS hang/slow response when connecting to Windows Storage Server 2003 shares

[Adam Tworkowski]

I am seeing occasional hangs/very slow response while accessing shares on Windows Storage Server 2003.  There are two of these devices.  Shares are mounted via CIFS in /etc/fstab within two different branches (say /net/nas1, /net/nas2).  The client system is SLES9 with current patches.

The symptom is that on occasion an "ls" issued within one of root directories (/net/nas2) where six shares are mounted "hangs" for some time.  The logs show that each share is "evaluated" in serial until all are done at which point the "ls" ceases to hang and the root of each share is listed.

So far, we have only conclusively seen the issue on one Linux server and not another (only two are presently in the position to the see the issue.

Both servers are set up identically (or so I can tell) in terms of CIFS.  The only main difference between the servers is that the affected server was moved from on network to another, while unaffected one has always had the same IP.

I see the following when I do the ls and it stop here at the hang.

Nov 21 10:45:27 <myserver> kernel:  fs/cifs/inode.c: CIFS VFS: in cifs_revalidate as Xid: 14216 with uid: 0
Nov 21 10:45:27 <myserver> kernel:  fs/cifs/inode.c: Revalidate:  inode 0xef6639d4 count 1 dentry: 0xf20d1670 d_time 0 jiffies 930883372
Nov 21 10:45:27 <myserver> kernel:  fs/cifs/inode.c: Getting info on

Once the hang for the specific share is finished the logs continue:

\.DS_Store
Nov 21 10:45:59 stage11 kernel:  fs/cifs/transport.c: For smb_command 50
Nov 21 10:45:59 stage11 kernel:  fs/cifs/transport.c: Sending smb of length 94
Nov 21 10:45:59 stage11 kernel: | 0x00 0x00 0x00 0x5e 0xff 0x53 0x4d 0x42  |  _ _ _ ^ \377 S M B
Nov 21 10:45:59 stage11 kernel: | 0x32 0x00 0x00 0x00 0x00 0x00 0x01 0xc0  |  2 _ _ _ _ _ _
Nov 21 10:45:59 stage11 kernel: | 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00  |  _ _ _ _ _ _ _ _
[snip]

I believe that evaluation of each share hangs at "Getting info on" and then continues until the process is complete.

I can provide more logs if necessary.  If anyone can point at what stage in the CIFS dialogue this is occurring, that would be useful.  The domain control is smb-ldap based.

Regards,

Adam Tworkowski, atworkowski at masterfile.com
Systems Administrator, Computer Department
Masterfile Corporation, www.masterfile.com
416-929-6745 x.2287

Regards,

Adam Tworkowski, atworkowski at masterfile.com