Register error code 103009 for push not enabled on a cluster

[lmars]

Registers 103009 / push_not_enabled_for_cluster, plus the regenerated protocol/errors.json.

Why

A dedicated cluster provisioned without push has no push queues, so every push publish fails at enqueue. Today that reaches the customer as a bare 50000 "Failed to send message to queue", which gives them nothing to act on. This came up in an internal thread, where a customer's publishes were failing and the underlying cause (NO_ROUTE from the broker, because the push queues weren't declared) was only visible to us in our own logs.

The answer in that situation is always the same and only Ably can act on it, so the code says so, and its message names support directly rather than making the reader reach the docs to find out what to do.

The entry

• Title — Push notifications not enabled for the cluster
• Summary — what happened, then that push has to be turned on by Ably on a dedicated cluster
• Body — What you should do leads with contacting support and notes the error won't clear on its own or on retry. Why it happens explains the shared-vs-dedicated distinction and that registrations and subscriptions keep working, which is the usual way this presents. What you'll see gives the message and status, and points at [meta]log:push for channel-triggered pushes, where the publisher gets no response to inspect.

npm run validate:errors passes and protocol/errors.json was regenerated with npm run generate:errors (no hand edits).

Related

The realtime side is a separate change on a branch of the same name, which returns 103009 wherever it can tell push isn't available (a broker NO_ROUTE, no push queue keys, no push queue URLs in any region, and no AMQP hosts in Go's frontdoor). Publishing the page to ably.com/docs needs a follow-up PR against ably/docs to bump the submodule and regenerate.

🤖 Generated with Claude Code