Add a centralised table with the wall time limits for different types of worker invocations #27247
Description
Proposed changes
First clarify that this is not about CPU time in the slightest, CPU time is the standard answer: configurable up to 5 minutes, default yadayada…
Then provide this table which I wish I had
| Worker invocation source | Wall Time Limit |
|---|---|
| Workflows (outside step) | 30 minutes |
| Workflows (inside step) | Unlimited (default configuration 5 minutes) |
| Queue handlers | 15 minutes (up to 100 retries possible though) |
| Incoming external HTTP request | Unlimited until the caller disconnects, then 30 second extension possible using ctx.waitUntil(). Can rarely we evicted when CF does updates, 30s grace period in that case. |
| Durable objects RPC calls | Unlimited long as the caller stays connected |
| Cron triggers | 15 minutes |
| Durable object alarm handlers | 15 minutes |
Subject Matter
Assuming a worker is waiting for a rate limit before doing its work, where's the best place put the worker architecturally
Content Location
https://developers.cloudflare.com/workers/platform/limits/
Additional information
Would be great to centralise, rn this needs visiting like 6 docs pages + some aren't documented at all