My Pulsar cluster installation using the Pulsar helm chart (v3.0.0) gets stuck on the startup of the broker and proxy services. The respective init containers do not complete. Here’s what I’ve done: 1. Followed the official guide to installed Pulsar into Kuberentes. Used all the default values from the guide. 2. Installed the helm chart. 3. After about 6min, all of the services are running, except for the broker and the proxy. The broker and proxy services do not start up, even after an hour of waiting. 4. Opened a terminal inside the

wait-zookeeper-ready

init container of the broker. The command

echo "ruok" | nc pulsar-mini-zookeeper 2181 ; echo

returns “imok”, which means that the container can reach zookeeper, and zookeeper is running. 5. Retrieved the logs of the

wait-zookeeper-ready

container using the command

kubectl logs pulsar-mini-broker-0 -c wait-zookeeper-ready --namespace pulsar

. The logs show that the init container could not reach Zookeeper. The interesting observation is that the checks were only ran once at the very beginning of installing the pulsar chart, when zookeeper was not ready yet. The checks are not continuously executed, so the init container can never complete. 6. When I manually delete the broker pod, it restarts and successfully enters running state. Then the proxy server also enters running state. What could be the problem? And how could I get the helm install to work?