> On Jan. 14, 2015, 10:09 a.m., Adam B wrote: > > Were you able to reproduce the issue without your changes and prove that it > > was fixed with your changes, or are you guessing here?
Pasting in Till's summary of the problem: "After some investigation, it turns out that my two-fold solution had a serious flaw — it all stems from the way CyrusSASL works — in the end, my code re-adds the aux-prop-plugin for every new instance of the master (libprocess-)process. CyrusSASL does bind its context to the os-process (transparently). So when the tests keep using the same os-process (which they certainly do), my patch keeps adding more (identical) plugins. All those plugins remain active and at some point this will trigger delays big enough to break slow machines (e.g. travis-ci)." - Adam ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/29887/#review68061 ----------------------------------------------------------- On Jan. 14, 2015, 9:36 a.m., Till Toenshoff wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/29887/ > ----------------------------------------------------------- > > (Updated Jan. 14, 2015, 9:36 a.m.) > > > Review request for mesos and Adam B. > > > Bugs: MESOS-2050 > https://issues.apache.org/jira/browse/MESOS-2050 > > > Repository: mesos-git > > > Description > ------- > > see summary. > > > Diffs > ----- > > src/authentication/cram_md5/authenticator.hpp 1819ef4 > src/authentication/cram_md5/auxprop.hpp b894386 > src/authentication/cram_md5/auxprop.cpp cf503a2 > > Diff: https://reviews.apache.org/r/29887/diff/ > > > Testing > ------- > > make check (with enhanced verbosity, GLOG_v=2 and gtest_repeat=10000) > > > Thanks, > > Till Toenshoff > >
