On Sun, Feb 09, 2020 at 03:25:48PM +0100, Theo Buehler wrote: > On Sat, Feb 08, 2020 at 03:22:52PM -0500, Kurt Mosiejczuk wrote: > > On Sat, Feb 08, 2020 at 07:26:31PM +0100, Theo Buehler wrote: > > > I can add arm64 to sthen's table. Builds & works ok:
> > > i386 builds OK tests OK > > > macppc builds OK tests OK > > > amd64 builds OK tests OK > > > arm64 builds OK tests OK > > > armv7 runtime/tests fail (no regression, in-tree fail too) > > Well, sparc64 builds OK, but the tests fail. In-tree version passes all > > tests. > I got access to a sparc64 test machine and haven't managed to reproduce > this failure. Build and tests were OK three times in a row. The machine > I'm on is slightly slower than yours. My test machines is a 1.5Ghz v210, not the cluster, so you may or may not be slower... > I'm now running a loop building and testing, maybe I'll hit it... > If I read your log and the test itself correctly, one of the clients got > stuck and didn't send anything to the server, then you hit the default > test timeout of 10 minutes. I'm not sure how worrying this is. > It's also not clear whether incident is related to this update or to > sparc64 at all. Could you perhaps try to reproduce it on your machine? I ran the tests a bunch of times last night. At first I was thinking it was only the first test run that would fail, but it seems to actually be an intermittent failure. I got a failure and then 3 passes and then a failure. I may try running the original tests multitple times to see if we have a failure crop up there over multiple runs. --Kurt