> On June 27, 2016, 6:27 p.m., Vinod Kone wrote: > > src/webui/master/static/js/controllers.js, line 149 > > <https://reviews.apache.org/r/49228/diff/4/?file=1431212#file1431212line149> > > > > task.statuses[0] could be a terminal state if a task went directly from > > TASK_STAGING to TASK_LOST, i.e., a task was removed before any update for > > it was received by the master (e.g., agent removed). > > Tomasz Janiszewski wrote: > What is desired behaviour for this situation? Now start and stop time > will be equal. Should start time be `null` when first status is terminal? > > Vinod Kone wrote: > looks like start time is capturing when the master receives a > TASK_RUNNING update and not when the task launch request from the scheduler > was received by the master. so i would prefer to not set `task.start_time` if > the master didn't receive that update.
I added condition that will not set `task.start_time` when first status state is terminal. - Tomasz ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/49228/#review139622 ----------------------------------------------------------- On June 27, 2016, 9:05 p.m., Tomasz Janiszewski wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/49228/ > ----------------------------------------------------------- > > (Updated June 27, 2016, 9:05 p.m.) > > > Review request for mesos, Benjamin Mahler, haosdent huang, Ross Allen, and > Vinod Kone. > > > Repository: mesos > > > Description > ------- > > Only set the terminal time if the last state is terminal. > > > Diffs > ----- > > src/webui/master/static/js/controllers.js > be3fa1f1fba9eda7ebcc9bf33142fe4307dd8d90 > > Diff: https://reviews.apache.org/r/49228/diff/ > > > Testing > ------- > > Generate orpahne tasks that is alive. It should have not set stopped time. > > > Thanks, > > Tomasz Janiszewski > >
