Gapis timeout

[dsrbecky]

My gapis keeping exiting on a big trace:
W 00:17:00.031 gapis Stopping GAPIS server as it has been idle for more than 3.333333333s (--idle-timeout)
I am not stopping any process, and running just at info log-level.

Also, should we it an error? It took me a while to notice it all the other log messages.

It also seems that gapis will explicitly kill gapir process on timeout. Can we just let gapir die on its own? I mistook the abrupt death for driver crash.

[ben-clayton]

"3.333333333s" is a bug, it's actually 10 seconds.

I've seen this as well, but I'm not sure what the correct solution is.
It seems odd / wrong that GAPIC hasn't sent a ping to GAPIS within 10 seconds when it should be pinging at 1 second intervals.

I guess it's possible that with the workload of opening a large trace the heartbeat thread is just not scheduled. If so, the only short-term solution I can think of is to bump the timeouts. @pmuetschard proposed scanning for process IDs, but that involves per-platform code.

Also, should we it an error?

Sure.

It also seems that gapis will explicitly kill gapir process on timeout. Can we just let gapir die on its own?

When gapis times out, the main function returns, the main context is cancelled so that things clean up, and any outstanding processes are terminated so we don't orphan them. I'd argue that killing child processes is the correct thing to do when we exit - but maybe we should be using an interrupt instead of forcefully closing the child process. Unfortunately that isn't supported on windows.

[dsrbecky]

Increased the timeout in #777
It still dies after a minute though

[AWoloszyn]

FYI, I got the 1M timeout on Windows for a particularly large trace. Things had not hung yet though.