Closed stephenpascoe closed 9 years ago
That probably explains it, as all the other counts seemed accurate and jobs had been building up yesterday due to some overnight failures. The previous version placed such tasks both in upstream failed/disabled and pending. It might be confusing to users to change it, but I can see why you did. I think I can get used to either solution. Are upstream failed and upstream disabled all pending?
All upstream failed/disabled are pending. This seems logical to me as a way to not count pending tasks twice. I've confirmed the counts seem to add up when we have pending and failed tasks.
From https://github.com/spotify/luigi/pull/1086
Are any of these pending tasks upstream failures or disabled? I'm counting these independently. Maybe this is also related to the server limit (see below)?