Horizon needs Redis available to start

[holysoles]

Horizon requires Redis/Valkey being available to start properly. This can fail during initial deployment

[2025-01-25 05:51:24] production.ERROR: Connection refused [tcp://valkey-primary:6379] {"exception":"[object] (Predis\\Connection\\ConnectionException(code: 111): Connection refused [tcp://valkey-primary:6379] at /var/www/vendor/predis/predis/src/Connection/AbstractConnection.php:144)

When pixelfed is deployed in a single pod, this is problematic since Horizon exits, but the rest of the pod continues to run (I think this is because of the CMD script of the container?)

When pixelfed is deployed with a separate worker pod, this is less problematic, but the pod will restart a few times.

Not sure of the best solution right now, but figured I'd document this.

[jessebot]

Hmmm, we can separate the deployments and put a liveness/readiness probe on the horizon deployment to retry x amount of times. Maybe have it try 10 times to be safe? I feel like as long as it eventually becomes healthy, that's ok?