[
https://issues.apache.org/jira/browse/MESOS-1559?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14065775#comment-14065775
]
Yan Xu commented on MESOS-1559:
-------------------------------
Use something like this on Jenkins. This can be generalized and put in Mesos
./support folder.
{code}
cmd="MESOS_VERBOSE=1 GLOG_v=1 make check GTEST_SHUFFLE=1"
start=$(date +"%s")
$cmd &
pid=$!
duration=$((160 * 60))
while [ $(($(date +"%s") - $start)) -lt $duration ]; do
running=`jobs | grep "Running" | wc -l`
if [ $running -eq 0 ]; then exit 0; fi
sleep 10
done
echo "Process still running after $duration seconds. Attaching gdb..."
tmp=`mktemp -t XXXXX`
echo "thread apply all bt" > $tmp; gdb attach $pid < $tmp
echo "Killing the process"
kill $pid
{code}
> Allow jenkins build machine to dump stack traces of all threads when timeout
> ----------------------------------------------------------------------------
>
> Key: MESOS-1559
> URL: https://issues.apache.org/jira/browse/MESOS-1559
> Project: Mesos
> Issue Type: Improvement
> Components: build, jenkins
> Reporter: Jie Yu
> Assignee: Yan Xu
> Priority: Minor
>
> Many of the time, when jenkins build times out, we know that some test
> freezes at some place. However, most of the time, it's very hard to reproduce
> the deadlock on dev machines.
> I would be cool if we can dump the stack traces of all threads when jenkins
> build times out. Some command like the following:
> {noformat}
> echo thread apply all bt > tmp; gdb attach `pgrep lt-mesos-tests` < tmp
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.2#6252)
