Open this page in two browsers, pick the other user as recipient,
send a message. Their tab pops a card; the value they click comes
back as your live.push reply. Or schedule it 3 seconds
out and watch the 6-field live.cron tick drain it.
Sending as Breezy Turnip (85a0077d)
Nothing yet. When someone pushes you a notification, the card appears here with reply buttons.
Nothing pending. Set a schedule above to enqueue a notification.
Each entry names the worker that handled that step, and marks how the push reached its recipient. Those are two different pieces of cluster machinery. Differing worker ids on a schedule and its fire mean the cluster-shared queue and the leader-gated cron did their job: one instance enqueued it, another drained it. The connection registry is not involved in that - its job is reaching a recipient's socket wherever it happens to be connected, and the badge is what reports it: relayed via cluster when the recipient was on another instance, same instance when no hop was needed. (One local instance shows one worker id and no relays.)
Nothing yet.