Looks like still problematic, Thomas. 12/11/14 18:10:36 INFO bsp.TestBSPMasterGroomServer: Temp file length: 20 12/11/14 18:10:36 INFO bsp.TestBSPMasterGroomServer: Client finishes execution job. 12/11/14 18:10:36 INFO bsp.FileInputFormat: Total input paths to process : 1 12/11/14 18:10:36 INFO bsp.FileInputFormat: numSplits: 3 12/11/14 18:10:36 INFO bsp.FileInputFormat: computeSplitSize: 6 (6, 1, 33554432) 12/11/14 18:10:36 INFO bsp.FileInputFormat: Total # of splits: 4 12/11/14 18:10:36 WARN bsp.BSPJobClient: No job jar file set. User classes may not be found. See BSPJob#setJar(String) or check Your jar file. 12/11/14 18:10:36 INFO bsp.JobInProgress: num BSPTasks: 4 12/11/14 18:10:36 INFO bsp.JobInProgress: Job is initialized. 12/11/14 18:10:36 INFO bsp.BSPJobClient: Running job: job_201211141810_0001 12/11/14 18:10:36 ERROR bsp.SimpleTaskScheduler: Could not schedule all tasks! 12/11/14 18:10:36 ERROR bsp.SimpleTaskScheduler: Scheduling of job Pagerank could not be done successfully. Killing it!
On Wed, Nov 14, 2012 at 6:14 PM, Thomas Jungblut <[email protected]> wrote: > That was fixed in TRUNK long time ago. > > 2012/11/14 Sandy Ding <[email protected]> > >> You are welcome, Edward. >> In my case, I run hama on my VM under psudeo-distributed mode. >> bsp.local.tasks.maximum = 10 >> bsp.tasks.maximum = 3 >> I didn't set task number. >> When the file is somehow split into 4(4 tasks), job fails; when it's split >> into 3, job succeeds. >> And my hama version is 0.5.0. >> >> >> 2012/11/14 Edward J. Yoon <[email protected]> >> >> > Oh, when you set the number of tasks to 3 or more, Job fails. >> > >> > Right? >> > >> > Thanks Sandy. It looks like a bug. >> > >> > On Wed, Nov 14, 2012 at 11:56 AM, Sandy Ding <[email protected]> >> > wrote: >> > > A B D >> > > B C D >> > > C A >> > > D C >> > > won't work. >> > > >> > > a b d >> > > b c d >> > > c a >> > > d c >> > > works under name pr.sim.abcd, but won't work with other names, like >> > > pr.in.xx. >> > > >> > > a b d >> > > b c d >> > > c a >> > > won't work. >> > > >> > > >> > > >> > > 2012/11/13 Edward J. Yoon <[email protected]> >> > > >> > >> Hi, >> > >> >> > >> Can you attach your input file here so that I can check what's wrong? >> > >> >> > >> On Tue, Nov 13, 2012 at 10:28 PM, Sandy Ding <[email protected] >> > >> > >> wrote: >> > >> > Hi, >> > >> > >> > >> > I am recently quite confused about how hama splits input file. >> > >> > I ran hama pagerank with a very simple input file (with only 4 >> > vertices >> > >> and >> > >> > 6 edges). The file is split into 4 parts(while there are 3 tasks) >> and >> > the >> > >> > job failed. The bsp master log shows that : >> > >> > Scheduling of job pagerank could not be done successfully, killing >> it. >> > >> > And then zookeeper session timeout. >> > >> > The job then succeed dramatically after I change the vertices name >> of >> > the >> > >> > input file (call this file1), but fail again when I try to delete >> one >> > >> line >> > >> > from the file. Even when I changed the file to be exactly the same >> as >> > >> file1 >> > >> > but with a different name, the job still failed. >> > >> > Can someone tell me how does the splitting part work? I am really >> > >> confused. >> > >> > >> > >> > >> > >> > Best, >> > >> > Sandy >> > >> >> > >> >> > >> >> > >> -- >> > >> Best Regards, Edward J. Yoon >> > >> @eddieyoon >> > >> >> > >> > >> > >> > -- >> > Best Regards, Edward J. Yoon >> > @eddieyoon >> > >> -- Best Regards, Edward J. Yoon @eddieyoon
