[
https://issues.apache.org/jira/browse/HDFS-14403?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16869039#comment-16869039
]
Íñigo Goiri commented on HDFS-14403:
------------------------------------
I opened a JIRA for the DN to use this kind of fairness (HDFS-14589).
I wanted to double check that this was not already in the air.
>From the code in DataXceiverServer and the DataNode it doesn't look like this
>is done at all.
Regarding this JIRA itself, I have a couple minor comments:
* The location of {{applyWeights()}} in {{ProcessingDetails}} is a little
dangerous; should we make sure that weight and timing are the same size? I'm
not sure the {{applyWeights()}} naming is the most intuitive either. It looks
more like a {{getCost()}}. I would actually just move the full code to
{{WeightedTimeCostProvider}}. Not sure it adds much value to add it to
{{ProcessingDetails}}.
* Do we want to add a couple unit tests with corner cases? Like having no
requests, etc?
* Use logger format for the logs (using {}).
* Use lambda for the {{GenericTestUtils#waitFor()}}.
* I would add some high level text about the cost based approach not only
describing the fields but describing what the goal is and how one would use it
and set it up "end to end"-ish.
> Cost-Based RPC FairCallQueue
> ----------------------------
>
> Key: HDFS-14403
> URL: https://issues.apache.org/jira/browse/HDFS-14403
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: ipc, namenode
> Reporter: Erik Krogen
> Assignee: Christopher Gregorian
> Priority: Major
> Labels: qos, rpc
> Attachments: CostBasedFairCallQueueDesign_v0.pdf,
> HDFS-14403.001.patch, HDFS-14403.002.patch, HDFS-14403.003.patch,
> HDFS-14403.004.patch, HDFS-14403.005.patch, HDFS-14403.006.combined.patch,
> HDFS-14403.006.patch, HDFS-14403.007.patch, HDFS-14403.008.patch,
> HDFS-14403.009.patch, HDFS-14403.010.patch, HDFS-14403.011.patch,
> HDFS-14403.branch-2.8.patch
>
>
> HADOOP-15016 initially described extensions to the Hadoop FairCallQueue
> encompassing both cost-based analysis of incoming RPCs, as well as support
> for reservations of RPC capacity for system/platform users. This JIRA intends
> to track the former, as HADOOP-15016 was repurposed to more specifically
> focus on the reservation portion of the work.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
