[
https://issues.apache.org/jira/browse/HADOOP-4576?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12653209#action_12653209
]
Amar Kamat commented on HADOOP-4576:
------------------------------------
Comments :
{{JobQueuesManager.java}} :
1) No need to import {{AtomicInteger}}. Also plz remove extra diffs.
{{CapacityTaskScheduler.java}} :
1) You cant simply change the constructor def. Overload it and deprecate the
other if needed.
2) Extra diffs w.r.t.
{code}
- supportsPriority?"YES":"NO"));
+ supportsPriority?"YES":"NO"));
{code}
3) {quote}
+ sb.append(String.format("* Scheduling information can be off by " +
+ "maximum of %d seconds\n", pollingInterval/1000));
{quote}
Use {{StringUtils}} for formatting time.
{{TestCapacityScheduler.java}}
1) {quote}
+ scheduler.assignTasks(tracker("tt1")); // heartbeat
+ p.selectJobsToInitialize();
{quote}
The ordering doesnt seem right. You submit 5 jobs, try to assign tasks (which
should be no-op) and then you init the jobs.
2) Shouldnt we also check/test the timing issue that after _poll-interval_
units of time the values are correct. Something like
- add jobs
- check the queue-sched-info
- allow the jobs to be inited by the poller i.e wait for _poller-interval_
time
- check again to see if the change is made.
2) Plz mark the start and end of a new sub-test using comments
> Modify pending tasks count in the UI to pending jobs count in the UI
> --------------------------------------------------------------------
>
> Key: HADOOP-4576
> URL: https://issues.apache.org/jira/browse/HADOOP-4576
> Project: Hadoop Core
> Issue Type: Bug
> Components: contrib/capacity-sched
> Affects Versions: 0.19.0
> Reporter: Hemanth Yamijala
> Assignee: Sreekanth Ramakrishnan
> Attachments: HADOOP-4576-1.patch, HADOOP-4576-2.patch,
> HADOOP-4576-3.patch
>
>
> The UI for capacity scheduler displays 'pending tasks' counts. However the
> capacity scheduler does not update these counts to be the actual values for
> optimization purposes, for e.g. to avoid walking all pending jobs on all
> heartbeats. Hence this information is not very accurate.
> Also, while 'running tasks' counts are useful to compare against capacities
> and limits, 'pending tasks' counts do not add too much user value. A better
> count to display would be the number of running and pending jobs.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
