Skip to content

Fix intermittent "jdbc registry, kafka messaging with embedded cache" CI failure - #3762

Draft
sophokles73 with Copilot wants to merge 1 commit into
masterfrom
copilot/fix-failing-github-actions-job
Draft

Fix intermittent "jdbc registry, kafka messaging with embedded cache" CI failure#3762
sophokles73 with Copilot wants to merge 1 commit into
masterfrom
copilot/fix-failing-github-actions-job

Conversation

Copilot AI commented Jul 18, 2026

Copy link
Copy Markdown

The jdbc/kafka/embedded CI matrix job intermittently fails with CoAP adapter freezes and cascading TelemetryCoapIT DTLS timeouts (HTTP 503). Root cause: confluentinc/cp-kafka:8.0.3 carries Confluent Platform overhead (REST proxy, Schema Registry stubs, extra JVM agents) that pushes the test runner—already hosting 5 adapters, Command Router, JDBC registry, Auth Server, PostgreSQL, and Kafka—over its memory budget. Observable in run 87962894532: Kafka producers hit UNKNOWN_TOPIC_OR_PARTITION during metadata refresh, CoAP adapter goes silent at 18:48:18Z, all subsequent DTLS handshakes time out.

Changes

  • .github/workflows/ci.yml: swap docker.io/confluentinc/cp-kafka:8.0.3docker.io/apache/kafka:4.0.0 for the jdbc matrix entry
    • Same Kafka 4.0 wire protocol; CI comment unchanged
    • Consistent with mongodb-7.0 entry already using docker.io/apache/kafka:4.1.1
    • All KAFKA_* env vars in tests/pom.xml are supported by the official Apache image

…bedded CI job

The Confluent Platform image (cp-kafka:8.0.3) carries additional Confluent
services overhead compared to the official Apache Kafka image. Under
constrained GitHub Actions runner memory (~7 GB total), the extra resource
pressure caused intermittent CoAP adapter freezes in run 87962894532:
Kafka producers hit UNKNOWN_TOPIC_OR_PARTITION during metadata refresh,
eventually making DTLS handshakes time out with HTTP 503.

Switch to docker.io/apache/kafka:4.0.0 which:
 - Tests the same Kafka 4.0 protocol version (CI comment unchanged)
 - Uses the same KAFKA_* environment variable convention already used by the
   mongodb-7.0 matrix entry (apache/kafka:4.1.1)
 - Reduces memory/CPU footprint on the CI runner
Copilot AI changed the title Fix "Use jdbc registry, kafka messaging with embedded cache" CI job failure Fix intermittent "jdbc registry, kafka messaging with embedded cache" CI failure Jul 18, 2026
Copilot AI requested a review from sophokles73 July 18, 2026 03:42
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants