[cifs-protocol] Connection turn down problem in W2k8R2?

Obaid Farooqi obaidf at microsoft.com
Thu Jan 5 15:05:04 MST 2012


Hi Volker:
Thank you for contacting Microsoft regarding your inquiry on W2k8R2 product behavior. I have informed product group about the issue and we'll investigate as appropriate. If we need help to reproduce this issue, I or someone else will be in touch.

Regards,
Obaid Farooqi
Escalation Engineer | Microsoft

Exceeding your expectations is my highest priority.  If you would like to provide feedback on your case you may contact my manager at nkang at Microsoft dot com


-----Original Message-----
From: Volker Lendecke [mailto:Volker.Lendecke at SerNet.DE] 
Sent: Thursday, January 05, 2012 10:22 AM
To: dochelp at winse.microsoft.com
Cc: pfif at tridgell.net; cifs-protocol at samba.org; Jim Pinkerton
Subject: Connection turn down problem in W2k8R2?

Hi!

Right now I'm trying to test and improve Samba's cleanup routines. Testing those against W2k8R2 revealed that it might have some slight problem here as well.

Not sure this is the right forum, but this is my most direct channel :-)

I've added some code to Samba to artificially simulate a crashing connection helper process, triggered by a magic packet. In the attached trace, it is frame 43. It is a frame with a normal NBSS header but instead of 0xFF"SMB" it has the 4 letters "exit" as SMB header.

The attached trace tests brlock cleanup. Frame 16 creates a file, frame 19 brlocks it. Frame 41 proves it's locked. Then comes frame 43 which W2k8R2 properly responds to: It sends a RST, because we sent garbage. I would have expected that the brlock from frame 19 is cleaned up, but not so: Frame 46 proves it's still locked. Even more: MMC looking at the local open files still shows that file open, although the TCP connection does not exist anymore. So the only way to access that file is to go to the server console and forcibly close that file.

I'm sure this should NOT end up in the docs, I'd think this is just a W2k8R2 server bug. If required, I'm happy to help you reproduce this in your lab.

With best regards,

Volker Lendecke



More information about the cifs-protocol mailing list