# Kafka server fails to start

**URL:** <https://forum.confluent.io/t/kafka-server-fails-to-start/4188>\
**Category:** Containers\
**Created:** [16 February 2022 16:05 UTC](https://forum.confluent.io/t/kafka-server-fails-to-start/4188 "2022-02-16T16:05:44Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![alexei.ivanov](https://sea1.discourse-cdn.com/flex019/user_avatar/forum.confluent.io/alexei.ivanov/32/5049_2.png) [@alexei.ivanov](https://forum.confluent.io/u/alexei.ivanov)\
**Post date:** [16 February 2022 16:05 UTC](https://forum.confluent.io/t/kafka-server-fails-to-start/4188/1 "2022-02-16T16:05:44Z")

</div>

Kafka server fails to start due to the following error:

`Readiness probe failed: dial tcp 10.244.2.23:9071: connect: connection refused`

Kafka server deployed on AKS using [this example](https://github.com/confluentinc/confluent-kubernetes-examples/blob/1ce5d79c48fa4b9590ca8369d13e1c916f582f89/general/cp-version/confluent-platform-7.0.0.yaml). The condition of the k8s cluster is normal. Confluent operator is up and running. Zookeeper starts up fine.

Steps taken so far:

1. Re-deployed the confluent cluster
2. Increased timeout & initial delay of the probes
3. Tried using different image version

First error in the Kafka pod log:

`[ERROR] 2022-02-16 15:49:25,245 [controller-event-thread] state.change.logger error - [Controller id=0 epoch=5130] Controller 0 epoch 5130 failed to change state for partition _confluent-controlcenter-7-0-1-0-group-stream-extension-rekey-1 from OfflinePartition to OnlinePartition`

Anyone familiar with this issue? Any help or tips will be highly appreciated.
