Status

Current state: Draft [One of "Under Discussion", "Accepted", "Rejected"]

Discussion thread: here [Change the link from the KIP proposal email archive to your own email thread]

JIRA: NA

Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).

...

This KIP aims to improve the procedure described in Motivation above, making it more robust to hardware failure. The new procedure should avoid generating new broker id except when bringing up a new cluster, while maintaining good characteristic of old procedure. Let’s take a closer look what information is available:

...

Note step 1 is expensive as it needs to read a lot of data from ZooKeeper (~1s for 1k topics with 1ms each). An alternative approach could also be used but is more complicated to implement.

...

Another Way of Looking for Missing Broker Id

The approach proposed in this KIP is expensive and is linear to the number of topics, which probably works for thousands of topic but is difficult to scale even further.

...