hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hemanth Yamijala (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HADOOP-3842) There is a window where the JobTracker is in the RUNNING state (i.e ready to accept jobs) and never executes them.
Date Mon, 13 Oct 2008 17:36:45 GMT

     [ https://issues.apache.org/jira/browse/HADOOP-3842?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Hemanth Yamijala updated HADOOP-3842:
-------------------------------------

         Priority: Major  (was: Blocker)
    Fix Version/s:     (was: 0.19.0)
                   0.20.0

bq. Prior to HADOOP-3412, job tracker could accept jobs without even offering service (i.e
without JobTracker.offerService() being called). 

I see that in the latest code, job tracker calls {{TaskScheduler.start}} as the first statement
in {{offerService}} and the {{interTrackerServer}} is started after that. I also checked that
all schedulers currently implemented add their listeners to the JobTracker in the {{start}}
method itself. Hence, currently, this issue doesn't exist.

I am moving this out of the Hadoop 0.19 blocker list based on this, but keeping the issue
open - as it is possible that the schedulers don't add their listeners at startup theoretically.
I think this fact needs to be made more explicit and the contract well defined.

But for now, things do work fine. Amar, if this diagnosis is incorrect, can you please explain
and mark it back as a blocker ?

> There is a window where the JobTracker is in the RUNNING state (i.e ready to accept jobs)
and never executes them.
> ------------------------------------------------------------------------------------------------------------------
>
>                 Key: HADOOP-3842
>                 URL: https://issues.apache.org/jira/browse/HADOOP-3842
>             Project: Hadoop Core
>          Issue Type: Bug
>          Components: mapred
>            Reporter: Amar Kamat
>            Assignee: Hemanth Yamijala
>             Fix For: 0.20.0
>
>
> Prior to HADOOP-3412, job tracker could accept jobs without even offering service (i.e
without {{JobTracker.offerService()}} being called). In such a case the job stays in JT's
memory and job execution was guaranteed. With HADOOP-3412, {{JobTracker.submitJob()}} adds
the job to JT's local structures and passes it to the scheduler. Scheduler gets initialized
in {{JobTracker.offerService()}} and hence calling {{JobTracker.submitJob()}} before calling
{{JobTracker.offerService()}} is actually a no-op. The job stays in JT's memory but never
gets initialized. This is 
> - backward incompatible 
> - erroneous as there is a window where the jobtracker is ready to accept jobs, accepts
them and never executes them.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message