Also, I know the swimlanes UI has been discussed but I couldn't see an answer in the mailing lists to what the red lines at the end of container lifecycles are (I see a bunch that say -100 in bold red print).
On Tue, Sep 16, 2014 at 10:55 PM, Thaddeus Diamond < [email protected]> wrote: > You know what, looks like it's because it's a Tez session that is not yet > complete. YARN can't give us application logs until the AM returns. Would > be good in the future to support during-session dumps so I don't have to > restart the whole session. Should I file a JIRA? > > On Tue, Sep 16, 2014 at 10:30 PM, Jonathan Eagles <[email protected]> > wrote: > >> Thad, the swimlanes tool may need some added robustness for the case >> above. Can you capture the yarn logs for this job to help in debugging? >> >> jeagles >> >> On Tue, Sep 16, 2014 at 8:46 PM, Thaddeus Diamond < >> [email protected]> wrote: >> >>> Hey guys, >>> >>> I'm trying to use the swimlanes tool (very nice!) but am getting an >>> error. I have Tez revision a1dd829 being run and when I run the swimlanes >>> bash script (added set -x) I get: >>> >>> [thaddeusdiamond@hadapt ~]$ ./yarn-swimlanes.sh >>> application_1410882226668_0008 >>> + APPID=application_1410882226668_0008 >>> ++ which yarn >>> + YARN=/usr/bin/yarn >>> ++ mktemp >>> + TMP=/tmp/tmp.kmPS10lCNt >>> + echo 'Fetching yarn logs for application_1410882226668_0008' >>> Fetching yarn logs for application_1410882226668_0008 >>> + /usr/bin/yarn logs -applicationId application_1410882226668_0008 >>> + grep HISTORY >>> + python swimlane.py -o application_1410882226668_0008.svg >>> /tmp/tmp.kmPS10lCNt >>> Traceback (most recent call last): >>> File "swimlane.py", line 201, in <module> >>> sys.exit(main(sys.argv[1:])) >>> File "swimlane.py", line 121, in main >>> log = AMLog(args[0]).structure() >>> File "/home/thaddeusdiamond/amlogparser.py", line 203, in structure >>> am.containers = containers >>> AttributeError: 'NoneType' object has no attribute 'containers' >>> >>> Thanks, >>> Thad >>> >> >> >
