Split auxiliary processes our of the ScyllaDB container

[tnozicka]

We are running multiple processes within ScyllaDB container which is against the best practices but most importantly a cgroup is assigned to a particular container. The issues that stem from that are:

• other processes steal cpu quota from ScyllaDB and affect latency
• coupling the lifetimes - if the sidecar fails, it kills ScyllaDB too
• multiple unrelated logs are mixes into 1 container and neither can be parsed when collected
• zombies can be left around

There are 2 classes of processes that are run within ScyllaDB container. The ones coming from the scylla-operator binary that gets copied there through a shared filesystem with mostly static linking and provides probes, sidecar controller and startup logic. Then there are also the processes coming from the ScyllaDB container itself that use the anti-pattern with supervisord and running multiple processes there.

Tasks

Preview Give feedback

1. Split operator auxiliary processes out of ScyllaDB container and remove the need to copy the operator binary there #1943
   2 of 6
   kind/epic priority/important-soon triage/accepted +3
2. Split scylla container supervisord processes into separate containers and remove supervisord #1769
   2 of 7
   kind/epic priority/important-longterm +2

[scylla-operator-bot]

The Scylla Operator project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 30d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out

/lifecycle stale

[tnozicka]

/remove-lifecycle stale
/triage accepted