I'd look at your code honestly, it's been working great for me both in personal and work projects - Cloudflare Developers

© 2026 Hedgehog Software, LLC

More

Communities Docs About Terms Privacy

Walshy

WWalshy I'd look at your code honestly, it's been working great for me both in personal ...

sathoro

S

sathoro•5/28/23, 8:19 AM

I think it's mostly when manually doing sentry.captureExceptionsentry.captureException and also sometimes when there is an error deep in a library then there is no context

sathoro

S

sathoro•5/28/23, 8:22 AM

actually I have a great example from yesterday because it is from the same webhook route

sathoro

S

sathoro•5/28/23, 8:23 AM

missing stack trace (only shows the library code)

sathoro

S

sathoro•5/28/23, 8:24 AM

actually I don't want to show screenshot of the second one but its code in the same route itself that manually calls throw new Errorthrow new Error and the full stack trace is there

sathoro

S

sathoro•5/28/23, 8:25 AM

the other issue I get is with errors sometimes not de-duplicating themselves. is this something I need to config differently?

sathoro

S

sathoro•5/28/23, 8:27 AM

also with some DO errors I sometimes get errors like this with zero stack trace. I think itty-durableitty-durable is probably losing the context somehow since they use a proxy object or something. this one does get de-duplicated though

PencilNavigator

P

PencilNavigator•5/28/23, 9:13 AM

guys all my workers just got about 200-2k requests all at 9:00 together

PencilNavigator

P

PencilNavigator•5/28/23, 9:13 AM

what kind of wierd issue is this

PencilNavigator

P

PencilNavigator•5/28/23, 9:13 AM

by all i mean all worker service deployed in my account

Ian

I

Ian•5/28/23, 3:06 PM

What's the best way to store logs from workers thonk

thonk

, current options seems to all require either R2 or an external destination to store them, with each storing costing a R2 write

Walshy

W

WalshyOP•5/28/23, 3:10 PM

logpush is the way and r2 is one option. You can also use others though like Backblaze which would be cheaper

They're batched and sampled (depending on your option) so you don't pay per request but yeah, it will likely cost a write at wherever you store

Walshy

WWalshy logpush is the way and r2 is one option. You can also use others though like Bac...

Ian

I

Ian•5/28/23, 3:17 PM

I have no enterprise for logpush tho

Ian

I

Ian•5/28/23, 3:18 PM

Ahh there is workers trace events logpush

Walshy

W

WalshyOP•5/28/23, 3:18 PM

Workers Logpush doesn't require ENT

Walshy

W

WalshyOP•5/28/23, 3:18 PM

just Workers Paid

Ian

I

Ian•5/28/23, 3:18 PM

FrogeThumbsUp

Walshy

W

WalshyOP•5/28/23, 3:34 PM

It ca run for a long time

Walshy

W

WalshyOP•5/28/23, 3:34 PM

It can only use 30 seconds of CPU time (or 15 min in cron triggers)

Walshy

W

WalshyOP•5/28/23, 3:35 PM

the wall clock time is what is "unlimited", it can run for as long as it can

Walshy

W

WalshyOP•5/28/23, 3:35 PM

the limit is on the CPU time just like Bundled

Walshy

W

WalshyOP•5/28/23, 3:35 PM

agree the way we word/represent it is confusing

Walshy

WWalshy It can only use 30 seconds of CPU time (or 15 min in cron triggers)

Isaac McFadyen

I

Isaac McFadyen•5/28/23, 3:36 PM

Think you mean wall time?

Walshy

W

WalshyOP•5/28/23, 3:36 PM

no?

Isaac McFadyen

I

Isaac McFadyen•5/28/23, 3:37 PM

As far as I understood it, it was allowed 30s of wall time? (but realistically unlimited - 30s is the minimum they'll give you for runtime restarts)

Walshy

W

WalshyOP•5/28/23, 3:37 PM

nah 30 seconds of CPU time

Isaac McFadyen

I

Isaac McFadyen•5/28/23, 3:37 PM

Oh, interesting.

Walshy

W

WalshyOP•5/28/23, 3:37 PM

the 30 second buffer is in case of runtime restarts

Walshy

W

WalshyOP•5/28/23, 3:37 PM

but it could run for an hour if it wanted

Isaac McFadyen

I

Isaac McFadyen•5/28/23, 3:37 PM

Yeah I think I'm getting those numbers confused

Walshy

W

WalshyOP•5/28/23, 3:38 PM

(we do a bad job at explaining all this notlikemeow

notlikemeow

)

Erisa

E

Erisa•5/28/23, 3:38 PM

its implied on here with the cpu runtime 30s and then duration unlimited with an asterisk leading to the restarts disclaimer

semicolondev

Ssemicolondev Now I'm getting 502

Walshy

W

WalshyOP•5/28/23, 3:39 PM

A 502 from a Worker isn't something I'd expect.

If you ran out of CPU time you'd get a proper error. 502 comes from us trying to hit your origin

Walshy

W

WalshyOP•5/28/23, 3:40 PM

Are you trying to hit your origin / an external service? If so, I think the problem is from there

Walshy

W

WalshyOP•5/28/23, 3:40 PM

1 second is nothing

Erisa

E

Erisa•5/28/23, 3:41 PM

even our default http timeout is 100 seconds

Walshy

W

WalshyOP•5/28/23, 3:41 PM

OpenAI's internal APIs and external APIs are different

Erisa

E

Erisa•5/28/23, 3:41 PM

but if its actively streaming data that won't be hit either

Walshy

W

WalshyOP•5/28/23, 3:41 PM

Do you have a curl we can run to see the behaviour?

Erisa

E

Erisa•5/28/23, 3:42 PM

and also yeah my first thought for 502 from a worker is that its being thrown from an upstream origin

Walshy

W

WalshyOP•5/28/23, 3:50 PM

yep

Walshy

W

WalshyOP•5/28/23, 3:50 PM

No, it won't be in catch