This is set by default to 2 hours, while the slave information updating is at 15second intervals.
In what situation can you see problems with us reducing the stalled slave threshold to 5 minutes? I know sometimes lightning does not report back for extended periods of time, but the slave should still be sending pings to mongo, correct?
The default that Deadline installs with is 10 minutes. 5 is probably fine as well, but 10 just gives a little more of a buffer.
The slave does the status updating in a separate thread that has a high priority, so it should be happening regardless of what the render is doing. The only situation that could cause a false positive is if the slave is temporarily unable to access the database for some reason.
Hm. Its 10 mins? Damn. Someone set it to 2hours lol.
Are the repo setting changes logged somewhere?
They’re not. We should probably be doing that though, with the granularity that we do for the job properties. I’ll put this on the wish list.