Workflows keep stacking up

[RyanNutt]

Not sure if this is a bug or how temporal is supposed to work. But I've noticed that the workflow list just keeps growing. Mine looks like this right now.

Shouldn't most of these go away as posts are sent?

And what sent me down the rabbit hole is that scheduled posts were silently not happening, and then after restarting the docker stack they all sent at once.

[id-ego]

The growing list and the stalled publishing are probably two separate things. Closed Temporal workflows normally remain visible until their retention period expires, so they do not disappear as soon as a post is sent.

The silent queue followed by a burst after restart does match the worker-startup failure described in #1748: the process and /health/status can look healthy even though the actual Temporal worker was never registered. Restarting once Temporal is ready then registers the worker and drains the backlog.

Before another restart, inspect every integration for overdue QUEUE items and pause or remove anything that must not go out; otherwise the entire backlog can publish at once. After restarting, verify that the worker is registered in Temporal and use one controlled post to confirm the full terminal/provider state rather than relying only on PM2 online or /health/status. If it recurs, the most useful evidence is the orchestrator startup block around TemporalConnectionFactory; the durable fix is to retry or fail startup loudly and make health reflect the registered worker. Workflow-history retention/cleanup can then be handled separately from this publishing failure.