Skip to content

Only log Kafka commit failures every 10m and if failing for > 10m. #36685

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Nov 3, 2025
Merged

Only log Kafka commit failures every 10m and if failing for > 10m. #36685

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
5b8afe2
Only log Kafka commit failures if failing for > 10 minutes.
tomstepp Oct 30, 2025
5c11361
Simplify commit failed logging conditions
tomstepp Oct 30, 2025
9bd0ca7
Removed unused member
tomstepp Oct 30, 2025
ad3d682
Keep info log for failed commits.
tomstepp Oct 31, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 19 additions & 3 deletions sdks/java/io/kafka/src/main/java/org/apache/beam/sdk/io/kafka/KafkaUnboundedReader.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -157,7 +157,7 @@ public boolean advance() throws IOException {
*/
while (true) {
if (curBatch.hasNext()) {
// Initalize metrics container.
// Initialize metrics container.
kafkaResults = KafkaSinkMetrics.kafkaMetrics();

PartitionState<K, V> pState = curBatch.next();
Expand Down Expand Up @@ -374,6 +374,7 @@ public boolean offsetBasedDeduplicationSupported() {
private static final Duration RECORDS_DEQUEUE_POLL_TIMEOUT_MIN = Duration.millis(1);
private static final Duration RECORDS_DEQUEUE_POLL_TIMEOUT_MAX = Duration.millis(20);
private static final Duration RECORDS_ENQUEUE_POLL_TIMEOUT = Duration.millis(100);
private static final Duration MIN_COMMIT_FAIL_LOG_INTERVAL = Duration.standardMinutes(10);

// Use a separate thread to read Kafka messages. Kafka Consumer does all its work including
// network I/O inside poll(). Polling only inside #advance(), especially with a small timeout
Expand All @@ -392,6 +393,7 @@ public boolean offsetBasedDeduplicationSupported() {
private AtomicReference<@Nullable KafkaCheckpointMark> finalizedCheckpointMark =
new AtomicReference<>();
private AtomicBoolean closed = new AtomicBoolean(false);
private Instant nextAllowedCommitFailLogTime = Instant.ofEpochMilli(0);

// Backlog support :
// Kafka consumer does not have an API to fetch latest offset for topic. We need to seekToEnd()
Expand Down Expand Up @@ -612,6 +614,7 @@ private void commitCheckpointMark() {
if (checkpointMark != null) {
LOG.debug("{}: Committing finalized checkpoint {}", this, checkpointMark);
Consumer<byte[], byte[]> consumer = Preconditions.checkStateNotNull(this.consumer);
Instant now = Instant.now();

try {
consumer.commitSync(
Expand All @@ -621,11 +624,24 @@ private void commitCheckpointMark() {
Collectors.toMap(
p -> new TopicPartition(p.getTopic(), p.getPartition()),
p -> new OffsetAndMetadata(p.getNextOffset()))));
nextAllowedCommitFailLogTime = now.plus(MIN_COMMIT_FAIL_LOG_INTERVAL);
} catch (Exception e) {
// Log but ignore the exception. Committing consumer offsets to Kafka is not critical for
// KafkaIO because it relies on the offsets stored in KafkaCheckpointMark.
LOG.warn(
String.format("%s: Could not commit finalized checkpoint %s", this, checkpointMark), e);
if (now.isAfter(nextAllowedCommitFailLogTime)) {
LOG.warn(
String.format(
"%s: Did not successfully commit finalized checkpoint for > %s. Current checkpoint: %s",
this, MIN_COMMIT_FAIL_LOG_INTERVAL, checkpointMark),
e);
nextAllowedCommitFailLogTime = now.plus(MIN_COMMIT_FAIL_LOG_INTERVAL);
} else {
LOG.info(
String.format(
"%s: Could not commit finalized checkpoint. Commit will be retried with subsequent reads. Current checkpoint: %s",
this, checkpointMark),
e);
}
}
}
}
Expand Down
Loading