[
https://issues.apache.org/jira/browse/YARN-7200?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17263356#comment-17263356
]
Szilard Nemeth commented on YARN-7200:
--------------------------------------
Hi [~akshink],
Thanks for the explanation, makes sense.
I also checked the code and I'm with the same opinion, I don't see a way in the
SLS framework that would allow delayed executions of certain AMs so the
scenario I described with [my comment
here|https://issues.apache.org/jira/browse/YARN-7200?focusedCommentId=17252048&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17252048]
is not possible in reality.
Some thoughts:
1. SLSCapacityScheduler / SLSFairScheduler:
The code block that is added is the same for both:
{code}
if (SLSRunner.getRemainingApps() == 0) {
try {
getSchedulerMetrics().tearDown();
SLSRunner.exitSLSRunner();
} catch (Exception e) {
e.printStackTrace();
}
}
{code}
I can see that there's no common parent for these classes, so don't spend time
with code deduplication here as there's another jira for that: YARN-10552.
However, invoking printStackTrace is not the best, as we want to leverage the
underlying logging system to print the exception details into the configured
outputs (files, console, anything).
Printstacktrace only prints it to the standard error, so this is a limitation.
Please use a LOG.error statement, add a message and pass the exception as an
argument.
2. org.apache.hadoop.yarn.sls.scheduler.SchedulerMetrics#tearDown
It's okay that metricsLogBW is set to null.
I can see 3 write calls of this field in
org.apache.hadoop.yarn.sls.scheduler.SchedulerMetrics.MetricsLogRunnable.
How it is guaranteed that these won't be invoked after teardown has been
executed?
Please fix these 2 issues and we're good to go.
> SLS generates a realtimetrack.json file but that file is missing the closing
> ']'
> --------------------------------------------------------------------------------
>
> Key: YARN-7200
> URL: https://issues.apache.org/jira/browse/YARN-7200
> Project: Hadoop YARN
> Issue Type: Bug
> Components: scheduler-load-simulator
> Reporter: Grant Sohn
> Assignee: Agshin Kazimli
> Priority: Minor
> Labels: newbie, newbie++
> Attachments: YARN-7200-branch-trunk.patch, YARN-7200.002.patch,
> YARN-7200.003.patch, YARN-7200.004.patch, snemeth-testing-20201113.zip
>
>
> File
> hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/SchedulerMetrics.java
> shows:
> {noformat}
> void tearDown() throws Exception {
> if (metricsLogBW != null) {
> metricsLogBW.write("]");
> metricsLogBW.close();
> }
> ....
> {noformat}
> So the exit logic is flawed.
--
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
