What is your actual setup?

johtso•6/23/24, 4:50 PM

a simple _worker.js in a pages project, set up with a custom domain foo.example.com. it successfully receives requests, including a different webhook from the same source. the other webhook (a body full of XML) never gets to my worker

WalshyOP•6/23/24, 4:50 PM

yeah check security events, if nothing in there then cf seems to be fine

johtso•6/23/24, 4:51 PM

I don't think I have anywhere to check security events

WalshyOP•6/23/24, 4:51 PM

oh right external dns

WalshyOP•6/23/24, 4:51 PM

then yeah this isn't a cf issue

WalshyOP•6/23/24, 4:51 PM

i'd chat to that caller that isn't working

johtso•6/23/24, 4:52 PM

it just seems impossible that it isn't a cf thing.. because if I cname foo.example.com to a different domain the request comes through

WWalshy then yeah this isn't a cf issue

johtso•6/23/24, 4:54 PM

do you say this because cf wouldn't be applying any firewall in that case?

WalshyOP•6/23/24, 4:57 PM

yup

johtso•6/23/24, 4:57 PM

oh crazy

johtso•6/23/24, 4:58 PM

I'm literally stumped then. I don't see how the sender of a http request would fail to send that request only if the domain they are sending to ends up going to a cloudflare worker

johtso•6/23/24, 5:00 PM

the fact that it also doesn't work when transparently tunnelled through a netlify site makes me think that cf must not be allowing the request through based on what it looks like, rather than it being a problem with the sender

johtso•6/23/24, 5:00 PM

but I appreciate the fact that I don't control the sender makes it hard to debug

johtso•6/23/24, 5:00 PM

hence spending a whole day trying to figure this out

Nob•6/23/24, 6:48 PM

I couldn't find a pricing calculator for the standard pricing model, so I created one:
https://standard-model-pricing-calculator.pages.dev/

johtso•6/23/24, 8:03 PM

@Walshy tried putting the same code on deno.deploy and forwarding requests to that and it works fine

WalshyOP•6/23/24, 8:05 PM

Still doesn't point to cf sadly haha

WalshyOP•6/23/24, 8:05 PM

We've seen sites do weird things when going to CF IPs

WalshyOP•6/23/24, 8:05 PM

I can't think of anything on CF that would prevent it, so if you're sure your code isn't invoked. I'd talk to that sender

WWalshy I can't think of anything on CF that would prevent it, so if you're sure your co...

johtso•6/23/24, 8:14 PM

yeah, I really don't know enough about networking and http to know where a problem could come from

johtso•6/23/24, 8:18 PM

@Walshy but the fact that the request is going via a netlify rewrite should mean that to the requester it makes no difference where the response ultimately comes from, and it's impossible to tell

johtso•6/23/24, 8:18 PM

unless I'm misunderstanding how it works

WalshyOP•6/23/24, 8:18 PM

Yes, to the caller they'd see netlify. So, does netlify have any info?

WWalshy Yes, to the caller they'd see netlify. So, does netlify have any info?

johtso•6/23/24, 8:20 PM

nope, another black box, but I've eliminated netlify from the equation because I get the same behaviour if I use cloudflare pages

WalshyOP•6/23/24, 8:21 PM

Are you able to add headers to the caller?

WWalshy Are you able to add headers to the caller?

johtso•6/23/24, 8:25 PM

can't do anything with the caller, just had a look and don't think netlify lets you do anything with headers when rewriting.. what I was thinking of doing was putting a simple function on deno deploy that would forward the request to my worker, then if that doesn't work I'd be able to see what happens

WalshyOP•6/23/24, 8:26 PM

Yeah even having a status code would help a lot

johtso•6/23/24, 8:38 PM

ok, moment of truth

johtso•6/23/24, 8:39 PM

@Walshy bingo! It's returning 403

WalshyOP•6/23/24, 8:41 PM

ok so something is blocking it :Hmm:

WalshyOP•6/23/24, 8:41 PM

send me the pages.dev

WalshyOP•6/23/24, 8:41 PM

i need to sleep but i can take a look for 5 mins

johtso•6/23/24, 8:44 PM

thankyou!!

oof2win2•6/23/24, 10:30 PM

is generating IDs with nanoid very expensive computationally for workers? some requests taking nearly 400ms of cpu time seems a bit excessive

oof2win2•6/23/24, 10:32 PM

like i was generating at most 40 ids per request which doesn't seem that bad to me

oof2win2•6/23/24, 10:36 PM

running a benchmark of generating the ID gives me a p999 of 525ns525ns and i don't think that adds up here (i'm also running database queries and sending external requests but that's it and those go into wall time)

safah222•6/23/24, 11:28 PM

Hello guys

safah222•6/23/24, 11:29 PM

I want to detect if the traffic was comming from verified bot , (( if verified bot )) --> do that

safah222•6/23/24, 11:29 PM

how I can do that with cloudflare worker , I have busniess plan

Rowin•6/23/24, 11:33 PM

I believe it's

request.cf.botManagement.verifiedBot

request.cf.botManagement.verifiedBot

Rowin•6/23/24, 11:33 PM

https://github.com/cloudflare/workerd/blob/bd09e5b12988b76277835b6f75100833f7f88158/types/defines/cf.d.ts#L412

AngusMa•6/24/24, 5:17 AM

Does

wrangler.toml

wrangler.toml

in the code editor from the dashboard work?
I added

tail_consumers = [{service = "tail"}]

tail_consumers = [{service = "tail"}]

in the file but it didn't work.
I'm sure my tail worker is called

tail

tail

and the worker has

tail(event)

tail(event)

function.

AAngusMa Does `wrangler.toml` in the code editor from the dashboard work? I added `tail_c...

ItsWendell•6/24/24, 8:39 AM

Tail workers are quite hard to debug, you can try the dashboard logs and potentially create an endpoint in that tail worker that mocks some of the tail events to test if your worker is doing what you expect to do.

ItsWendell•6/24/24, 8:46 AM

Question in regards to service bindings, reading environment variables and latencies. I've a worker called Edge Config that exposes 2 RPC functions,

setSecret

setSecret

and

getSecret

getSecret

setSecret

setSecret

updates the secrets of itself using the Cloudflare REST API.

getSecret

getSecret

returns the environment secret binding. I expected it to be close to 0 latency for doing an RPC call to Edge Config and returning a secret, but it ends up being around

147ms

147ms

for the RPC call to get an environment variable. The whole reason I have this setup is to have close to 0 read latency to get some dynamic config values.

It's a Pages -> Workers service binding, both have smart placement turned off. Is this latency normal, how can I further reduce this? Any other recommendations to have dynamic config with close to 0 latency for reads?

IItsWendell Tail workers are quite hard to debug, you can try the dashboard logs and potenti...

AngusMa•6/24/24, 3:02 PM

I added this in the

fetch

fetch

function to test and it works well, so I think the problem isn’t on the

tail

tail

function

await this.tail({
  test:true
})

await this.tail({
  test:true
})

AngusMa•6/24/24, 4:33 PM

What is https://internet-measurement.com/ ? It keeps requesting to my worker on different ports like 8443, 8880 and 2086

AAngusMa What is https://internet-measurement.com/ ? It keeps requesting to my worker on ...

Nob•6/24/24, 4:47 PM

well you can read maybe

AngusMa•6/24/24, 4:49 PM

How do I block them?

AAngusMa What is https://internet-measurement.com/ ? It keeps requesting to my worker on ...

Chaika•6/24/24, 4:52 PM

there's tons of companies and projects which scan the entire web and ip space, not particularly harmful,espec not to a worker which isn't going to be vuln to anything

Chaika•6/24/24, 4:52 PM

if you want to try to block those, disable your workers.dev for that worker and use Custom Domains/Routes to bind your worker to your domain, and you can use a rule like

not cf.edge.server_port in {80 443}

not cf.edge.server_port in {80 443}

to block all traffic to non-standard ports

What is your actual setup?

Similar Threads

Similar Threads

Similar Threads