[
https://issues.apache.org/jira/browse/INLONG-590?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ASF GitHub Bot updated INLONG-590:
----------------------------------
Labels: pull-request-available (was: )
> Optimize the implementation of DefaultLoadBalancer class
> --------------------------------------------------------
>
> Key: INLONG-590
> URL: https://issues.apache.org/jira/browse/INLONG-590
> Project: Apache InLong
> Issue Type: Improvement
> Components: inlong-tubemq
> Reporter: Guocheng Zhang
> Assignee: bigdaronlee132
> Priority: High
> Labels: pull-request-available
> Time Spent: 10m
> Remaining Estimate: 0h
>
>
> TubeMQ uses the server-side rebalance mode to allocate tasks. At present, the
> task allocation mode is relatively single, and the implementation of this
> part is relatively hard-coded, and the readability and maintainability are
> not good enough.
> Existing solutions take all the partitions of the topic set subscribed by the
> consumer group according to the number of consumers in the consumer group,
> and then extract the partitions at consumer number interval, and allocate the
> partitions to the corresponding client; if the client or partition has
> increasing or decreasing, the client with more partitions subscriptions will
> release the partitions through partial balance, and will be allocated the
> released partitions to consumers with less or no subscription partitions for
> balanced consumption.
> If you are interested, you can try to abstract this piece. At the same time,
> you can also propose your rebalance plan, which is compatible on the basis of
> existing rebalance.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
