[
https://issues.apache.org/jira/browse/HADOOP-11361?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15365687#comment-15365687
]
Vinayakumar B commented on HADOOP-11361:
----------------------------------------
bq. If one thread enters the above method with Time.now() - jmxCacheTS >=
jmxCacheTTL being false, then getAllMetrics will continue to be false, then
when updateAttrCache() is called, it will hit NULL lastRecs
>From your earlier comment linked, if {{Time.now() - jmxCacheTS >=
>jmxCacheTTL}} is false, doesnt it return from there itself?
bq. It looks to me that it's actually not a race condition issue (otherwise
please would you please explain what race condition you are seeing),
Race condition is there between two threads calling {{updateJmxCache()}} at
same time.
{code} if (getAllMetrics) {
MetricsCollectorImpl builder = new MetricsCollectorImpl();
getMetrics(builder, true);
}
synchronized(this) {
updateAttrCache();
if (getAllMetrics) {
updateInfoCache();
}
jmxCacheTS = Time.now();
lastRecs = null; // in case regular interval update is not running
lastRecsCleared = true;
}{code}
1. Consider both threads t1 and t2 waiting at the entrance of
{{synchronized(this) {}}, i.e. line no : 183.
2. t1 gets inside and makes the {{lastRecs = null}} and {{lastRecsCleared=true}}
3. now since t2 already passed the line of reloading the {{lastRecs}} in
{{getMetrics(..)}} line, t2 will get NPE in {{updateAttrCache()}}.
bq. Given that the code is doing lastRecsCleared = true;, the comment "Not
making LastRecs null ..." is a bit contradictory.
To make it less contradictory, how about renaming {{lastRecsCleared}} to
something, {{reloadRecs}}?
bq. . So seems to me that the solution would be to fix
MetricsSourceAdapter#updateAttrCache to let it check whether lastRecs is NULL,
and not to access it if so.
This also works, since the {{lastRecs}} is used only to update {{attrCache}}
and {{infoCache}}, avoiding updating these in case of {{lastRecs == null}} will
not be a problem.
> Fix a race condition in MetricsSourceAdapter.updateJmxCache
> -----------------------------------------------------------
>
> Key: HADOOP-11361
> URL: https://issues.apache.org/jira/browse/HADOOP-11361
> Project: Hadoop Common
> Issue Type: Bug
> Affects Versions: 2.4.1, 2.5.1, 2.6.0
> Reporter: Brahma Reddy Battula
> Assignee: Brahma Reddy Battula
> Attachments: HADOOP-111361-003.patch, HADOOP-11361-002.patch,
> HADOOP-11361-004.patch, HADOOP-11361.patch, HDFS-7487.patch
>
>
> {noformat}
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.metrics2.impl.MetricsSourceAdapter.updateAttrCache(MetricsSourceAdapter.java:247)
> at
> org.apache.hadoop.metrics2.impl.MetricsSourceAdapter.updateJmxCache(MetricsSourceAdapter.java:177)
> at
> org.apache.hadoop.metrics2.impl.MetricsSourceAdapter.getAttribute(MetricsSourceAdapter.java:102)
> at
> com.sun.jmx.interceptor.DefaultMBeanServerInterceptor.getAttribute(DefaultMBeanServerInterceptor.java:647)
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
