I cannot literal get what that "rollover" mean

Pushed a fix, I'll monitor the errors now. Thanks @zszszsz

I never thought it could be like an infinite loop.

I was guess it could be scientific 🙂

I ran into that issue earlier today but I thought this condition will never hit so added a check. Let's see

Don't think I've seen this asked, I have a zone that I want to have a Worker on, apart from one subdomain where I plan to use Pages. I think I've figured out 2 ways of doing it, but is there any reason to choose one or the other? Custom Domain:

worker.example.tld -> worker
*.example.tld CNAME worker.example.tld
except.example.tld CNAME project.pages.dev

Routes: in worker config

except.example.tld/* -> no worker selected
*.example.tld/* -> worker

BTW, I absolutely love DX of Workers ❤️ Props to the entire team 👏 Going from fix to production is <5s.

Second option will probably work best. Two things to note,

except.example.tld/*

has to be set from the dash, since it doesn't pertain to any specific Worker. Secondly, you need some kind of DNS record to actually cover

*.example.tld

. I recommend an

AAAA

record pointing toward

100::

I don't think you can do a wildcard cname

I've been using wildcard CNAMEs since I switched to CF :D it's super helpful

sweet tysm, I have

*.example.tld CNAME example.tld

and it's proxied, everything seems to work including the bypass

Hello all

I am trying to implement a reverse proxy of WebSocket in cloudflare worker. What I want : Client (https://) => Worker (As wss:// request) => My Origin (As ws:// request). How to achieve this?

Hello everyone. We'll structure our architectural rework trough workers. To avoid code splitting, is it common to have only one worker handle all (graphql, rest api, ui serving, cache via r2 or d1, etc..) or really better to segment one service = one worker and let them communicate on edge ?

In general, I'd recommend cramming as much into a single Worker as possible, as once it is warm, every route is warm, while if you split it up, only the gateway will be warm.

Thank you for the answer. It makes sense. We'll see if we can fit it all into 5MB... And having one worker ping the others every minute, is that a bad practice?

If you are paying for it, they don't care, but it is kind of wasteful

enable minifying too

Are analytics engine datasets limited to one per account during beta?

Can you expand on this? I thought cloudflare workers didn't need to be "warmed" because of the custom runtime environment?

There is the no cold start system, but that only works on the first Worker triggered

If you need to trigger multiple Workers, then it might still need to cold start

Woah! Didn't know that, thank you!

is there a way workers can pass data to each other via service bindings that can't be spoofed by normal http clients? specifically, could i set some secret

isFromWorker: true

property that allows me to skip the normal authentication steps i have on another worker? is there a security reason i shouldn't do this?

You can add some kind of flag to the request.cf object, which can then be passed around service bindings.

awesome. and this can't be spoofed by someone calling https://myworker.example.workers.dev directly?

Nope

i love cloudflare

im working on an app where I connect to a remote server over ws and stream certain updates back to the client. I assume the worker gets evicted After the CPU time limit, but how would I go about having it open for longer periods (hours)? Do I use a DO or can I ping my worker (if so what is preferrable?)

A Durable Object would probably be best, since you aren’t always guaranteed to hit the same Worker when you ping