MQSeries.net :: View topic

Heinz57 · Posted: Tue Feb 13, 2007 7:16 am Post subject: biphttplistener advice

I've looked at several of the posts but none solve my issue. We are at WBIMB v5 csd 4 (I know it's behind but we're moveing to v6).

Without all the details of why, we run MQSIRELOAD every morning to eliminate stale odbc threads. We have about 40 EG's and our reload takes 5-10 minutes to complete. We have an MQ Clustered environment in 2 seperate geographical locations and fail back and forth to a single broker during reload by inhibiting/allowing queues. This is all automated.

Our new issue is with the biphttplistener. We have several flows that are built for http calls and they invoke other flows to hit the backend systems or otherwise satisfy the request. We are also in the process of implementing a Cisco Content Switch to route the traffic between the 2 sites.

Here's our problem;
Because we cannot control the start/stop of the listener (on it's own) we have requests hitting the broker during the mqsireload. These requests ultimately time out or fail. I realize that the biphttplistener.exe is tied to the broker service. Even if we stop/start the broker, our larger EG's take time to ramp up but the http requests keep flying in and timing out. Once the appropriate EG's are running, all works great !

Our question;
Has anyone else faced this issue with attempting to failover a server (similar to what we do with inhibitting queues) for the http traffic ? If you have a process, has it been automated ?

Any and all advice is appreciated...

Thanks