Hi Aniket, I need to create a JIRA and upload a patch for review to fix that.
New improvements were made for important use cases that replaced the default
which I will move to another partitioner so folks can use it and revert the
original restoring expected default out of the box behavior.
Hi,
We have a blocking async producer.
We noticed:
1. when the metadata brokers are down, the client no longer attempts to
reconnect to any of them.
2. zookeeper.connect is no longer recognized as a valid configuration,
hinting that in fact the zookeeper balancing is no longer supported on
Yes, the zk option is no longer available in the producer in 0.8 to
simplify the logic in the client. Load balancing in the producer is still
supported though.
I am not sure what you mean by "blocking async" producer. The producer has
an async mode, but then the send() calls are not blocking. The
Hi,
Thank you for answer.
How is the balancing done without zookeeper? Is it done locally to
producer thread with round robin, or do we need to use a VIP with a TCP
balancer for each partition, with the hosts having that partition
replicas, or by other means?
Indeed, the producers reconnect, and
The producer in 0.8 doesn't depend on a VIP to connect to the brokers.
Instead, it obtains metadata from a list of brokers (or a VIP in front of
the brokers) and finds out the host/port of the broker for the leader
replica of each partition in a topic. It then establishes socket
connections to thos
No, this is all within the same DC. I think the problem has to do with the LB.
We've upgraded our producers to point directory to a node for testing and after
running it all night, I don't see any more connections then there are supposed
to be.
Can I ask which LB are you using? We are using A1
