That I don't know Jun. I dug a little into the Ruby source yesterday, and it appeared that we might be affected by the ruby client issuing a topic metadata request explicitly. I was concerned that we were triggering the behaviour seen in https://issues.apache.org/jira/browse/KAFKA-1507 However, I'll check your suggestion on the ZK bypass.
On 23 February 2015 at 17:32, Jun Rao <[email protected]> wrote: > Does the ruby library write to ZK directly to create topics? That will > bypass the checking on the broker side. > > Thanks, > > Jun > > On Mon, Feb 23, 2015 at 3:06 AM, David Corley <[email protected]> > wrote: > > > Hey all, > > I'm trying to run some basic error-handling validation with some client > > code, and I'm attempting to handle an UnknownTopicOrPartitionException. > To > > set this scenario up, I wanted to attempt fetch messages from a topic I > > know doesn't exist. To that end, I've got a 3-broker cluster with: > > * auto.create.topics.enable=false > > * each broker restarted after setting the above > > > > The logs for each broker show the following after the restart: > > [2015-02-23 10:13:30,315] INFO Property auto.create.topics.enable is > > overridden to false (kafka.utils.VerifiableProperties) > > > > However, every time I try to fetch messages from the broker, I get no > > exception, and kafka-list-topics shows the previously non-existent topic > is > > now created! > > > > I'm using poseidon/poseidon-cluster ruby libraries for the comms, but > don't > > believe they should be over-riding the broker configuration in terms of > > topic creation. Nothing in source code suggests anything, and I don't > > believe the protocol allows for it anyway. > > > > We're using Kafka 0.8.0 final for all brokers. Anyone come across this > > issue before? > > > > /Dave > > >
