Turns out there is a bug in the JDK:

https://issues.apache.org/jira/browse/ZOOKEEPER-427?focusedCommentId=12716020&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#action_12716020

We are looking to see if we can solve this via some workaround in 3.2. Currently the only solution is to restart the server that experiences this problem.

Patrick

Patrick Hunt wrote:
According to your trace I see you are using jvm 1.6.0_03-b05

One of the bugs fixed in:
http://java.sun.com/javase/6/webnotes/6u4.html
specifically:
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6403933
seems to have a description very close to what you are seeing.

Perhaps you can try running with the latest (or as recent as you can get) jvm?

Patrick

Satish Bhatti wrote:
stat
Zookeeper version: 3.1.1-755636, built on 03/18/2009 16:52 GMT
Clients:
 /127.0.0.1:42460[1](queued=0,recved=0,sent=0)
 /172.16.0.178:34283[1](queued=0,recved=925109,sent=925109)

Latency min/avg/max: 0/0/220
Received: 925109
Sent: 925109
Outstanding: 0
Zxid: 0x50051c7b3
Mode: follower
Node count: 11
Connection closed by foreign host.


On Tue, Jun 2, 2009 at 1:39 PM, Ted Dunning <ted.dunn...@gmail.com> wrote:

I am sooo glad I put in the fourth item. This is clearly not overload in
that sense.

Do you have logs from the ZK?

What does the stat command return (telnet to the ZK inquestion, type "stat"
(without the quotes)).

On Tue, Jun 2, 2009 at 1:35 PM, Satish Bhatti <cthd2...@gmail.com> wrote:

I have a grand total of 3 files, each holding a single long!  Could it
really be in gc hell because of this?  I passed in -Xmx256m on the
command
line.



--
Ted Dunning, CTO
DeepDyve


Reply via email to