[
https://issues.apache.org/jira/browse/KAFKA-479?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13439164#comment-13439164
]
Chris Riccomini commented on KAFKA-479:
---------------------------------------
As you can see from hprof:
CPU SAMPLES BEGIN (total = 6315) Tue Aug 21 16:45:05 2012
rank self accum count trace method
1 97.80% 97.80% 6176 300932 sun.nio.ch.EPollArrayWrapper.epollWait
2 0.08% 97.88% 5 301417
scala.util.parsing.json.Lexer$$anonfun$string$1$$anonfun$apply$13$$anonfun$apply$14.apply
3 0.06% 97.94% 4 301411 java.lang.Thread.currentThread
4 0.05% 97.99% 3 301433 sun.misc.Unsafe.getInt
5 0.03% 98.02% 2 301342 java.lang.Character.toString
6 0.03% 98.05% 2 301360 java.nio.HeapByteBuffer.<init>
7 0.03% 98.08% 2 301412 java.lang.Character.toString
8 0.03% 98.12% 2 301431 java.lang.Integer.toString
9 0.03% 98.15% 2 301446 sun.nio.ch.EPollArrayWrapper.interrupt
10 0.03% 98.18% 2 301454 sun.nio.ch.FileDispatcher.write0
11 0.03% 98.21% 2 301492 java.lang.Shutdown.halt0
12 0.02% 98.23% 1 300020 java.util.zip.ZipFile.read
13 0.02% 98.24% 1 300032 java.util.zip.ZipFile.open
14 0.02% 98.26% 1 300055 java.util.zip.ZipFile.open
15 0.02% 98.27% 1 300059 java.lang.StringCoding$StringEncoder.encode
And the trace is:
TRACE 300932:
sun.nio.ch.EPollArrayWrapper.epollWait(EPollArrayWrapper.java:Unknown line)
sun.nio.ch.EPollArrayWrapper.poll(EPollArrayWrapper.java:210)
sun.nio.ch.EPollSelectorImpl.doSelect(EPollSelectorImpl.java:65)
sun.nio.ch.SelectorImpl.lockAndDoSelect(SelectorImpl.java:69)
sun.nio.ch.SelectorImpl.select(SelectorImpl.java:80)
org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1066)
> ZK EPoll taking 100% CPU usage with Kafka Client
> ------------------------------------------------
>
> Key: KAFKA-479
> URL: https://issues.apache.org/jira/browse/KAFKA-479
> Project: Kafka
> Issue Type: Bug
> Environment: java 1.6.0_21
> Kafka 0.7.14
> Reporter: Chris Riccomini
> Priority: Blocker
> Attachments: java.hprof.txt, java.hprof.txt
>
>
> Hey Guys,
> I'm seeing a very strange bug after I upgraded to Kafka 0.7.14. On my
> consumer side, the process seems to run with very high CPU usage. I turned on
> HPROF, and it's showing that 75% of the CPU usage is going to Java's NIO
> EPoll/select calls.
> There are supposedly work arounds that Mina and a few others employ. This
> seems like more of a ZK bug- have you guys seen this before?
> Any ideas?
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
