I have this Farm, with two WFE’s using Windows NLB. One of the WFE’s works great, however the second WFE thinks its a great idea to throw the dreaded “Unexpected Error” message once a day for a two web apps sharing a common app pool. I think “how odd…one WFE is not in sync with the other”, so I recycle the app and all is good. As it turns out this begins to be a 24 hour pattern of a manual recycle of the app pool, obviously not acceptable.
Now…I know what you are thinking “Dude there is a memory leak somewhere, fix it”. And while I agree something is causing a memory leak I’m not sure it’s the web part everybody has in the back of their mind. After doing some research I found this forum post where it appears some others have the same problem.
I don’t have a pretty answer for this problem but I do have something that works a tad better than an scheduled IIS reset. I’ve been successful (knock on wood) in setting the recycle settings for the app pool in question to after the SharePoint search service runs.
Again…Not Pretty but seems to work so far.