Sometimes Kafka Strimzi service does not reconcile.
Following are the scenarios to be considered if the Kafka-Strimizi or Kafka-Edge does not reconcile:
- If the topic is created, but CR is not reconciled due to etcd issue, then after the etcd issue is resolved, you might have to restart the
strimzi-cluster-operator
,entity-operator
, oredge-entity-operator
by executing the following commands.kubectl get pods | egrep "strimzi-cluster-operator" | cut -f1 -d' ' | xargs kubectl delete pod kubectl get pods | egrep "internal-entity-operator" | cut -f1 -d' ' | xargs kubectl delete pod kubectl get pods -n kafka-edge | egrep "edge-entity-operator" | cut -f1 -d' ' | xargs kubectl delete pod -n kafka-edge
- If the topic is not created, you must restart the topic operator.
For internal Kafka, run the following command.
kubectl get pods | egrep "internal-entity-operator" | cut -f1 -d' ' | xargs kubectl delete pod
For Edge Kafka, run the following command.kubectl get pods -n kafka-edge | egrep "edge-entity-operator" | cut -f1 -d' ' | xargs kubectl delete pod -n kafka-edge
- If the topic is created, but CR not reconciled, execute the following command:
kubectl get kafkatopic NAME CLUSTER PARTITIONS REPLICATION FACTOR READY vsa-alerts internal 4 1 vsa-alerts---57fba39f4d6436566c678036822b624a918e720f internal 1 1 True vsa-baseline internal 1 1 vsa-baseline---73960ad4f7a1fcf1fb6f1fff79846a8d1f4f496f internal 1 1 True
In the output,
vsa-alerts
andvsa-baseline
are not reconciled. Here the topic is created with default partition of 1 as seen in the output which might lead to a performance impact. However, there is no functional impact.The partition can be increased by editing the CR as follows.- For internal Kafka, execute the following command:
kubectl edit kafkatopic vsa-alerts---57fba39f4d6436566c678036822b624a918e720f
Change partition to 4.
- For Edge Kafka, execute the following command:
kubectl edit kafkatopic -n kafka-edge metrics-ext-data---2c3fd3a5b7407dddf679abdbf87726bfad6a4dad
Change partition to 4.
- For internal Kafka, execute the following command:
Kafka Edge not getting reconciled as few Kafka topics are conflicting with auto-created topics
Problem
kubectl get kt -A NAMESPACE NAME CLUSTER PARTITIONS REPLICATION FACTOR READY kafka-edge inv-topo-raw edge 4 3 kafka-edge inv-topo-raw---12b38b0b5a182d693f9783a3dd5e8b8cf14ba5bb edge 4 3 True kafka-edge vsa-events-raw edge 5 3 kafka-edge vsa-events-raw---68e94faa465ed3f2dbe939b97ad68eed2cfb72a7 edge 5 3 True kafka-edge vsa-metrics-raw edge 6 3 kafka-edge vsa-metrics-raw---9115db2760b5b9c3da2d2e28e3577aff10a9f2cc edge 6 3 True
Solution
- Stop all the Collectors and Enrichment.
- Delete the auto-created topics.
kubectl delete kt -n kafka-edge inv-topo-raw---12b38b0b5a182d693f9783a3dd5e8b8cf14ba5bb vsa-events-raw---68e94faa465ed3f2dbe939b97ad68eed2cfb72a7 vsa-metrics-raw---9115db2760b5b9c3da2d2e28e3577aff10a9f2cc Output kafkatopic.kafka.strimzi.io "inv-topo-raw---12b38b0b5a182d693f9783a3dd5e8b8cf14ba5bb" deleted kafkatopic.kafka.strimzi.io "vsa-events-raw---68e94faa465ed3f2dbe939b97ad68eed2cfb72a7" deleted kafkatopic.kafka.strimzi.io "vsa-metrics-raw---9115db2760b5b9c3da2d2e28e3577aff10a9f2cc" deleted
Wait for Kafka Edge to reconcile.
kubectl get app kafka-edge kafka-edge