Specific broker keeps being restarted constantly #10256
Replies: 2 comments 18 replies
-
Providing the log for that broker could be a good starting point. It could provide some clues about why it's not ready and probe is failing maybe. |
Beta Was this translation helpful? Give feedback.
-
I guess you should start by sharing the full configurations and full logs from all the components. 1.5Gi is not much for Kafka. I normally use at least 2Gi just for simple Strimzi development without any load. But it is not something what will be an obvious issue. It will depend on how it is used etc. Also, if you are running in KRaft mode, 6 mixed nodes seem like a misconfiguration. You should for example have 3 mixed nodes and 3 brokers or 3 controller-only nodes and 3 broker-only nodes etc. That said, given the current Apache Kafka limitations, there is no easy way to change it. |
Beta Was this translation helpful? Give feedback.
-
Hello.
I have a KRaft-based Kafka Cluster with 6 brokers. The cluster was recently migrated to KRaft. We have 1.5GiB on each broker (requests and limits).
Although, since a few days ago, we are experiencing issues because there is a broker constantly getting restarted.
The error on broker 3 is the following:
It seems the probes are constantly timing out only for this broker. We are using the default configurations for the probes.
At first, we thought it was OOM error but there is no clear evidence of that. Furthermore, this is only happening on this broker.
Thank you in advance.
Beta Was this translation helpful? Give feedback.
All reactions