Filed KAFKA-1062, including trivial patch.
/Sam
On Sep 19, 2013, at 5:52 PM, Neha Narkhede wrote:
> Agreed. This is a regression and is not easy to reason about. This is a
> side effect of reading the partitions as a set from zookeeper. Please can
> you file a JIRA to get this fixed? Feel free
Hello Sam,
I agree that with the fix we still should sort the partition list before
hand it to the assignment algorithm. I will try to make a follow-up patch
to fix this.
Guozhang
On Thu, Sep 19, 2013 at 8:17 AM, Sam Meder wrote:
> The latest consumer changes to read data from Zookeeper during
Agreed. This is a regression and is not easy to reason about. This is a
side effect of reading the partitions as a set from zookeeper. Please can
you file a JIRA to get this fixed? Feel free to upload a patch as well.
Thanks,
Neha
On Thu, Sep 19, 2013 at 8:17 AM, Sam Meder wrote:
> The latest c
The latest consumer changes to read data from Zookeeper during rebalance have
made the consumer rebalance code incompatible with older versions (making
rolling upgrades without downtime hard). The problem relates to how partitions
are ordered. The old code seems to have returned the partitions s
