Problem with WAFS not caching certain servers

[balderslade]

The system is one Core and two Edge devices, plus the CM. The core is a WAE-512 and the edge devices are NME-WAEs, all running 4.0.19.14. WCCP v2 being used for redirection.

Core and Edge file services are enabled and running in transparent mode with auto-discovery enabled. Preposition is not used.

We ran some tests to make sure CIFS caching was functioning properly and from some servers it was fine - a 5MB file took 1.5 mins on the first pass over the WAN, and under 1 second on subsequent attempts. The nearest Edge device reported cache utilisation and a 'sh cifs session list' showed the session, while 'sh cifs auto-discovery host-db' shows the auto discovered file server. All good.

But when pulling files off the main file server the session does not show under 'sh cifs session list' and the server doesn't get auto-discovered. I added the server explicitly to the core-edge connection and verified this by doing a 'sh cifs auto host', but sessions to that host still don't appear in the cifs session list and files aren't being cached on the edge device.

The connections are being 'optimized' by the WAE devices and I can see the connection being (highly) compressed, the file's just aren't entering the cache.

The file server is Windows 2003 Storage Server, which isn't named explicitly on the OS compatibility matrix, but I can't see why it would be appreciably different to standard 2K3.

Any ideas or suggestions?

[dstolt]

Check if the server has SMB Signing enabled. If you see SMB signing error in the Rx.logs, the server will not accelerated via the CIFS AO.

For more on SMB Signing, check out http://support.microsoft.com/?kbid=887429

Dan

[balderslade]

Checked the server in question and it's not configured for SMB signing. Captured some packets as an added precaution and it's not setting signing in the negotiate protocol response.

BTW, where is this Rx log?

Thanks for your help so far. Anything else to check?

Brendan

[dstolt]

Brendan,

local1/logs/actona/RxLogging.log

Another good place to look for Rx logs would be to look in is:

local1/errorlog/actona/RxInternal.log

Look for the IP address of your server by using "find-pattern match x.x.x.x RxLogging.log"

What happens if you let the server be autodiscovered? You will see more log messages and can see more in the CLI's "sh cifs auto-discory status|host-db|last" and "sh policy-engine application dynamic"

Double check your core and edge services are running and that the connectivity directive is configured (if CIFS works on the other file servers, then it's probably working).

Are your edge WAEs within 5 ms of the files server? Is it further then 25 ms from the core WAE? Both of those will cause CIFS AO bypass.

Dan