This message was deleted.

Hey @Tim Rannikko I ran into a similar problem when installing a Docker cluster with Zookeeper on an M3 Mac laptop. That installation script requires Python and assumes that you have pyyaml installed. I tried a few things and eventually did this:

pip3 install ruamel.yaml

which is what's required when using Python3. For your installation, I would recommend installing

perl

which I think is only required to do the installation, not to run the cluster.

Thanks, @Mike Sherman. Yes I had already installed Python, as that was mentioned in the documentation. Because of the error I also installed Perl, and it started to work. I later found that the Perl requirement is mentioned in the quick start guide but not in the clustered deployment guide that I was following. https://druid.apache.org/docs/latest/tutorials/cluster

Sorry, you're right @Tim Rannikko I didn't look closely at the script you were trying to run, which starts a non-Zookeeper cluster. In your situation, it appears that perl

is

required.

@Mike Sherman Also, I'm now figuring out that if you really want to use clustering, you're better off using Kubernetes for auto-scaling, etc. I've never used Kubernetes before, so I think I need to learn that now 😅

Looks like there is a perl script called

supervise

what runs perl as part of the startup without Zookeeper.

I have zookeeper 3 node cluster on separate servers

Nice, I haven't spent a lot of time with Zookeeper, other than using it in simple ways. I was going to check out what's on the znodes, seems like that should tell me more about who's doing what in the cluster... 🤔

Yes, I don't know much about Zookeeper or distributed systems in general either. It seems that Kafka is phasing out Zookeeper in favor of KRaft.

Wow, that was a year ago, pretty significant shift - https://www.baeldung.com/kafka-shift-from-zookeeper-to-kraft

Yeah, I'm not sure either. It's probably due to some unique issues specific to Kafka and the added complexity of an extra management layer. Managing metadata internally without Zookeeper makes Kafka deployment and management much easier now.