fix: align kafka consumer fetch ceiling to 64 MiB broker max (#442)

* fix: align kafka consumer fetch ceiling to 64 MiB broker max

librdkafka defaults max.partition.fetch.bytes to 1 MiB and fetch.max.bytes
to 50 MiB, both below the 64 MiB a DLQ re-publish can reach. An
oversized-but-valid record is accepted/stored by the broker yet trips
MSG_SIZE_TOO_LARGE on consume, wedging the partition in a re-seek loop
(observed: DLQ replay group at ~243 MiB in-flight, 0 committed partitions).

Raise consumer fetch ceilings to match the producer/broker 64 MiB
max.message.bytes. Config-only; applies to both primary and _dlq groups.

* fix: add missing imports to ServerSpec (threadDelay, catch, finally)

* style: trim fetch-ceiling comment to terse inline form
This commit is contained in:
Anthony Alaribe 2026-06-22 16:38:11 +02:00 • committed by GitHub
commit fbf0e2fefd
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 7 additions and 0 deletions