Dynamic SQL

Hi Len,

 

 

Can you elaborate on the failure symptoms? ie, do all requests to the application get an application level error? Do some requests get a TCP reset?

 

 

Are the app servers communicating with the databases through the BIG-IP? Are the databases being load balanced? If you remove the HTTP class(es) from the VIP(s) and completely bypass ASM, does the issue still occur?

 

 

Aaron

Ok let me answer the best that I can.

 

 

Cfg:

 

1.Each server has two VIPs and two pools and an snat mapping public to private IP.

 

2.One vip is for port 80 and the other vip for all ports.

 

3.This is only the ASM4100, there is no load balancing LTM, etc.

 

4.From any node in the pools behind the ASM, I can access everything in any external network just like the ASM is not there and vice versa back into the node.

 

5.The ASM is in transparent mode.

 

6.The DB’s are in another network that the pools cannot directly route to so they connect via their public nat address.

 

 

Problem:

 

1.From the nodes in the pools, I can create and connect via ODBC to any SQL server that is not clustered

 

2.From the nodes in the pools, I can create but cannot connect via ODBC to any SQL server that is clustered

 

3.The web apps, when moved behind the ASM, give SQL errors like there is no connection to the backend DB.

 

 

I do not see any TCP resets, there are no errors in the logs.

 

 

So, yes the SQL connection is going through the F5, but I dont see anywhere that the F5 was/is blocking.

 

Can you establish any TCP connection from the 4100 to the clustered SQL servers?

 

 

Are the clusters using NIC teaming? If so, check SOL3009 (Click here) for details on an issue with "load balancing" traffic to a server with teamed NICs. The gist is that you should configure 802.3ad link aggregation instead.

 

 

 

Aaron

No teaming. I just verified. I will do another packet capture on the F5 to see if there is anything revealing. By the way, thanks for the help.