Please answer these questions before submitting your issue.

What version of gRPC and what language are you using?

1.4.2 - C++

What operating system (Linux, Windows, …) and version?

Windows 7.

What runtime / compiler are you using (e.g. python version or version of gcc)

Visual Studio 2015 Update 3

What did you do?

Server setup
One thread dedicated to picking events from the completion queue and another thread processes it. The thread responsible for processing also shuts down the server when required.

Set GRPC_ARG_MAX_CONNECTION_IDLE_MS to 1 hour and set GRPC_ARG_MAX_CONNECTION_AGE_MS to (INT_MAX - 1)

Client setup
3 client threads using sync api and infinite timeout (default). Threads repeatedly call a number of rpcs in a never ending loop.

After letting it run for few minutes, exited client process and then called a graceful shutdown on the server.

What did you expect to see?

Server shutdown gracefully.

What did you see instead?

Server crashed randomly (not always). 2 out of 3 times.

See the attached screenshot with various relevant variables and callstack.
Crash 1

Crash 2