[sheepdog-users] zookeeper followers hungs
Andrew J. Hobbs
ajhobbs at desu.edu
Thu Apr 17 15:06:42 CEST 2014
I just noticed the interior exception. gnu.java.nio from libgcj.so. It
would be worthwhile running openjdk or oracle official and see if the
issue goes away, as it looks like it may be specific to gcj java.
This link is to a zookeeper mailing list that talks about this very issue.
So root cause actually appears to be use of gcj rather than a more
standard openjdk or official distribution.
On 04/16/2014 09:40 PM, Hitoshi Mitake wrote:
> On Sat, Apr 12, 2014 at 11:02 PM, Maxim Terletskiy <terletskiy at emu.ru> wrote:
>> I'm using three nodes zookeeper cluster and have problems with follower
>> nodes. They are working normal about 12-24 hours, after that process starts
>> to using 100% of one cpu core and stops to responding. Restart helps but not
>> for long. Does anyone have similar problem?
>> There are warnings in logs:
>> 2014-04-11 18:19:57,415 - WARN
>> [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181:NIOServerCnxnFactory at 218] - Ign
>> oring unexpected runtime exception
>> at gnu.java.nio.SelectionKeyImpl.readyOps(libgcj.so.10)
>> at java.lang.Thread.run(libgcj.so.10)
>> Maybe my config is wrong?
>> Zookeeper version is 3.4.5, OS CentOS 6.5 x86_64.
> Sorry, I cannot provide solution for it currently. Could you add an
> issue to our tracker?
>> sheepdog-users mailing lists
>> sheepdog-users at lists.wpkg.org
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 353 bytes
More information about the sheepdog-users