Status
Current state: "Under Discussion"
Discussion thread: here [Change the link from the KIP proposal email archive to your own email thread]
JIRA: KAFKA-12473
Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).
Motivation
Now that 3.0 is coming up, we can change the default ConsumerPartitionAssignor to something better than the RangeAssignor. The original plan was to switch over to the StickyAssignor, but now that we have incremental cooperative rebalancing, we should consider using the new CooperativeStickyAssignor instead: this will enable the consumer group to follow the COOPERATIVE protocol, improving the rebalancing experience OOTB.
In Kafka, we currently support the following assignors:
Supported strategy | Feature | Drawback | |
RangeAssignor | Eager | Current default value. The 1st assignor we have | possible to generate heavily skewed assignments when the consumer topic subscriptions are not identical |
RoundRobinAssignor | Eager | Improvement for RangeAssignor to have balanced assignment | Didn't consider the overheads of reassignment |
StickyAssignor | Eager | Improvement for RoundRobinAssignor/RangeAssignor to preserve the existing assignments to reduce some of the overheads of a reassignment | Will have stop-the-world issue when doing rebalance |
CooperativeStickyAssignor | Eager, Cooperative | To be default value in 3.0 as in this KIP described, by having multiple rounds of rebalance to avoid the stop-the-world issue as described in KIP-429 |
As above table showed, since we already introduced a better assignor strategy, we should change the default assignor now.
Public Interfaces
With this KIP, the default value of partition.assignment.strategy
changes from "RangeAssignor"
to "CooperativeStickyAssignor"
- It won't affect to the current consumers if current consumers have explicitly set the
partition.assignment.strategy
value - If the consumers rely on the default value of
partition.assignment.strategy
, users can do 1 of the following options:Set the
partition.assignment.strategy
config to one of the non-cooperative assignors during the rolling upgrade if they wish to remain on EAGER and/or perform the upgrade in just a single rolling bounce.- Upgrading the consumer (refer to the following upgrade path section)
Proposed Changes
- Change the default setting of
partition.assignment.strategy
in implementation and tests - When changing the default assignor we need to make sure this is clearly documented in the upgrade guide for 3.0
Compatibility, Upgrade path
this will require users to follow the upgrade path laid out in KIP-429 to safely perform a rolling upgrade. Please note: KAFKA-12477 is doing the upgrade experience improvement to reduce the safe upgrade path to just a single rolling bounce. It's still in progress so far. Before it completed, we still need to follow the following upgrade path:
<Copy from KIP-429>
From the user's perspective, the upgrade path of leveraging new protocols is similar to switching to a new assignor. For example, assuming the current version of Kafka consumer is 2.2 and "range" assignor is specified in the config (or no assignor is configured, which is identical as the RangeAssignor is the default below 3.0). The upgrade path would be:
- The first rolling bounce is to replace the byte code (i.e. swap the jars) and introduce the cooperative assignor: set the assignors to "cooperative-sticky, range" (or round-robin/sticky/etc if you are using a different assignor already). At this stage, the new versioned byte code sends both assignors in their join-group request, but will still choose EAGER as the protocol since it's still configured with the "range" assignor, and the selected rebalancing protocol must be supported by all assignors. in the list. The "range" assignor will be selected to assign partitions while everyone is following the EAGER protocol. This rolling bounce is safe.
- The second rolling bounce is to remove the "range" (or round-robin/sticky/etc) assignor, i.e. only leave the "cooperative-sticky" assignor in the config. At this stage, whoever has been bounced will then choose the COOPERATIVE protocol and not revoke partitions while others not-yet-bounced will still go with EAGER and revoke everything. However the "cooperative-sticky" assignor will be chosen since at least one member who's already bounced will not have "range" any more. The "cooperative-sticky" assignor works even when there are some members in EAGER and some members in COOPERATIVE: it is fine as long as the leader can recognize them and make assignment choice accordingly, and for EAGER members, they've revoked everything and hence did not have any pre-assigned-partitions anymore in the subscription information, hence it is safe just to move those partitions to other members immediately based on the assignor's output.
The key point behind this two rolling bounce is that, we want to avoid the situation where leader is on old byte-code and only recognize "eager", but due to compatibility would still be able to deserialize the new protocol data from newer versioned members, and hence just go ahead and do the assignment while new versioned members did not revoke their partitions before joining the group. Note the difference with KIP-415 here: since on consumer we do not have the luxury to leverage on list of built-in assignors since it is user-customizable and hence would be black box to the consumer coordinator, we'd need two rolling bounces instead of one rolling bounce to complete the upgrade, whereas Connect only need one rolling bounce.
Rejected Alternatives
If there are alternative ways of accomplishing the same thing, what were they? The purpose of this section is to motivate why the design is the way it is and not some other way.