# Clean up legacy soju Deployment

## Context

The pod `soju-d985d7c99-v8s6w` has been in CrashLoopBackOff for 36+ hours (423 restarts). Root cause: a manually-created legacy `soju` Deployment references a ConfigMap with `db sqlite3 /data/soju.db`, but the soju image is built with `nosqlite` (PostgreSQL only). This legacy Deployment predates the soju-operator and is not used by anything -- all active bouncers (apple, josie, doggy, pupy) are operator-managed with correct PostgreSQL configs.

## Resources to delete

All three were created manually via the OpenShift web console on 2026-03-03:

1. **Deployment** `soju` -- the crashing deployment (selector: `app: soju, deployment: soju`)
2. **Service** `soju` -- ClusterIP pointing at the dead pod (nothing references it)
3. **ConfigMap** `soju-config` -- contains the broken `db sqlite3` config

## Safety check

- Chat proxies to `soju-shared` (per `chat/nginx.conf`), not `soju`
- Operator-managed bouncers have their own services: `apple`, `josie`, `doggy`, `pupy`
- No other service, route, or config references the `soju` service

## Steps

```bash
oc delete deployment soju -n irc-josie-cloud
oc delete service soju -n irc-josie-cloud
oc delete configmap soju-config -n irc-josie-cloud
```

## Verification

```bash
oc get deployment,service,configmap -n irc-josie-cloud | grep -E '^(NAME|.*soju)' | grep -v 'soju-db\|soju-operator'
```

Should show only `soju-operator` deployment and operator-managed bouncer configs (e.g. `josie-config`, `apple-config`).