[
https://issues.apache.org/jira/browse/HADOOP-4035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12642861#action_12642861
]
Vivek Ratan commented on HADOOP-4035:
-------------------------------------
Here's a summary of what this patch should be doing.
TTs report the amount of free memory available on their node (as described in
HADOOP-3759 and HADOOP-4439), which is equal to the total VM assigned for
Hadoop tasks on that node (_mapred.tasktracker.tasks.maxmemory_) minus the VM
guaranteed to the tasks that are already running. The CapacityScheduler looks
at this free memory to decide if it can run a task, which has its own memory
needs. If the task requires more memory than is available on the TT, the
Scheduler returns nothing to the TT (thus forcing it to finish up what it is
running and eventually having enough free memory).
In order to make sure that no job asks for memory that is more than what a TT
has available, we should have a cluster-wide limit on the amount of VM a job
can ask for its tasks. If this limit is set, and a job asks for too much, the
job should not be accepted by the JT in submitJob(). If the limit is not set,
jobs cannot be rejected based on their memory requirements.
> Modify the capacity scheduler (HADOOP-3445) to schedule tasks based on memory
> requirements and task trackers free memory
> ------------------------------------------------------------------------------------------------------------------------
>
> Key: HADOOP-4035
> URL: https://issues.apache.org/jira/browse/HADOOP-4035
> Project: Hadoop Core
> Issue Type: Bug
> Components: contrib/capacity-sched
> Affects Versions: 0.19.0
> Reporter: Hemanth Yamijala
> Assignee: Vinod K V
> Priority: Blocker
> Fix For: 0.20.0
>
> Attachments: 4035.1.patch, HADOOP-4035-20080918.1.txt,
> HADOOP-4035-20081006.1.txt, HADOOP-4035-20081006.txt, HADOOP-4035-20081008.txt
>
>
> HADOOP-3759 introduced configuration variables that can be used to specify
> memory requirements for jobs, and also modified the tasktrackers to report
> their free memory. The capacity scheduler in HADOOP-3445 should schedule
> tasks based on these parameters. A task that is scheduled on a TT that uses
> more than the default amount of memory per slot can be viewed as effectively
> using more than one slot, as it would decrease the amount of free memory on
> the TT by more than the default amount while it runs. The scheduler should
> make the used capacity account for this additional usage while enforcing
> limits, etc.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
