Repository navigation
fix: lift the Kafka sink record ceiling to 16 MiB - #116
Merged
Merged
Conversation
The size cap in front of Vector's Kafka sink fell back to librdkafka's own message.max.bytes default of 1,000,000 bytes, and the generated producer set no ceiling either, so any record over 999,872 bytes was dropped on every deployment. Every other stage's producer and every topic scalo creates carry 16 MiB.
The sink's librdkafka options now take scalo's MESSAGE_MAX_BYTES where no layer sets message.max.bytes, so a central profile or sink.librdkafka_options still decides it. The size cap reads the same merged value, so the cap and the producer cannot disagree. A record over the ceiling is still held back by the cap and counted once in pipeline_dead_letters_dropped_total{reason="too_large"}.
scalo's scaling feature is now named in Cargo.toml. It already arrived through cli-service, and the runtime already writes the scaling_pressure gauge, which a new test reads off a running binary.
Refs #115
unsafe-libyaml is flagged unmaintained, not vulnerable. It arrives via figment and serde_yaml_ng and parses operator-supplied config YAML only. Tracked in hyperi-io/scalo-rs#341.
|
Released in v1.0.49 -- https://github.com/hyperi-io/dfe-transform-vector/releases/tag/v1.0.49 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Vector's Kafka sink dropped any record over 999,872 bytes, on every deployment. The size cap in front of it fell back to librdkafka's own
message.max.bytesdefault of 1,000,000, and the producer we generate set no ceiling either, so a record that got past the cap was refused by librdkafka anyway. Every other stage's producer, and every topic scalo creates, carries 16 MiB.What changed:
MESSAGE_MAX_BYTES(16 MiB) where no layer setsmessage.max.bytes. The central profile orsink.librdkafka_optionsstill decides it where set, so the value dfe-infra#678 writes on managed Kafka is the ceiling there.pipeline_dead_letters_dropped_total{reason="too_large"}. This app has no DLQ (docs/DESIGN.md), and a record over the Kafka ceiling would not fit a Kafka DLQ topic either.scalingis now named in the scalo features. It already arrived throughcli-service, and the runtime already writes thescaling_pressuregauge. It reads 0: the supervisor feeds the pressure no components and no memory reading.Tests:
too_large. Red on main with Vector's own totalsreceived: 2.0, sent: 0.0, oversize: 2.0-- the cap held back both records.message.max.bytes: 16777216and caps at 16777088; a deployment'smessage.max.bytes(2000000, 8388608) is the ceiling for both the producer and the cap.scaling_pressureas a gauge and/scaling/pressureanswers 200. This one passes on main too.Refs #115