hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Billie Rinaldi (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-1635) Implement a Leveldb based ApplicationTimelineStore
Date Thu, 06 Feb 2014 22:31:20 GMT

    [ https://issues.apache.org/jira/browse/YARN-1635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13893906#comment-13893906
] 

Billie Rinaldi commented on YARN-1635:
--------------------------------------

bq. 1. Change ATSPutError -> ATSPutErrorCode? Otherwise, the name is duplicate with ATSPutError
POJO.
I don't understand.  The ATSPutError in the patch is the POJO.

bq. 3. "atsEntity.setEvents(new ArrayList<ATSEvent>());" is not necessary, right? Same
for the followup invoke of collection setters.
That is true, but I am hoping that we switch the POJOs to default the objects to null, and
this code works no matter what the defaults are.  But I'll change it if I must.

bq. 4. Is it better to combine two conditions as "primaryFilters && key[prefixlen]
== PRIMARY_FILTER_COLUMN[0]", and check boolean flag first? Same for the following conditions.
I don't think so.  If they are combined, then it will keep checking the other conditions in
the if/else, when it should just continue to the next key/value pair because this k/v contains
data that is not being requested.

bq. 5. Does CurrentMap work here? Then, we can reduce the synchronization.
I'm not sure what CurrentMap does.  LRUMap is not thread safe, so I was getting ConcurrentModificationException
in the LeveldbApplicationTimelineStore.getStartTime method when I didn't have it wrapped in
a synchronized map.

bq. 7. Catch IOException only.
Should we handle runtime exceptions somewhere?  Maybe in ATSWebServices?

> Implement a Leveldb based ApplicationTimelineStore
> --------------------------------------------------
>
>                 Key: YARN-1635
>                 URL: https://issues.apache.org/jira/browse/YARN-1635
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Vinod Kumar Vavilapalli
>            Assignee: Billie Rinaldi
>         Attachments: YARN-1635.1.patch, YARN-1635.10.patch, YARN-1635.2.patch, YARN-1635.3.patch,
YARN-1635.4.patch, YARN-1635.5.patch, YARN-1635.6.patch, YARN-1635.7.patch, YARN-1635.8.patch,
YARN-1635.9.patch
>
>
> As per the design doc, we need a levelDB + local-filesystem based implementation to start
with and for small deployments.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

Mime
View raw message