Sounds good. I do worry about the ones that hang in the tests.
The one that hit me yesterday though was a very long delay collecting test results. It wasn’t even in our Maven run.
And, really, the other issue was simply running with only 3 executors. Things were pretty smooth when I dropped that to 2 a few weeks ago but we need way more than that, especially as I start to propagate/cherry pick updates I made for Arduino on the 9.0
branch. But HIPP just can’t handle it.
I've seen hangs locally too. It seems to happen more often when the machine is under load. From experience, those hangs are often real bugs in the code or flaws in the tests. One
interesting experiment to try is create a VM with just one core, some tests fail much more easily.
One thing we could do is to open bugs for the different hangs and intermittent failures (with logs attached). In Trace Compass, we then linked those bugs to a "meta bug"  so we can track progress and duplicates more easily.
If you are OK with this approach, I can open such a bug and document a few that I've seen. Of course, someone still needs to look at the bugs but hopefully it will be easier to split up the work.
Back from vacation today. Put together my first fix for the Qt new project wizard. But my verify job has hung. I notice Sergey’s has hung too. In fact you look at the history and we have an abysmal record of failures. And they really appear to be more
infrastructure than the actual tests.
Any thoughts on how we can improve this situation. I’d love for people to be able to volunteer slaves for the farm but I’m not sure how well that works in this environment. For example, I’d love to just hook up a VM on my laptop for a few hours at a time.
We talked about this before and I did have a machine on AWS helping out though it didn’t work particularly well. What’s the right way to do this.