This issue started about 3 weeks ago. Randomly, we've run into one or more of our three app tier servers performance dropping to a crawl. After logging in, we can see the CPU is maxed and a w3wp process at >99% . Opening IIS and checking the work processes, I can see the TFS app pool as the culprit. I've let it go, thinking it was a process that was taking a long time, but once I get enough users complaining, we end up restarting the app pool.
I've looked through the event logs, can't find anything (other than some timeouts after the CPU maxes). The workaround is to restart the app pool, but that kills any connections. This causes builds to fail, along with any long running action (branching, merging, etc) for machines that are connected to that app server. Where else can I look to see what's causing this? I turned on failed request logging...got nothing so far.
They are all same OS, Server 2012 Standard, and same version of TFS, 2013 RTM.