fuse silently reconnecting?

[boucher]

Apologies upfront for this being somewhat vague, but it seemed like filing an issue would be the best route to get some feedback.

I'm running fuse/fuse-bindings inside of a docker container, and consuming that filesystem inside of other docker containers, which mostly works great. But I'm running into an issue where occasionally something is causing the "consumer" docker containers to lose access to the fuse filesystem ("Transport endpoint is not connected" error).

I can reproduce a similar state by just restarting the process that's running the fuse-fs -- any pre-existing mounts that point to something inside the fuse-fs fail. Re-mounting them again works. This hints to me that maybe somehow fuse is being remounted internally in response to some event or error, which isn't surfaced to me at the node level? Looking through the code I didn't see anything obvious though.

Here's a repo I made with instructions on how to reproduce the restart failure mode:
https://github.com/boucher/docker-fuse-restart-test

Any ideas?

[boucher]

I'm pretty sure this turned out to be a case of responding to the same event twice (missing return in an if block). Happy to close this ticket, but is there any way to add detection / better error reporting for this kind of mistake?