The immediate path is to confirm whether the old IIS worker process is actually crashed, stuck during shutdown, or only appears as a stale PID.
- Check the real cause in Event Viewer.
- On the server, open Run with Win+R, type eventvwr, and open Event Viewer.
- In the Application log, look for:
- Application Error, Event ID 1000
- .NET Runtime, Event ID 1026
- For IIS worker process failures, focus on w3wp.exe as the faulting application.
- Record the Faulting module name and Exception code.
- Interpret the exception code.
- 0xC0000005: access violation.
- 0xe0434352: unhandled CLR exception in application code. Check the matching .NET Runtime 1026 event details, especially Description and Exception Info.
- 0xC00000fd: stack overflow, commonly caused by infinite recursion.
- Capture a dump of the affected w3wp.exe instead of trying to force-kill it repeatedly.
- This is the most useful way to help developers identify why the old process is hanging onto the gRPC port.
- Safer options from the documented workflow are:
- Windows Error Reporting (WER) first
- Debug Diagnostic Tool
- ProcDump
- Start with WER.
- Check Event Viewer for Windows Error Reporting events.
- Look in the directories named in those events for .mdmp or .hdmp files.
- If dumps already exist, provide them to the developers or support.
- If not, configure HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Windows Error Reporting\LocalDumps with:
- DumpFolder = path on a non-system drive with enough space
- DumpCount = 3
- DumpType = 2 for a full dump
- Warning: registry changes are risky. Back up the relevant key first and use a non-system drive for dump storage.
- If the problem must be reproduced, use ProcDump on the IIS worker process.
- Download ProcDump.
- Extract it to a folder on a drive other than the system drive.
- Open Command Prompt as administrator.
- In IIS Manager:
- select the server
- open Worker Processes
- note the PID for the affected w3wp.exe
- Run:
procdump -ma -e 1 -f "<typeOfException>" PID -
-macaptures a full dump,-e 1captures on first exception, and-ffilters by exception type.
- If the issue happens during app pool shutdown, check shutdown timing.
- IIS application pools need time to finish in-flight requests during shutdown.
- If ShutdownTimeLimit was reduced too far from the default 90 seconds, IIS can log shutdown-related warnings and the pool may not stop cleanly under load.
- If long-running requests are expected, increase ShutdownTimeLimit and capture a dump of w3wp.exe while the shutdown problem is occurring.
- About
taskkillreporting “process not found”.- On Windows Server 2012 / 2012 R2 under heavy load,
Taskkill.exe /PID <pid>can fail with:-
ERROR: The process "<pid>" not found.
-
- That behavior is documented, so this result alone does not prove the process is gone cleanly.
- On Windows Server 2012 / 2012 R2 under heavy load,
- After a dump is collected, analyze it.
- Open DebugDiag2 Analysis.
- Use Add Data Files to load the dump.
- Run Start Analysis.
- In the report:
- for 0xC0000005, find the failing call stack and identify application code
- for 0xC00000fd, look for repeating functions indicating recursion
- for 0xe0434352, identify the call stack returning the application exception
This gives developers actionable evidence without immediately restarting the server. The key step is collecting a dump from the hung or crashing w3wp.exe and correlating it with Application Error 1000 / .NET Runtime 1026 events.
References: