IIS worker process recycling under load; escalated to Microsoft support

[People & Places]

The IIS application pool fronting the internal actuation service recycles under moderate load with an access-violation in a native module, dropping in-flight requests. Reproduced but not root-caused in-house; opened a paid support case with Microsoft.

[M. Ross]

Crash dumps point at a native ISAPI filter, not our code. Beyond what I can debug here. Raising a case with Microsoft with the dumps attached.

[D. Pike]

While we wait — could the health-check be failing a lookup and recycling the pool? i.e. DNS?

[M. Ross]

Microsoft came back: known issue in that filter version, private hotfix supplied (KB on the case). Applied, soak-tested, no recycles in a week. Good support, credit where due. Not DNS, Pike, but nice try. Closing.