You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hadoop.apache.org by nagarjuna kanamarlapudi <na...@gmail.com> on 2012/11/20 07:33:53 UTC
Secondary JobTracker
After working on hadoop-0.20 for some time. I was just wondering why we
don't have seondary jobtracker just like secondary name node.
Can we come up with some reasonalbe answers to this ?
Re: Secondary JobTracker
Posted by Harsh J <ha...@cloudera.com>.
The function of a Secondary NameNode is to take checkpoints, not
failover. If you are looking for HA JobTracker (not yet available) or
Recoverable JobTracker functionality (already present today IIRC),
look at the parent JIRA
https://issues.apache.org/jira/browse/MAPREDUCE-2288 for some ideas
that were floated around.
In any case, JobTracker is an MR1 component and does not exist anymore
in the 2.x releases, and the primary goal right now is to have the
YARN's RM go HA, as tracked by
https://issues.apache.org/jira/browse/YARN-149 and associated JIRAs.
On Tue, Nov 20, 2012 at 12:03 PM, nagarjuna kanamarlapudi
<na...@gmail.com> wrote:
> After working on hadoop-0.20 for some time. I was just wondering why we
> don't have seondary jobtracker just like secondary name node.
>
> Can we come up with some reasonalbe answers to this ?
--
Harsh J
Re: Secondary JobTracker
Posted by Harsh J <ha...@cloudera.com>.
The function of a Secondary NameNode is to take checkpoints, not
failover. If you are looking for HA JobTracker (not yet available) or
Recoverable JobTracker functionality (already present today IIRC),
look at the parent JIRA
https://issues.apache.org/jira/browse/MAPREDUCE-2288 for some ideas
that were floated around.
In any case, JobTracker is an MR1 component and does not exist anymore
in the 2.x releases, and the primary goal right now is to have the
YARN's RM go HA, as tracked by
https://issues.apache.org/jira/browse/YARN-149 and associated JIRAs.
On Tue, Nov 20, 2012 at 12:03 PM, nagarjuna kanamarlapudi
<na...@gmail.com> wrote:
> After working on hadoop-0.20 for some time. I was just wondering why we
> don't have seondary jobtracker just like secondary name node.
>
> Can we come up with some reasonalbe answers to this ?
--
Harsh J
Re: Secondary JobTracker
Posted by Harsh J <ha...@cloudera.com>.
The function of a Secondary NameNode is to take checkpoints, not
failover. If you are looking for HA JobTracker (not yet available) or
Recoverable JobTracker functionality (already present today IIRC),
look at the parent JIRA
https://issues.apache.org/jira/browse/MAPREDUCE-2288 for some ideas
that were floated around.
In any case, JobTracker is an MR1 component and does not exist anymore
in the 2.x releases, and the primary goal right now is to have the
YARN's RM go HA, as tracked by
https://issues.apache.org/jira/browse/YARN-149 and associated JIRAs.
On Tue, Nov 20, 2012 at 12:03 PM, nagarjuna kanamarlapudi
<na...@gmail.com> wrote:
> After working on hadoop-0.20 for some time. I was just wondering why we
> don't have seondary jobtracker just like secondary name node.
>
> Can we come up with some reasonalbe answers to this ?
--
Harsh J
Re: Secondary JobTracker
Posted by Harsh J <ha...@cloudera.com>.
The function of a Secondary NameNode is to take checkpoints, not
failover. If you are looking for HA JobTracker (not yet available) or
Recoverable JobTracker functionality (already present today IIRC),
look at the parent JIRA
https://issues.apache.org/jira/browse/MAPREDUCE-2288 for some ideas
that were floated around.
In any case, JobTracker is an MR1 component and does not exist anymore
in the 2.x releases, and the primary goal right now is to have the
YARN's RM go HA, as tracked by
https://issues.apache.org/jira/browse/YARN-149 and associated JIRAs.
On Tue, Nov 20, 2012 at 12:03 PM, nagarjuna kanamarlapudi
<na...@gmail.com> wrote:
> After working on hadoop-0.20 for some time. I was just wondering why we
> don't have seondary jobtracker just like secondary name node.
>
> Can we come up with some reasonalbe answers to this ?
--
Harsh J