Worker process recycles every couple minutes with recommended settings
Hi everyone, According to Joel Oleson's blog post at http://blogs.msdn.com/b/joelo/archive/2007/10/29/sharepoint-app-pool-settings.aspx, for my server farm with a WFE, app/search server, and db server with an average around 150 current connections at any one time, I should set my worker process to recycle when max virtual or max physical hits around 800MB-1GB. When I do this though it will recycle every couple minutes. Any insights?? Does this indicate an issue elsewhere in my farm that I need to look into, or is it OK? Right now I have completely removed these settings and am "letting it ride." We have just 3GB total ram on our server and generally it looks like it sits around 1.5GB being used. Scott more info - the WFE is using Windows Server 2003 64-bit with WSS 3.0 installed.
March 31st, 2011 12:22pm

Hi Scott, Do you have any custom code or workflows on that server ? Are there any crawling shedules active ? Do you have particullary large lists ? Inspect the logs of your server and check for 'Possible excess of undispose objects or something like that.' Take a look at the IIS and Asp.Net performance counters, some of them allow to track number of requests, garbase collections, application pool recycles. Try PAL on codeplex that analyses all the perf counters and that will maybe help you to track down the issue. I know its not a direct answer and more a set of clues, but i hope this will help somehow. Regards
Free Windows Admin Tool Kit Click here and download it now
April 3rd, 2011 8:05pm

Thanks I've been trying to track this down. Oddly enough something that helped the RAM usage a lot was just creating a new application pool and assigning my IIS instance to that. For some reason this cut the range from 1.2 GB - 2.0 GB down to 500MB - 1.4 GB. I don't know why. We don't have any custom workflows. It's out of the box WSS. We do have third-party web parts installed though. It appears that using our Bamboo web parts like a list rollup is the main culprit behind the high RAM usage. There are some errors in the log like you mentioned that point to the Bamboo web parts. Something else I have noticed that is really strange, is that if I track "Current Connections" in the performance monitor, they average 50-100 but the Current NonAnonymous Users + Current Anonymous Users trackers add up to average around 5. Is that normal? I feel like these two numbers should not be so far from each other. I would not have thought there would be 10+ connections needed per visitor. Also aren't "current connections" supposed to close out once a page is loaded? The "Current Connections" seem to increase throughout the day. Even when there are 0 current users, there are 40+ connections open.
April 5th, 2011 4:36pm

Hi scott, The happy to see you could at least find some cause to you problem. These errors are not fatal and sometimes just almost unavoidable. So dont strive to zero, but there are some numbers n that error about he number you have excessed. That can be an indication of real memory leaks. For the request to connection ratio depends on what the performance counter is. IIS connections, this could be much more elevated than the users, because one page, makes a couple of request to get images, scripts etc so that can explain the diff, and yes they should disapear after a time. SQL connections that should not be that high, at least i dont think so, but those connection can stay open indefinetly. Any crawling shedules, orso ? does it still occures when the search is stopped ? Any automated monitoring tools ? warm up scripts ? Regards
Free Windows Admin Tool Kit Click here and download it now
April 5th, 2011 4:56pm

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics