Juwei, Its odd that a killed job should get "recovered" back into running state. Can you not simply disable the JT recovery feature (I believe its turned off by default)?
On Fri, Jul 1, 2011 at 10:47 PM, Juwei Shi <[email protected]> wrote: > Thanks Harsh. > > There are "recover" jobs after I re-boot mapreduce/hdfs. > > Is there any other way to delete the status records of the running jobs? > Then they will not recover after restarting JT? > > 2011/7/2 Harsh J <[email protected]> >> >> Juwei, >> >> Please do not cross-post to multiple lists. I believe this question >> suits the mapreduce-user@ list so am replying only on there. >> >> On Fri, Jul 1, 2011 at 9:22 PM, Juwei Shi <[email protected]> wrote: >> > Hi, >> > >> > I faced a problem that the jobs are still running after executing >> > "hadoop >> > job -kill jobId". I rebooted the cluster but the job still can not be >> > killed. >> >> What does the JT logs say after you attempt to kill a job ID? Does the >> same Job ID keep running even after or are you seeing other jobs >> continue to launch? >> >> -- >> Harsh J > > -- > - Juwei > -- Harsh J
