Hi Chriss, I'm the maintainer of memcached on Fedora and RHEL, On Thu, 10 Sep 2020 22:58:03 -0000 "Chris Lamb" <[email protected]> wrote: > Hi Moises, > > > We run the tests afterward on a dedicated machine.. > > The question here is not when or where these tests are run - can you > reproduce these test failures at all on this other machine? I can > reproduce them locally here (NB. with TLS enabled during the build). > > If you can/cannot, sharing the test log would be a good next step. I can not. The testsuite on Fedora does not hang. I remember a situation similar to this. When i first enabled tls in Fedoras memcached the testsuite hanged because test certificates had short key and Fedora has hardened crypto environment, thus memcached crashed during testing and suite was not prepared for that.
I can show you a test log from Fedora Rawhide where we have 1.6.7. https://tkorbar.fedorapeople.org/test.log I will try to install debian VM and reproduce this. > > > ...we do not run memcacheds test suite during build because it is cpu > > load dependent and not stable enough for us to rely on it during build. > > As an unrelated remark, if the tests are CPU-load dependent (and > therefore non-deterministic) I would suggest you file that as an > upstream issue. A randomly-failing testsuite is, as I am sure you are > aware, not good software engineering practice. > Yeah, i know that the suite is not ideal but i did not yet get to handle that. It is stable on normal build machines but if you run it on some powerful server CPU then techniques like "sleep(0.5)" starts to fail. I will keep in mind to fix this. > Thanks for looking into this so far. > > Regards, > > -- > ,''`. > : :' : Chris Lamb > `. `'` [email protected] 🍥 chris-lamb.co.uk > `- > > Regards.
