[
https://issues.apache.org/jira/browse/YARN-6492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17100369#comment-17100369
]
Jonathan Hung edited comment on YARN-6492 at 5/6/20, 1:14 AM:
--------------------------------------------------------------
OK thanks [~maniraj...@gmail.com] for the explanation. Sorry for the long
delay, took some time to grok the latest 007 patch.
* Can we rename getPartitionQueueMetrics to something different? My initial
confusion was that getPartitionQueueMetrics for QueueMetrics and
PartitionQueueMetrics serve different purposes...the former for queue*partition
and the latter for partition only. It's especially confusing in the case of
PartitionQueueMetrics#getPartitionQueueMetrics, since this has nothing to do
with queues. We can update the comment for
PartitionQueueMetrics#getPartitionQueueMetrics as well, it also says Partition
* Queue.
* Mentioned this earlier, can we remove the {noformat} if (parent != null) {
parent.setAvailableResourcesToUser(partition, user, limit);
}{noformat}
check in QueueMetrics#setAvailableResourcesToUser? I think it should be
addressed here rather than YARN-9767.
* I don't think the asserts in TestNodeLabelContainerAllocation should change.
leafQueue.getMetrics should return metrics for default partition. I think we
still need to check in QueueMetrics#setAvailableResourcesToUser and
QueueMetrics#setAvailableResourcesToQueue whether partition is null or empty
string. (This will break updating partition queue metrics, so we need to find a
way to distinguish whether we're updating default partition queue metrics or
partitioned queue metrics.)
* Mentioned before, can we update everywhere we're creating a new metricName
for partition/user/queue metrics to use a delimiter? e.g. {noformat}String
metricName = partition + this.queueName + userName;{noformat}. Otherwise
there's a chance that these metric names could collide.
was (Author: jhung):
OK thanks [~maniraj...@gmail.com] for the explanation. Sorry for the long
delay, took some time to grok the latest 007 patch.
* Can we rename getPartitionQueueMetrics to something different? My initial
confusion was that getPartitionQueueMetrics for QueueMetrics and
PartitionQueueMetrics serve different purposes...the former for queue*partition
and the latter for partition only. It's especially confusing in the case of
PartitionQueueMetrics#getPartitionQueueMetrics, since this has nothing to do
with queues. We can update the comment for
PartitionQueueMetrics#getPartitionQueueMetrics as well, it also says Partition
* Queue.
* Mentioned this earlier, can we remove the {noformat} if (parent != null) {
parent.setAvailableResourcesToUser(partition, user, limit);
}{noformat}
check in QueueMetrics#setAvailableResourcesToUser? I think it should be
addressed here rather than YARN-9767.
* I don't think the asserts in TestNodeLabelContainerAllocation should change.
leafQueue.getMetrics should return metrics for default partition. I think we
still need to check in QueueMetrics#setAvailableResourcesToUser and
QueueMetrics#setAvailableResour cesToQueue whether partition is null or empty
string. (This will break updating partition queue metrics, so we need to find a
way to distinguish whether we're updating default partition queue metrics or
partitioned queue metrics.)
* Mentioned before, can we update everywhere we're creating a new metricName
for partition/user/queue metrics to use a delimiter? e.g. {noformat}String
metricName = partition + this.queueName + userName;{noformat}. Otherwise
there's a chance that these metric names could collide.
> Generate queue metrics for each partition
> -----------------------------------------
>
> Key: YARN-6492
> URL: https://issues.apache.org/jira/browse/YARN-6492
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: capacity scheduler
> Reporter: Jonathan Hung
> Assignee: Manikandan R
> Priority: Major
> Attachments: PartitionQueueMetrics_default_partition.txt,
> PartitionQueueMetrics_x_partition.txt, PartitionQueueMetrics_y_partition.txt,
> YARN-6492.001.patch, YARN-6492.002.patch, YARN-6492.003.patch,
> YARN-6492.004.patch, YARN-6492.005.WIP.patch, YARN-6492.006.WIP.patch,
> YARN-6492.007.WIP.patch, partition_metrics.txt
>
>
> We are interested in having queue metrics for all partitions. Right now each
> queue has one QueueMetrics object which captures metrics either in default
> partition or across all partitions. (After YARN-6467 it will be in default
> partition)
> But having the partition metrics would be very useful.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
