Onboarding papercuts

[nightkr]

A (relative unorganized) bunch of issues I encountered when trying to set up a testing cluster by following https://docs.stackable.tech/home/getting_started.html:

• Clarify supported/recommended distros (Debian 10, CentOS 7/8) (Start addressing papercuts #96)
• Curl is used before being installed (Start addressing papercuts #96)
• Inconsistent CRD installation methods (copy/paste vs curl), consider adding an amalgamation YAML?
• Do we need to specify the repository with both the Repository k8s object and APT/YUM source lists?
• Do we need to install the JVM manually? (shouldn't the specific app packages depend on this?)
• Deb packages and systemd commands are wrong (need to remove -server suffix) (Start addressing papercuts #96)
• Deb convention is that packages automatically enable required systemd services (I believe this is different on RHEL/CentOS?)
• Log notes should probably refer to journalctl rather than syslog (Start addressing papercuts #96)
• Systemd services should implement sd-notify protocol so that systemd can notify the user immediately that startup failed
• Kubeconfig also needs to be copied from /etc/rancher/k8s/k8s.yaml to /root/.kube/config on controller node (never documented) (Start addressing papercuts #96)
• Example ZookeeperCluster doesn't conform to CRD schema (seems to be an indentation error, works if I change it to https://gist.github.com/teozkr/4f373aed9e7d6c25a5d47fa290d8563b) (Start addressing papercuts #96)
• Zookeeper operator keeps failing with ERROR stackable_operator::controller: Reconciliation finished with an error, this should not happen, please file an issue err=QueueError { source: InitialListFailed { source: Api(ErrorResponse { status: "404 Not Found", message: "\"404 page not found\\n\"", reason: "Failed to parse error data", code: 404 }), backtrace: Backtrace(()) }, backtrace: Backtrace(()) }
• Zookeeper CRD YAML doesn't include command CRDs
• Stackable Agent dies until manual restart if K8s API is unavailable (Agent currently doesn't restart when losing K8s API access agent#317)
• Required ZooKeeper config fields aren't provided in example (dataDir, initLimit, syncLImit) (Start addressing papercuts #96)
• ZooKeeper fails to connect to peers, because they are configured to listen by hostname (which Debian binds to localhost by default) (Start addressing papercuts #96)
• Operators don't mention which CRDs it's failing to find (Controller should report errors querying related resources separately kube-rs/kube#640)
• ZooKeeper isn't reconciling config changes without a manual Restart command (Config changes (in CRD object) aren't propagated to cluster without an explicit Restart command zookeeper-operator#227)
• ZooKeeper isn't setting a clientPort by default, but assumed by Kafka (Start addressing papercuts #96)
• Stackable Agent doesn't add any Events on Pod about package installation progress (leaving the Pod looking inexplicably stuck as Registered)

[lfrancke]

I'll just assume that most of these are now outdated and will close the issue. I had a quick glance at the list and it looks like most of it is related to Agent stuff.