-1

I have a .Net based IIS Website that appears to be going down and coming back up without logging it. The only way I'm able to tell this is happening is from our CMS logging the time that it needs to re-initialize, and our Application Dynamics agents logging the time that they need to come back online. We thought this was the App Pool restarting at first, but we have all logging for the App Pool turned on and there are no entries when these events occur. After the event there's a period of slowness on the website that I've been attributing to a recompile time, but again no logs to indicate this.

Metrics on the server during these events appear to be standard and do not deviate from their norms. None of the normal suspects are higher than usual (CPU < 20%, Memory < 15%).

We've tried turning off all services on the server, increasing the worker process count from 1 to various amounts up to 10. Morning restarts of the app pools involved. Daily restarts of the server. Added various logging to Event Viewer.

This is happening during the middle of the day under normal load. There's around 1000 transactions per minute hitting the App Pool.

Yasterdly
  • 1
  • 1
  • You say it's not the app pool going down. But if there's no requests to the site in a while, it will idle out. The worker process will be killed. It's normally not a big deal. You can increase the idle timeout (default is 20 min I think). Or disable it. See https://serverfault.com/questions/42571/is-there-any-harm-in-disabling-the-idle-timeout-for-an-application-pool – mason Jun 30 '23 at 03:14
  • For .NET Framework based web apps, the most likely cause is app domain reload, https://www.tessferrandez.com/blog/2008/11/06/troubleshooting-appdomain-restarts-and-other-issues-with-ETW-tracing.html – Lex Li Jun 30 '23 at 05:39
  • To clarify, this is happening during the middle of the day under normal load. There's around 1000 transactions per minute hitting the App Pool. – Yasterdly Jun 30 '23 at 15:28
  • Instead of leaving new information shattered down in comments, edit the question to include all of them. – Lex Li Jun 30 '23 at 15:34

2 Answers2

0

I would suggest a look at the Reliability Monitor for any crashes, errors, or other events on the server. Click Start, type Reliability and the search should produce the 'View Reliability History' in Control Panel.

To clarify, does the w3svc continue running the entire time the site is up/down?

Protocol
  • 1
  • 2
  • I checked the Reliability Monitor, no reported crashes. – Yasterdly Jun 29 '23 at 23:03
  • All worker processes maintain their PID, and their start times through the events as well. Because of that I didn't check to see if w3svc was running. It looks like my w3svc's last start time was on June 17th, so I'm assuming it's staying up through the events. – Yasterdly Jun 29 '23 at 23:09
  • As it’s currently written, your answer is unclear. Please [edit] to add additional details that will help others understand how this addresses the question asked. You can find more information on how to write good answers [in the help center](/help/how-to-answer). – Community Jul 02 '23 at 09:44
0

Debug Diagnostics Tool (DebugDiag) can be a lifesaver. It can createand analyze IIS crash dumps. Maybe you can find the cause of the crash from the call stack.

IIS Application Pool Crash and Debug Diag: https://learn.microsoft.com/en-us/archive/blogs/parvez/iis-application-pool-crash-and-debug-diag

YurongDai
  • 1,362
  • 1
  • 2
  • 7