Publishing
Retry and backoff explained
When a publish hits a transient error, Signal retries with increasing delay up to a bounded number of attempts before marking it failed.
Last updated June 14, 2026
Not every publishing error is permanent. A network timeout or a temporary platform 5xx usually succeeds on a second try. Signal distinguishes transient errors (worth retrying) from permanent ones (a malformed request, a revoked token) and only retries the transient ones.
Retries are bounded and use backoff — each attempt waits longer than the last — so a struggling platform isn't hammered and a genuinely broken item doesn't loop forever. After the attempt budget is exhausted, the item is marked failed and surfaced to you.
Retried vs not retried
- ·Retried: timeouts, transient 5xx, token-refresh hiccups.
- ·Not retried: permanent rejections like a malformed payload or a policy block — retrying wouldn't help.
Common mistakes
- !Manually retrying a permanently-failed item without fixing the cause — it will fail the same way.