Description

Producer.Close() hanged indefinitely (I waited 10 mins). Before calling Close, I called Flush with the limit of 10,000 ms. Its return value was 186,962 (this many remaining unflushed upon return). Also, as the Close call was hanging, Producer.Len() would return 186,431.

How to reproduce

Don't have a clean method, but I have a code that was producing nearly 70K messages per second (based on Kafka's metrics) to a topic with two partitions when this error occurred.

Checklist

Please provide the following information:

[x] confluent-kafka-go and librdkafka version (LibraryVersion()): 17040127 1.4.2
[x] Apache Kafka broker version: Docker image docker.io/bitnami/kafka:2-debian-10 (kafka_2.12-2.6.0)
[x] Client configuration: ConfigMap{...}: "acks": 1, "bootstrap.servers": "..."
[x] Operating system: Linux

$ uname -a
Linux f0634a3cfcda 5.4.53.1.amd64-smp #1 SMP Fri Jul 24 10:28:09 CEST 2020 x86_64 GNU/Linux
$ go version
go version go1.15 linux/amd64

[ ] Provide client logs (with "debug": ".." as necessary)
[x] Provide broker log excerpts: Nothing special -- Flush and Close were called at about 21:08

kafka_1         | [2020-09-27 21:06:19,674] INFO [GroupMetadataManager brokerId=1001] Removed 0 expired offsets in 0 milliseconds. (kafka.coordinator.group.GroupMetadataManager)
kafka_1         | [2020-09-27 21:07:00,045] INFO [ProducerStateManager partition=topic-1] Writing producer snapshot at offset 1956018171 (kafka.log.ProducerStateManager)
kafka_1         | [2020-09-27 21:07:00,342] INFO [Log partition=topic-1, dir=/data/01] Rolled new log segment at offset 1956018171 in 2196 ms. (kafka.log.Log)
kafka_1         | [2020-09-27 21:07:11,468] INFO [ProducerStateManager partition=topic_name-0] Writing producer snapshot at offset 4789613528 (kafka.log.ProducerStateManager)
kafka_1         | [2020-09-27 21:07:11,778] INFO [Log partition=topic_name-0, dir=/data/00] Rolled new log segment at offset 4789613528 in 2607 ms. (kafka.log.Log)
kafka_1         | [2020-09-27 21:08:08,054] INFO [ProducerStateManager partition=topic_name-1] Writing producer snapshot at offset 1957162306 (kafka.log.ProducerStateManager)
kafka_1         | [2020-09-27 21:08:08,327] INFO [Log partition=topic_name-1, dir=/data/01] Rolled new log segment at offset 1957162306 in 1949 ms. (kafka.log.Log)
kafka_1         | [2020-09-27 21:08:14,160] INFO [ProducerStateManager partition=topic_name-0] Writing producer snapshot at offset 4790755843 (kafka.log.ProducerStateManager)
kafka_1         | [2020-09-27 21:08:14,319] INFO [Log partition=topic_name-0, dir=/data/00] Rolled new log segment at offset 4790755843 in 1175 ms. (kafka.log.Log)
kafka_1         | [2020-09-27 21:16:19,670] INFO [GroupMetadataManager brokerId=1001] Removed 0 expired offsets in 0 milliseconds. (kafka.coordinator.group.GroupMetadataManager)
kafka_1         | [2020-09-27 21:20:44,684] INFO [ProducerStateManager partition=topic_name-1] Writing producer snapshot at offset 1958297907 (kafka.log.ProducerStateManager)
kafka_1         | [2020-09-27 21:20:44,811] INFO [Log partition=topic_name-1, dir=/data/01] Rolled new log segment at offset 1958297907 in 6952 ms. (kafka.log.Log)