Reexamine the idle timer

[dluman]

It appears that some combination of circumstances leads to containers being left open for more than 24 hours in some cases. Unsure as to why this happens. We might need to verify that the inactivity timer is working correctly.

[dluman]

We need to observe, but I have an idea that this happens due to socket closure issues. I haven't been able to reproduce this. Here, we might need to emit a SIGUSER event?

[dluman]

So far, I've tracked this down to (potentially) using .end() and .destroy() -- I converted to using only .end(). Not sure that it helps in total. There still seems to be an issue with stranded containers.

[dluman]

So far, I've tracked this down to (potentially) using .end() and .destroy() -- I converted to using only .end(). Not sure that it helps in total. There still seems to be an issue with stranded containers.

This seems to not necessarily help the case, though it's probably the thing that solves the socket error event that requires restarting student sessions wholesale. A log would help greatly in this instance.