Hi Martijn, thanks for replying. I would also expect the behavior you describe below. AFAICT it was also like this with Flink 1.14. I am aware that Flink is using checkpointing for fault tolerance, but for example the Kafka offsets are part of our monitoring and this will lead to alerts. Other applications which use the Kafka client directly also do not show repeated commit failures once all Kafka brokers are online again. I think this occurs in Flink jobs using Flinks Kafka Connector directly (KafkaSource) or via a Kafka SQL Connector based application.
Will try to write a small job to verify this behavior, as we also use flink-avro-confluent-registry which makes it harder to understand the root of the issue. Best regards, Christian Von: Martijn Visser <martijnvis...@apache.org> Datum: Montag, 13. Juni 2022 um 12:05 An: Christian Lorenz <christian.lor...@mapp.com> Cc: "user@flink.apache.org" <user@flink.apache.org> Betreff: Re: Kafka Consumer commit error This email has reached Mapp via an external source Hi Christian, I would expect that after the broker comes back up and recovers completely, these error messages would disappear automagically. It should not require a restart (only time). Flink doesn't rely on Kafka's checkpointing mechanism for fault tolerance. Best regards, Martijn Op wo 8 jun. 2022 om 15:49 schreef Christian Lorenz <christian.lor...@mapp.com<mailto:christian.lor...@mapp.com>>: Hi, we have some issues with a job using the flink-sql-connector-kafka (flink 1.15.0/standalone cluster). If one broker e.g. is restarted for maintainance (replication-factor=2), the taskmanagers executing the job are constantly logging errors on each checkpoint creation: Failed to commit consumer offsets for checkpoint 50659 org.apache.flink.kafka.shaded.org.apache.kafka.clients.consumer.RetriableCommitFailedException: Offset commit failed with a retriable exception. You should retry committing the latest consumed offsets. Caused by: org.apache.flink.kafka.shaded.org.apache.kafka.common.errors.CoordinatorNotAvailableException: The coordinator is not available. AFAICT the error itself is produced by the underlying kafka consumer. Unfortunately this error cannot be reproduced on our test system. From my understanding this error might occur once, but follow up checkpoints / kafka commits should be fine again. Currently my only way of “fixing” the issue is to restart the taskmanagers. Is there maybe some kafka consumer setting which would help to circumvent this? Kind regards, Christian Mapp Digital Germany GmbH with registered offices at Dachauer, Str. 63, 80335 München. Registered with the District Court München HRB 226181 Managing Directors: Frasier, Christopher & Warren, Steve This e-mail is from Mapp Digital and its international legal entities and may contain information that is confidential or proprietary. If you are not the intended recipient, do not read, copy or distribute the e-mail or any attachments. Instead, please notify the sender and delete the e-mail and any attachments. Please consider the environment before printing. Thank you. Mapp Digital Germany GmbH with registered offices at Dachauer, Str. 63, 80335 München. Registered with the District Court München HRB 226181 Managing Directors: Frasier, Christopher & Warren, Steve This e-mail is from Mapp Digital and its international legal entities and may contain information that is confidential or proprietary. If you are not the intended recipient, do not read, copy or distribute the e-mail or any attachments. Instead, please notify the sender and delete the e-mail and any attachments. Please consider the environment before printing. Thank you.
