[
https://issues.apache.org/jira/browse/HADOOP-15901?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16676223#comment-16676223
]
Vinayakumar B commented on HADOOP-15901:
----------------------------------------
Thanks [[email protected]] for the pointers.
Yes! its depressing to see this. Previously I me only changed usage of
{{Time.now()}} to {{Time.monotonicNow()}} in many places. Should I go back now?
JDK guys says in this
[JDKBug|https://bugs.java.com/bugdatabase/view_bug.do?bug_id=6458294] that
monotocity of {{System.nanoTime()}} implementation depends on underlying
platform. If the underlying platform supports MONOTONIC_CLOCK then
{{System.nanoTime()}} should be monotonic, otherwise results will be same as
{{System.currentTimeMillis()}}.
Many blogs pointed out problem in {{Windows XP with SP2}} due to TSC counter
(core wise counter). SP3 onwards, power management timer (pmtimer) was enabled,
which solved this problem.
Even though with problematic {{System.nanoTime()}}, one of the answers it was
mentioned that difference can be ~ +/- 3ms.
But with {{System.currentTimeMillis()}} this could be even minutes when NTP is
running. This could lead to serious miscalculations.
This [blog|http://btorpey.github.io/blog/2014/02/18/clock-sources-in-linux/]
was really informative.
So what we should do? Remove all usages of {{System.nanoTime()}} in Hadoop?
> IPC Client and Server should use Time.monotonicNow() for elapsed times.
> -----------------------------------------------------------------------
>
> Key: HADOOP-15901
> URL: https://issues.apache.org/jira/browse/HADOOP-15901
> Project: Hadoop Common
> Issue Type: Bug
> Components: ipc, metrics
> Reporter: Vinayakumar B
> Assignee: Vinayakumar B
> Priority: Major
> Attachments: HADOOP-15901-01.patch
>
>
> Client.java and Server.java uses {{Time.now()}} to calculate the elapsed
> times/timeouts. This could result in undesired results when system clock's
> time changes.
> {{Time.monotonicNow()}} should be used for elapsed time calculations within
> same JVM.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
