[
https://issues.apache.org/jira/browse/FALCON-991?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14572084#comment-14572084
]
Srikanth Sundarrajan commented on FALCON-991:
---------------------------------------------
[~ajayyadava]
{code}
+ coordJob = client.getCoordJobInfo(coord.getId(),
"nominaltime>=" + start + ";nominaltime<" + end,
+ 0, Integer.MAX_VALUE);
{code}
Would retrieving all the coord actions for a coord cause OOM within Falcon with
this change ? From the looks of it, this wont be any worser than what it was
before this. Please confirm
Also can the exception be added to LOG statement ?
{code}
+ LOG.error("Unable to getcoordinator job for:{} for
start:{} and end: {}", coord.getId(),
+ start, end);
{code}
should perhaps be
{code}
+ LOG.error("Unable to getcoordinator job for:{} for
start:{} and end: {}", coord.getId(),
+ start, end, e);
{code}
> consolidate get coord actions in OozieWorkflowEngine
> ----------------------------------------------------
>
> Key: FALCON-991
> URL: https://issues.apache.org/jira/browse/FALCON-991
> Project: Falcon
> Issue Type: Improvement
> Reporter: Shwetha G S
> Assignee: Ajay Yadava
> Attachments: FALCON-991-v1.patch, FALCON-991-v2.patch,
> FALCON-991.patch
>
>
> OozieWorkflowEngine.getCoordActions() gets instances by calling
> oozieclient.getCoordAction() for each action. Its not efficient to call oozie
> API for each instance. Instead, doSummaryJobAction() gets all instances in
> bulk using oozieclient.getCoordJobInfo(). But this uses offset which won't
> work with purging of coord actions. A new API was added in oozie to get coord
> actions between timestamps. We should use that for both usecases
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
