On Thu, Apr 17, 2025 at 09:16:54AM +0200, Paul Gevers wrote:
On 17-04-2025 05:15, Bo YU wrote:I also noticed there are architectural differences between the boxes, so it is possible that there really is a problem other than speed.Hmm, I am afraid the latter is true.So, speed? Does increasing the timeout help, or does it need an unreasonable timeout on those unmatched/licheepi4a machines?If the timeout needs to be unreasonable, we can add the package to the reject_list until we only have faster hardware. Otherwise, just increase the timeout.
What's reasonable? I don't really want to just keep uploading packages with longer and longer timeouts to see what works, that's why I was asking if it was possible to run manually on one of the machines having issues to see if there's something wrong here other than the timeout. I changed the timeout from 1min to 5min. On some of the machines the test finishes in ~2min, on the ones where it fails it's now dying after almost 10min. Is it just that there's that much difference in the speed of the machines? (From those numbers it's taking longer to even get to the test timeout than it takes for an entire run when it works.)