Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Zookeeper >> mail # user >> Unexpected NodeCreated event after a reconnection.


Copy link to this message
-
Unexpected NodeCreated event after a reconnection.
Hi all,

Today I've observed a wrong behavior on our production environment that I
would like to share with you because I think that it is wrong.

One "zookeeper server" was consuming about 25% of CPU and its log only shows
lines like the following (as if it is was on some kind of loop):

2011-03-10 00:00:00,000 - WARN  [NIOServerCxn.Factory:
0.0.0.0/0.0.0.0:2301:NIOServerCnxn$Factory@272][] - Ignoring unexpected
runtime exception
java.lang.NullPointerException

After its restart, two "zookeeper client" instances received a Disconnected
followed by a SyncConnected and several "NodeCreated" when those clients
never set a "create" watcher since they only invoke getData o getChildren
methods. That's why I think there is something wrong.

Furthermore after a content modification all the "zookeeper clients"
received a NodeDataChanged except for those which where connected previously
to restarted server.

The environment is: Solaris 10 + Zookeeper 3.3.2 + J6u17 64 bits.

Please let me known if there is some kind of information that I could
provide in order to perform some follow-up.

Br,
/César.