Cloudflare Developers•2y ago

hmm yea

Jjohtso Doesn't seem to be visible on mobile 😢

ChaikaOP•7/13/24, 4:09 AM

did they hide it again? what the heck

johtso•7/13/24, 4:10 AM

Hmm well at least it's consistent

ChaikaOP•7/13/24, 4:12 AM

Jjohtso Doesn't seem to be visible on mobile 😢

ChaikaOP•7/13/24, 4:12 AM

did you scroll over enough?

ChaikaOP•7/13/24, 4:12 AM

funny because they did actually remove it at one time on mobile and we had to get them to readd it lol

johtso•7/13/24, 4:12 AM

Oh.. it's scrollable

johtso•7/13/24, 4:13 AM

The joys of minimalist interfaces without scrollbars.. there's no way to know something's scrollable

ChaikaOP•7/13/24, 4:14 AM

editor isn't much better if you stick to vertical w/ a small screen lol

johtso•7/13/24, 4:14 AM

Should probably stick to coding on my laptop then

johtso•7/13/24, 4:15 AM

Alright that's enough chasing runtime quirk rabbit holes, better get some sleep

CChaika hmm yea

johtso•7/13/24, 10:27 AM

I think the problem might be inconsistent behaviour in local dev.. I'm thinking Date.now() might not be returning 0.. will try and test in a bit

Puliczek•7/13/24, 12:09 PM

Guys i just tested deploy.workers.cloudflare.com . Why it is so hard to just deploy other users workers? Requires: Github Auth, CF Acc Id, CF Token (no template for that), Fork repository, enable github action, deploy worker. Lol its really hard to do that for someone who is not using cloudflare. Why it is not just simple url to do all that staff with CF auth? It works like that way for other clouds like Azure.

Puliczek•7/13/24, 12:10 PM

One first time i thought that its not official installation mode, but looking on url looks official.

Puliczek•7/13/24, 12:14 PM

I think the best workflow should be enter URL -> redirect to CF Auth Page -> Security check are you sure you want to install this worker Yes/No -> Worker deployed.

Puliczek•7/13/24, 12:17 PM

btw. deploy.workers.cloudflare.com looks really clear and easy to understand, but I think it is just overcomplicated.

zensin•7/13/24, 1:54 PM

This platform has a lot of weirdness of learning curve quirks. After having finally gotten all its features figured out and working, I can say it was all worth it. Now plz give us full version of vectorize and all will be as it should be.

johtso•7/13/24, 2:36 PM

@Chaika https://github.com/cloudflare/workers-sdk/issues/6261

CChaika yes, although a point worth mentioning is you have to add them as a custom hostn...

whois | API man•7/13/24, 2:44 PM

Wait, and how can I add them as Hostnames through API? Does cloudflare workers also provide this feature?

whois | API man•7/13/24, 2:47 PM

Wait I forgot it’s with the domain's zone, I think there is an API for it

whois | API man•7/13/24, 2:50 PM

Yup there is

whois | API man•7/13/24, 2:50 PM

It is

https://api.cloudflare.com/client/v4/zones/zone_id/custom_hostnames

https://api.cloudflare.com/client/v4/zones/zone_id/custom_hostnames

rdbisme•7/14/24, 12:10 PM

Hello, is there any plan to support ICMP? I woke up this morning with an idea to use a Worker as a status reporter for my server just to discover we can't send ICMP.

Or do you have any suggestions on how to achieve that.

I was thinking about using TCP sockets... But I fear my fail2ban will ban the worker IP

Walshy•7/14/24, 12:27 PM

No plans for ICMP, are you just trying to get latency to the server? Ideally you want to healthcheck services for status checkers

You can also make fail2ban not ban Workers but of course opens a bit of a hole.

WWalshy No plans for ICMP, are you just trying to get latency to the server? Ideally you...

rdbisme•7/14/24, 1:36 PM

Thanks for your answer. I just wanted a, cheap, external service to ping my server and send me a notification on Telegram.

rdbisme•7/14/24, 1:36 PM

I can indeed expose an http service on the server for healtcheck. But ping is easy, no need to deploy anything on the server

Rrdbisme I can indeed expose an http service on the server for healtcheck. But ping is ea...

Idle•7/14/24, 4:45 PM

does a icmp response always mean that the http service is running and healthy

CChaika yes, although a point worth mentioning is you have to add them as a custom hostn...

whois | API man•7/14/24, 10:53 PM

But is it a monthly pay or per usage both API and Fallback Workers? (CF SaaS)

IIdle does a icmp response always mean that the http service is running and healthy 🤔

rdbisme•7/15/24, 1:03 AM

No. But that's exactly what I want. I want to check the machine is up. Not the http service

Wwhois | API man But is it a monthly pay or per usage both API and Fallback Workers? (CF SaaS)

ChaikaOP•7/15/24, 1:16 AM

you pay monthly for each hostname over 100 at $0.10 per hostname per month, and you pay for whatever worker execution costs as well

Tony Leung•7/15/24, 4:43 AM

hi, is there anyway to detect request is from a known bot or not from cloudflare workers?

TTony Leung hi, is there anyway to detect request is from a known bot or not from cloudflare...

Zack0322•7/15/24, 6:32 AM

https://developers.cloudflare.com/bots/reference/bot-management-variables/#workers-variables

Cloudflare Docs

Bot Management variables · Cloudflare bot solutions docs

Bot Management provides access to several new variables within the expression builder of Ruleset Engine-based products such as WAF custom rules.

ZZack0322 https://developers.cloudflare.com/bots/reference/bot-management-variables/#worke...

Tony Leung•7/15/24, 8:41 AM

so this features is only available for enterprise? I'm using pro plan

Tony Leung•7/15/24, 8:56 AM

btw is there anyway to use image transform on images stored in a private R2 bucket in workers? I'm using Image Transform to add watermark to my image stored in R2. I'm implementing it by creating a presigned url to R2 objects and pass it through

fetch()

fetch()

with

cf.image

cf.image

option then push it to Cache API. This implementation has a problem when Cache API is a data-center-bounded and presigned url changed every time I sign it, so I cannot take the advantages if image delivery cache. Users from multiple regions who request the same resource will make image transform transforms more than one time. Especially when crawlers and bots trying to scrape my page , they make a huge hit to my wallet (4-5k images daily, approx 60k for 2 weeks). I'm trying to mitigate it by returning the original image if the request comes from well known bots or does not has

Referer

Referer

value be my page, but it does not solve the Cache problem above. Is there any elegant solution for this? Thank you.

Name With Exactly 32 Characters!•7/15/24, 12:11 PM

is there an adapter for

node:http

node:http

function for a worker

Original message was deleted

ChaikaOP•7/15/24, 1:00 PM

same eta with every other cloudflare product: :soonsoontm:

Vero•7/15/24, 1:17 PM

sometimesss:bibicat:

SSam Before I start a new project, does Cloudflare Workers support Turbo (for monorep...

JP•7/15/24, 1:59 PM

I had to hoist packages using pnpm with turbo to get my monorepo to work.

Ashley•7/15/24, 2:45 PM

I have a hopefully easy one. We're using Cloudflare to front our websites, but just for DNS - so we have

www-staging.foo.com

www-staging.foo.com

that is controlled by Cloudflare, and we're using a CNAME to

staging.actual-origin.com

staging.actual-origin.com

let's say, and it is WordPress but not sure that matters but naturally it's expecting requests from

www-staging.foo.com

www-staging.foo.com

in terms of the host.

All well and good, that works fine, now what we want to do is have a worker route intercept some traffic and set some additional headers on the response. I thought this would be pretty simple - have a worker route for

/the-page-to-intercept

/the-page-to-intercept

with a Worker, call

await fetch(request)

await fetch(request)

to hit the origin, and then modify the response before returning it from the worker (creating a new response really).

However, what's actually happening is downstream (

staging.actual-origin.com

staging.actual-origin.com

) is returning an error that suggests the host isn't being sent through - if I hit the downstream URL in the browser I get the same error (the error is site not found, effectively). Is this just some quirk of CNAMEs where it's not correctly hitting the origin, and is there any way to fix this / am I doing something dumb? I thought about setting the host header but that's not allowed with the Fetch API pretty sure

AAshley I have a hopefully easy one. We're using Cloudflare to front our websites, but j...

Hard@Work•7/15/24, 2:55 PM

So, to confirm, your CNAME record is proxied, right?

Ashley•7/15/24, 2:56 PM

It is

HHard@Work So, to confirm, your CNAME record is proxied, right?

Ashley•7/15/24, 2:57 PM

It works delightfully if I mix up staging and production URLs

But then the hosts aren't the same so it makes sense it works nicely (e.g. if I instead hit

www.foo.com

www.foo.com

from

www-staging.foo.com

www-staging.foo.com

inside the worker) but there must be some quirk I'm missing here

Ashley•7/15/24, 2:59 PM

I'm also hoping you might tell me it's hitting cache or something and I've since fixed the problem and when the cache clears it'll work

AAshley I'm also hoping you might tell me it's hitting cache or something and I've since...

Hard@Work•7/15/24, 2:59 PM

Does your origin have any kind of logging?

HHard@Work Does your origin have any kind of logging?

Ashley•7/15/24, 3:00 PM

It does but it's managed by an external agency, so it's not not something I directly have access to to debug. In theory, should the above work then?

AAshley It does but it's managed by an external agency, so it's not not something I dire...

Hard@Work•7/15/24, 3:01 PM

I'm just struggling a bit to make a mental model of which requests have what headers, tbh...

Ashley•7/15/24, 3:05 PM

So in the browser the user will hit

www-staging.foo.com

www-staging.foo.com

, which by default will proxy using a CNAME to

staging.actual-origin.com

staging.actual-origin.com

all works fine.

ow, if I insert a worker into the mix, in the worker I can see if I log the request it has

www-staging.foo.com

www-staging.foo.com

as the URL. I've tried both

fetch('www-staging.foo.com')

fetch('www-staging.foo.com')

and

fetch(request)

fetch(request)

to no avail - both give the same result, assuming there's no caching funny-business going on.

Ashley•7/15/24, 3:23 PM

Okay, I found out a key piece of info: the downstream (

staging.actual-origin.com

staging.actual-origin.com

) requires a specific HTTP header to be set - this is done by a transform rule when proxied - but naturally as the worker intercepts that transform rule isn't applied. So in theory I just need to set that, then it should work

Ashley•7/15/24, 3:36 PM

What's the best way to set that header? As it doesn't appear to be working, I'm doing:

const newRequestInit = {
                headers: {
                    'X-custom-header': 'some-val',
                }
            }

            const newRequest = new Request(
                request,
                new Request(request, newRequestInit)
            );

return await fetch(newRequest)

const newRequestInit = {
                headers: {
                    'X-custom-header': 'some-val',
                }
            }

            const newRequest = new Request(
                request,
                new Request(request, newRequestInit)
            );

return await fetch(newRequest)

Ashley•7/15/24, 3:45 PM

Back to the drawing board, the header is set as transform rules apply before a worker

CChaika you pay monthly for each hostname over 100 at $0.10 per hostname per month, and ...

whois | API man•7/15/24, 4:31 PM

I see, I have a question.

I activated Custom Hostnames (CF for SaaS), but it asks for a fallback origin, which is a record inside my DNS (A, AAAA or CNAME) but where do I point this record?

const newRequestInit = { headers: { 'X-custom-header': 'some-val', } } const newRequest = new Request( request, new Request(request, newRequestInit) ); return await fetch(newRequest)