Odd BigIP behavior

Hi,

 

We have had two occasions now where our BigIP has started doing odd things such as randomly not connecting clients to pool members even though it shows the pool members as active and we can connect directly to the pool members without any issues. I'll attempt to describe what occurs. It's a little convoluted though.

 

 

We use PeopleSoft and WebSphere, PeopleSoft for our student system and PeopleSoft for our website. To integrate to two, we have some remote portlets on WebSphere that communicate to PeopleSoft via PeopleSofts Weblogic webservers. The URLs used are load balanced between the two Weblogic webservers by the BigIP.

 

 

Now although we haven't identified the exact cause yet, for some reason the Weblogic web servers grind to a halt with a large number of connections. It gets to the point where one becomes unreposive and the BigIP flags that member as down and then directs all the conenctions to the 2nd weblogic server. Then WebLogic on that server soon becomes unresponsive as well as it is being hammered. Although we stop/restart the Weblogic web servers, as soon as they come back up they are hammered with hundreds of connections again. If we reboot the WebLogic servers and the WebSphere nodes we then notice that the BigIP is no longer behaving as expected, at least with regard to the virtual servers responsible for PeopleSoft and WebSphere conenctions. What happens is that conenctions to pool members just fail. The managment interface on the BigIP shows the pools and members within the pools as being up. If we connect to the members directly then all is fine, but not via the BigIP. we tried stopping/starting the virtual servers to no avail. A bigip reboot did however resolve the issue.

 

 

I'm just wondering if anyone else has had the BigIP do odd things like this.

 

 

tia

 

 

Craig