kafka runtime_error ai_generated partial

org.apache.kafka.common.errors.ReplicaNotAvailableException: Replica for partition is not available

ID: kafka/replica-not-available

Also available as: JSON · Markdown · 中文

85%Fix Rate

84%Confidence

1Evidence

2023-09-20First Seen

Version Compatibility

Version	Status	Introduced	Deprecated	Notes
2.8.2	active	—	—	—
3.2.3	active	—	—	—
3.5.1	active	—	—	—

Root Cause

A follower replica for the requested partition is offline, not fully caught up, or the fetch request targets an unavailable replica.

generic

中文

请求的分区的跟随者副本离线、未完全同步，或拉取请求指向了不可用的副本。

Official Documentation

https://kafka.apache.org/documentation/#replication

Workarounds

90% success Check the broker logs for the specific partition and bring the offline replica online: # Identify the partition leader kafka-topics.sh --bootstrap-server localhost:9092 --describe --topic my_topic # Restart the broker hosting the unavailable replica or reassign replicas: kafka-reassign-partitions.sh --bootstrap-server localhost:9092 --reassignment-json-file reassign.json --execute
```
Check the broker logs for the specific partition and bring the offline replica online:
# Identify the partition leader
kafka-topics.sh --bootstrap-server localhost:9092 --describe --topic my_topic
# Restart the broker hosting the unavailable replica or reassign replicas:
kafka-reassign-partitions.sh --bootstrap-server localhost:9092 --reassignment-json-file reassign.json --execute
```
80% success Set the consumer to read from the leader only by setting client-side rack awareness or using a fetch from the leader (default behavior if replica selector is not configured).
```
Set the consumer to read from the leader only by setting client-side rack awareness or using a fetch from the leader (default behavior if replica selector is not configured).
```

中文步骤

Check the broker logs for the specific partition and bring the offline replica online:
# Identify the partition leader
kafka-topics.sh --bootstrap-server localhost:9092 --describe --topic my_topic
# Restart the broker hosting the unavailable replica or reassign replicas:
kafka-reassign-partitions.sh --bootstrap-server localhost:9092 --reassignment-json-file reassign.json --execute

Set the consumer to read from the leader only by setting client-side rack awareness or using a fetch from the leader (default behavior if replica selector is not configured).

Dead Ends

Common approaches that don't work:

95% fail
Restarting the consumer only retries the fetch, but the replica remains unavailable.
85% fail
Increasing fetch.max.wait.ms only delays the timeout, not the root cause.