Long wall times and short response times
Hey folks, we're seeing a worker with long wall times but very short response times. We're not using
context.wailtUntil anywhere, and this is a fairly simple request flow through Hono (which we've now logged the living daylights out of, including in Hono itself). We have a couple of Supabase REST API calls that complete quickly, and we seem to get to the point where we're returning the response within 200-500ms, but we end up with a CPU time of maybe 10-20ms and a wall time of 30s. The worker then gets cancelled, but appears to have correctly returned a response—all very mysterious. Finally, I've tried to hit everything with the dangling promise lint rules in @typescript-eslint and biome, but nothing pops out.
With all that said, I'm wondering if anyone knows what might be good to look out for to understand where the wall time is being spent—how do others go about debugging these scenarios?1 Reply
This… Appears to have gone away on its own. Which is not a particularly satisfying resolution… But is a resolution nonetheless.