> On July 25, 2013, 1:53 a.m., Vinod Kone wrote: > > hadoop/mesos/src/java/org/apache/hadoop/mapred/MesosScheduler.java, lines > > 76-79 > > <https://reviews.apache.org/r/11124/diff/8/?file=327717#file327717line76> > > > > Just curious, why the task trackers would never launch? The mesos slave > > kills the executor if it doesn't register within a timeout? Would that not > > be sufficient here? > > Did you see hung task trackers? > > Brenden Matthews wrote: > It seems to happen occassionaly on slaves that are busy. Sometimes it > will get 'stuck' somewhere between grabbing the executor from HDFS, > extracting it, and then launching the TaskTracker.
the first 2 cases should be handled by the slave, the latter probably not. - Vinod ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/11124/#review23810 ----------------------------------------------------------- On July 25, 2013, 1:28 a.m., Brenden Matthews wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/11124/ > ----------------------------------------------------------- > > (Updated July 25, 2013, 1:28 a.m.) > > > Review request for mesos. > > > Repository: mesos > > > Description > ------- > > Kill tasks that never properly launch. > > After trying to launch a task tracker, we'll wait up to 5 minutes before > giving up and killing the task. > > Review: https://reviews.apache.org/r/11124 > > > Diffs > ----- > > hadoop/mesos/src/java/org/apache/hadoop/mapred/MesosScheduler.java > 279f84e0f0c43ad3cfd9e4442010e706ee3565d9 > > Diff: https://reviews.apache.org/r/11124/diff/ > > > Testing > ------- > > Used in production at airbnb. > > make -j10 check && cd hadoop && make hadoop-2.0.0-mr1-cdh4.2.1 && make > hadoop-0.20.205.0 && make hadoop-0.20.2-cdh3u3 > > > Thanks, > > Brenden Matthews > >
