Configuration

Endpoint Configuration

Beyond the URL and signing secret, each Endpoint has a configuration surface for retry timing, bookkeeping, custom headers, bulk onboarding, and temporarily holding deliveries without disabling it.

Custom retry schedule

Override the delay curve between retry attempts — independent from how many attempts happen at all (Max Retries, a separate setting).

Retry count and retry timing are deliberately independent

There's no automatic defaulting between "how many times" (Max Retries) and "how long between each" (this schedule) in either direction — you set each one on its own, and nothing infers one from the other. This follows the same model Svix uses.

1–10 intervals, strictly ascending

Each interval must be between 5 seconds and 7 days, and every interval must be longer than the one before it. The dashboard's editor is row-based — pick a number and a unit (seconds/minutes/hours/days) per row rather than typing raw seconds.

Asymmetric validation against Max Retries

A schedule shorter than your Max Retries count is rejected at save time — there's no well-defined delay for the attempts beyond the end of the list. A schedule longer than Max Retries is allowed, with a non-blocking warning: the extra intervals are simply never reached, since Max Retries still governs how many attempts actually happen.

Unset means the platform default

An Endpoint with no custom schedule uses the standard curve (30s → 5m → 30m → 2h → 8h → 24h). "Reset to Application Default" restores whatever your Application's own default schedule is set to — not necessarily the platform curve, if you've changed the Application-level default (see Roles & Access).

Metadata

Arbitrary key-value tags for your own dashboard organization — e.g. subscriber name, plan tier. Purely bookkeeping: never consumed by delivery or routing logic, and never exposed to the subscriber via the Consumer Portal.

Limits

Up to 20 keys, 4 KB total. String values only — not a place to stash structured or large data.

Custom static delivery headers

Static headers attached to every delivery to this Endpoint, e.g. X-Tenant-ID.

Plain text — not encrypted

This field stores values in plain text and is not the place for credentials or secrets. If your subscriber's endpoint requires an auth header to accept your requests, use the Endpoint Auth setting instead — that field is envelope-encrypted at rest.

Reserved names are rejected

You can't set webhook-id, webhook-timestamp, webhook-signature, webhook-event, webhook-test, Content-Type, or Host — those are platform-controlled. If Endpoint Auth is also configured, its header wins and a conflicting custom header is rejected at save time.

Do not enter credentials or secrets here. This is the same warning you'll see in the dashboard when setting a custom header — it's a real distinction, not boilerplate.

Bulk import via CSV

Useful for migrating a list of endpoints from another provider instead of creating them one at a time.

Format

CSV with columns url,name,description. Up to 500 rows per import.

Every row goes through the same checks as a single Endpoint

HTTPS-only, SSRF blocklist, and challenge-response verification all apply per row — bulk import is not a bypass of those checks.

Partial success

One bad row (e.g. an http:// URL) doesn't sink the whole import — you get a per-row report of what succeeded and what failed and why.

Dashboard → Endpoints → Import Endpoints

Pause & resume

Hold all deliveries to an Endpoint temporarily — without disabling it, and without losing retry progress.

Pause (confirm dialog) → deliveries held, retry budget frozen → new events during pause also held → Resume: redeliver or discard

Distinct from Disabled and Circuit Broken

Disabled implies the Endpoint is decommissioned. Circuit Broken is an automatic response to repeated failures. Pause is a deliberate, admin-initiated hold meant to last hours or days — held deliveries don't consume any of the Endpoint's retry attempts while paused, so pausing for a day doesn't risk pushing anything into the Dead Letter Queue.

Two ways to resume

Resume & Redeliver immediately sends every held event, continuing from whatever retry attempt it was on before the pause (no reset). Resume & Discard clears the held events without delivering them — both choices are recorded in the Endpoint's Pause History.

Pause History (its own tab on the Endpoint Detail page) is a full read-only audit trail — who paused/resumed, when, and whether resume redelivered or discarded — visible to every role including Monitor.

← Roles & Access Signature Verification →