Natalie Klestrup Röijezon
Natalie Klestrup Röijezon
This is an attempt to fix #34. Currently (as of 46ce1b7) everything compiles, but `nix-build` fails because it's unable to locate the `test-jar` artifact of `"org.apache.logging.log4j" % "log4j-core" % "2.8.1"`....
Currently, we write images of the format `stackable/zookeeper:3.5.8`, but (according to the logs) the agent only actually tries to resolve the package name `zookeeper:3.5.8` against all repositories.
This causes systemd to (nearly) always report the system status as degraded. Can be cleared manually by running `systemctl reset-failed $namespace-$pod-$container.service`.
Currently this is emitted to the systemd journal like this: ``` Oct 06 12:35:36 stackable-node2 stackable-agent[1373]: [2021-10-06T10:35:36Z INFO stackable_agent::provider::states::pod::downloading] Looking for package: library-busybox:1.32.1 in known repositories Oct 06 12:35:36 stackable-node2...
This means that restarting the K8s apiserver leaves all running agents in limbo, and that care must be taken so that the apiserver is running before starting any new agents.
See #18 (I thought I had a fix, but it didn't seem to work).
# Description This is far from final (for example, we'd need to expose a configurable LoadBalancerClass, for example), but I wanted to see whether lb-op would be a good fit...