Allow cleanup of redis sockets to prevent event core leaking.

[schveiguy]

See issue #2245. This allows user code to clean up a RedisClient or RedisSessionStore before exiting the program to prevent messages of leaked eventcore drivers. Not attached to the function name, so if that needs to change, let me know.

[Geod24]

Any way this could be tested ?

[schveiguy]

I'm trying to think of how to test this without having an actual Redis server...

Indeed, there's only 2 unittests in this file, and they both do not work with the actual Redis server.

I'm open to ideas, but I don't know if it can be done.

[s-ludwig]

I'm trying to think of how to test this without having an actual Redis server...

Indeed, there's only 2 unittests in this file, and they both do not work with the actual Redis server.

There is a Redis integration test in the "tests" directory where this could be added. The only way to verify this that I can see is to manually count all open file descriptors/SOCKET handles, before and after. I'd not make this mandatory, but it would be nice to add a simple call in the existing test, so that ideally the current warnings go away (CI output).

[schveiguy]

There is a Redis integration test in the "tests" directory where this could be added.

OK, will do.

[schveiguy]

Updated

[schveiguy]

Ugh, I keep screwing up my redis test. I couldn't run the test locally because I don't have a redis server that I want to muck with. In any case, I had to force push again. If it turns green, I'll ping you to reapply the auto merge.

[schveiguy]

OK, ping @s-ludwig. The failure seems unrelated to the patch.

[schveiguy]

ping @s-ludwig again.

[s-ludwig]

Sorry, missed the first time! Restarted and toggling auto-merge.