> On Feb. 12, 2015, 7:47 p.m., Jie Yu wrote: > > Just realized that there still a race here with killTask (killTask happens > > between `_runTask` and `__runTask`). We could end up with the situation > > where the executor receives an KillTaskMessage with an unknown task since > > RunTaskMessage hasn't reached the executor yet. :( > > Timothy Chen wrote: > True if kill task is sent without any task status update yet, should we > have a new state for the executor so that we don't proxy until it's fully > ready? > Or what's in your mind to fix this?
I am thinking about using the queuedTasks because it's in fact queued (and not yet launched). - Jie ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/30945/#review72227 ----------------------------------------------------------- On Feb. 12, 2015, 7:30 p.m., Jie Yu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/30945/ > ----------------------------------------------------------- > > (Updated Feb. 12, 2015, 7:30 p.m.) > > > Review request for mesos, Ian Downes and Vinod Kone. > > > Bugs: MESOS-998 > https://issues.apache.org/jira/browse/MESOS-998 > > > Repository: mesos > > > Description > ------- > > Waited containerizer update to finish before sending task to the executor in > Slave::_runTask. > > > Diffs > ----- > > src/slave/slave.hpp 7a399f6df50c69b7e1e12d74f076fa57b6edb1b3 > src/slave/slave.cpp f39a876cdd6b580a7a75fd053e6923761bee7635 > > Diff: https://reviews.apache.org/r/30945/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Jie Yu > >
