We have a test cluster running 0.8 that is not behaving properly. It is almost
continuously spewing the following exception into its log:
2013-03-07 23:44:17,532 ERROR kafka.network.Processor: Closing socket for
/10.10.2.123 because of error
java.io.IOException: Resource temporarily unavailabl
Bob,
It seems you are probably reaching the limit of open files on that box. In
Kafka 0.8, we keep the file handles for all segment files open until they
are garbage collection. Depending on the size of your cluster, this number
can be pretty big. Few 10 K or so.
Thanks,
Neha
On Thu, Mar 7, 201
Thanks Neha,
So are you saying that on 0.7.2, to delete a topic I need only remove it's
data log directory from each broker, and the restart the brokers? Is it ok
if it's a rolling restart?
For some reason I thought I also had to do something on zookeeper servers
too, to remove the topics. Is t
Could this be caused by this bug (
http://bugs.sun.com/view_bug.do?bug_id=5103988) in java 5?
Thanks,
Jun
On Thu, Mar 7, 2013 at 3:57 PM, Bob Jervis
wrote:
> We have a test cluster running 0.8 that is not behaving properly. It is
> almost continuously spewing the following exception into its l
In Kafka 0.7.2, data log directory delete and then rolling restart will
work to get rid of the topic completely from zookeeper as well as the
broker.
This is because the broker registers the topics and partitions from its
data directory into zookeeper on startup.
Thanks,
Neha
On Thu, Mar 7, 2013
