Logs

Operator page

Developers and owners only. Log lines can hold member details.

What the server has been doing. Useful for pinning down when something broke, and for following one member, one webhook or one stuck event through everything the server did with it.

Newest first.

Click anywhere, or press Esc, to close.

Filtering

Level, search, and the Live or History source.

Click anywhere, or press Esc, to close.

• Level — start at ERROR. Widen only if you find nothing.
• Search — type an email address or an identifier to follow one thing through.
• Live or History — Live is a short in-memory tail, cleared when the server restarts. History is the durable record, kept 30 days, and 180 days for anything touching money. History is the server's own record, so it is the same in every environment, sandboxes too.
• Auto-refresh ticks every three seconds. In History it adds only the new lines at the top and leaves the rest where they are. Turn it off when you are reading carefully.
• Show older, under the list in History, loads the next 300 lines further back.
• A search matches exactly what you type. 100% finds "100%", not "1000".

Following one thing

In History, every line shows small grey chips underneath: the member it concerned, the event type (a Stripe webhook such as invoice.paid, or an outbox job such as MemberTagResync), the outbox event, the Stripe account, the admin who clicked, and the request. Click a chip and the screen narrows to that one thing; the bar above the list says what you are following, with an Open member link when it is a member. When it is an admin, What this admin did opens Audit for them.

Logs are for finding out why something went wrong. For who did what, use Audit. Audit keeps every change, is never cut short, and says who made it.

You can also arrive here already narrowed:

• Member page → Logs (top right) shows everything the server logged for that member: their own requests, the webhooks about their subscription, the sync jobs that ran for them.
• Health → Dead letters → Logs on a row shows every delivery attempt of that one event, which is usually where the real error message is.

Nothing under a chip means the line was not inside a request, webhook or delivery — a startup banner, a health tick — and there is nothing to follow.

What to do with what you find

Copy the timestamp and the message and give both to Thomas — or, better, the page address while you are following the member or event, so he opens the same view. Do not act on a log line on its own. Health and Dead letters tell you whether something is actually stuck.

Health has a log history row too. It says how big the log table and the Audit trail are and how far back the logs reach. It only turns amber if either grows past its limit, which means a retention setting or a log storm needs a look, never that data is lost. Old lines are cleared by a daily job, a little at a time, so the server never slows down for it.