----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/23700/#review49137 -----------------------------------------------------------
Talked to Jie. This script looks for the newest (most recently started) process in this session and this may sometimes be not what we want. If some tests themselves fork subprocesses and depending on whether setsid() is called what we attach to may be the test or the subprocess and we may want to debug either/both of them. I can revise the script to attach to all the ancestor processes of the newest process and I think this works better. Still, this script tries to be useful in a generic manner with best effort. I imagine there are other cases where the script needs to be tailored depending on where the anticipated bug is. - Jiang Yan Xu On July 29, 2014, 11:46 a.m., Jiang Yan Xu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/23700/ > ----------------------------------------------------------- > > (Updated July 29, 2014, 11:46 a.m.) > > > Review request for mesos, Ben Mahler, Dominic Hamon, and Jie Yu. > > > Bugs: MESOS-1559 > https://issues.apache.org/jira/browse/MESOS-1559 > > > Repository: mesos-git > > > Description > ------- > > Example usage: > > ./support/timed_tests.sh -m "make check GTEST_FILTER=" "MESOS_VERBOSE=1 make > check GTEST_SHUFFLE=1" $((120 * 60)) > > # Bypass the 'make' stage. > ./support/timed_tests.sh "MESOS_VERBOSE=1 make check GTEST_SHUFFLE=1" 3600 > > It works by setting a new sid for the test script so that all subprocesses > are in this new session. > > > Diffs > ----- > > support/timed_tests.sh PRE-CREATION > > Diff: https://reviews.apache.org/r/23700/diff/ > > > Testing > ------- > > Tested on Linux with and without Jenkins. > > > Thanks, > > Jiang Yan Xu > >