[
https://issues.apache.org/jira/browse/YARN-6064?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15805717#comment-15805717
]
Varun Saxena commented on YARN-6064:
------------------------------------
Thanks [~rohithsharma] for the patch. Had a glance at the patch. Few comments.
# We are querying flow context while querying flow apps. Is extra peek in app
to flow table required ? Cant we supply flow run ID as well from client ?
# I think the tests require APPLICATION_CREATED_EVENT to be passed for an entry
in app to flow table to be made. We are relying on it in code.
# Are changes in GenericEntityReader required ?
# In ApplicationEntityReader, below piece of code has been added. Here, we
check for getFilters() being null and then inside if block dereference filters
by calling getFilters().getFromId(). What if getFilters() is null ? It may not
be null in code flow (havent checked) but this should still be fixed. I do not
think we need to pass fromId to ApplicationRowKeyPrefix in this case.
{code}
369 if (getFilters() == null || getFilters().getFromId() == null) {
370 applicationRowKeyPrefix = new ApplicationRowKeyPrefix(
371 context.getClusterId(), context.getUserId(),
context.getFlowName(),
372 context.getFlowRunId(), getFilters().getFromId());
{code}
An observation : fromId is inclusive as start row in HBase is inclusive. Let us
say for pagination case I first get app100-app91. Then from UI side you will
send fromId as app91 or you expect at the client side to find out the next app
as app90 and query ? Or client needs to drop the extra record ? We need to make
this clear in the documentation whenever we write it.
I will invoke the build after committing YARN-5585 and do a more detailed
review after QA report.
> Support fromId for flowRuns and flow/flowRun apps REST API's
> ------------------------------------------------------------
>
> Key: YARN-6064
> URL: https://issues.apache.org/jira/browse/YARN-6064
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: timelinereader
> Reporter: Rohith Sharma K S
> Assignee: Rohith Sharma K S
> Labels: yarn-5355-merge-blocker
> Attachments: YARN-6064-YARN-5355.0001.patch
>
>
> Splitting out JIRA YARN-6027 for pagination support for flowRuns, flow apps
> and flow run apps.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
