You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 5 Next »

Status

Current state: Under Discussion

Discussion thread: here

JIRA: here

Pull Request: here

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

Motivation

Kafka Streams treats repartition topics differently to regular topics. Instead of setting arbitrary retention criteria and having the broker cleanup old records, Kafka Streams sets infinite retention on repartition topics and explicitly deletes records once they've been committed to the next topic in their Topology. Currently, this is done every time the Task is committed, resulting in explicit "delete records" requests being sent every commit.interval.ms milliseconds.

When commit.interval.ms is set very low, for example when processing.guarantee is set to exactly_once_v2, this causes delete records requests to be sent extremely frequently, potentially reducing throughput and causing a high volume of log messages to be logged by the brokers.

Public Interfaces

New configuration options

NameTypeImportanceDefaultDescription
repartition.purge.interval.msLongLOW30000The frequency in milliseconds with which to delete fully consumed records from repartition topics.  The default value is the same as the default for commit.interval.ms (30000).  Records are only deleted after a successful commit, so setting this to a value equal to or larger than commit.interval.ms will bound it the commit interval. (Note, unlike commit.interval.ms, the default for this value remains unchanged when processing.guarantee is set to exactly_once_v2).

Proposed Changes

Adding a new configuration option, repartition.purge.interval.ms, that configures the frequency these explicit record deletions are sent will resolve the issue, by enabling users to tune the commit.interval.ms and repartition.purge.interval.ms separately.

We will still wait for a commit before explicitly deleting repartition records, but we will only do so if the time since the last record deletion is at least repartition.purge.interval.ms. This means the lower-bound for repartition.purge.interval.ms  is effectively capped by the value of commit.interval.ms.

Compatibility, Deprecation, and Migration Plan

  • The interval between explicit delete requests for repartition records will no longer be coupled to commit.interval.ms. Default behaviour is unchanged, however:
    • When commit.interval.ms  is explicitly modified by the user, old repartition records will no longer be deleted on every commit.
    • When processing.guarantee is set to exactly_once_v2, since the default commit.interval.ms is changed internally to 100 ms, old repartition records will no longer be deleted on every commit.
    • Users can regain this coupling by explicitly configuring both commit.interval.ms and repartition.purge.interval.ms to the same value.

Rejected Alternatives

Modifying the explicit deletion of records to be completely independent of commits such that repartition.purge.interval.ms is strictly adhered to, irrespective of the value of commit.interval.ms was not explored, as the increased complexity of the changes may introduce bugs, with little additional benefit.

  • No labels