Thanks. Based on the docs and info I found I assumed the same when I changed it a few weeks ago. Tho

Thanks. Based on the docs and info I found I assumed the same when I changed it a few weeks ago. Though our unbound request count is now at 10 million and standard at 150. So I suspect something might have gone wrong. Guess I'll try a manual redeploy

Wwarflash Thanks. Based on the docs and info I found I assumed the same when I changed it ...

Cyb3r-Jak3•11/18/23, 2:14 PM

If you have unbound in your wrangler file then you’ll need to change it there as well

CCyb3r-Jak3 If you have unbound in your wrangler file then you’ll need to change it there as...

warflashOP•11/18/23, 3:03 PM

Ah yes, that was it. Thanks! Surprisingly it does seem like the standard model is going to be like 50% more expensive than unbound even though the calculator sound it'll be 70% cheaper. Odd

anziez•11/18/23, 7:07 PM

why can't I install wrangler D:

Aanziez why can't I install wrangler D:

Hello, I’m Allie!•11/18/23, 8:01 PM

Are you installing it globally?

kian•11/18/23, 8:26 PM

kian•11/18/23, 8:26 PM

Where's the issue? meowthinking

kian•11/18/23, 9:13 PM

That'll be if you put in lon, lat

kian•11/18/23, 9:13 PM

It should be lat, lon

Isaac McFadyen•11/18/23, 10:14 PM

Keep in mind it's using MaxMind's GeoIP database, so it's entirely possible for it to be inaccurate yes.

Unsmart•11/18/23, 10:15 PM

Well I would assume if they give a city the long/lat will always actually be in the city but the actual location it says might not be accurate yeah fastnod

UUnsmart Well I would assume if they give a city the long/lat will always actually be in ...

Isaac McFadyen•11/18/23, 11:11 PM

Yeah.

Isaac McFadyen•11/18/23, 11:11 PM

Like, for example, I use Starlink which is CGNAT. So the IP is Toronto (not where I am, about 2 hours away)

Isaac McFadyen•11/18/23, 11:11 PM

The lon/lat is inside Toronto but I'm not in Toronto lul

GuillaumeLakano•11/19/23, 1:13 PM

Hi! Is anyone already tried to compare script code in Javascript in a Workers with the same code in WASM to known if there is interest to code it for example in AssemblyScript ( a variant of TypeScript that generate a WebAssembly code ) ?

GGuillaumeLakano Hi! Is anyone already tried to compare script code in Javascript in a Workers wi...

Hello, I’m Allie!•11/19/23, 1:29 PM

If you are already writing in (Types)Script, then I wouldn't bother rewriting it in AssemblyScript(unless you are doing it as a learning exercise).

HHello, I’m Allie!If you are already writing in (Types)Script, then I wouldn't bother rewriting it...

GuillaumeLakano•11/19/23, 1:44 PM

I'm not used to do TypeScript, but as I've just discover this AssemblyScript that directly return the .wasm file, may be this could be a good change for my workers

GGuillaumeLakano I'm not used to do TypeScript, but as I've just discover this AssemblyScript tha...

Hello, I’m Allie!•11/19/23, 1:46 PM

Yeah, as a way to learn about how WASM works, it can be a great excercise. I just mean that where possible, raw JS may be preferred over WASM, due to performance reasons. With something like

workers-rs

workers-rs

, you get the benefits of Rust's borrow-checker(and all the other goodies), but if you are using something like AssemblyScript, then I wouldn't see much benefit there(performance-wise, anyway).

HHello, I’m Allie!Yeah, as a way to learn about how WASM works, it can be a great excercise. I jus...

GuillaumeLakano•11/19/23, 1:49 PM

Thank you for the precision

Bulbasaur•11/19/23, 9:31 PM

Are there plans for smart placement to support multiple "smart locations"? For example if you have a standard replicated database with 6 locations around the world, workers could hypothetically determine that it should use smart placement in those 6 locations, and not restrict itself to one of them

Bulbasaur•11/19/23, 9:32 PM

Today, smart placement unfortunately feels like a not great compromise between "always completely global" and just back to being "single region"

Bulbasaur•11/19/23, 9:51 PM

Even if it wasn't necessarily completely automated for every possible ingress or whatever, something like "evaluate smart placement per-country" would be a massive improvement

mooon•11/19/23, 10:41 PM

hey there, if i were to create a worker that responds with a pretty large image

, would it be automatically streamed? are there any automatic optimisations cloudflare does?

Mmooon hey there, if i were to create a worker that responds with a pretty large image ...

Bulbasaur•11/19/23, 11:15 PM

as long as you don't buffer the image in memory, then it will be streamed

Bulbasaur•11/19/23, 11:15 PM

the key is to avoid reading the

Response

Response

body of any request to the image you make into memory

Bulbasaur•11/19/23, 11:18 PM

So avoid calling all of these basically, except

Response.clone

Response.clone

is fine -

Bulbasaur•11/19/23, 11:19 PM

If you do buffer the body in the worker, then the worker will first wait all the time to fetch the image itself and load it into memory, then respond with it, slowing it down.

Bulbasaur•11/19/23, 11:19 PM

On the client side (e.g in a browser), whether or not the worker buffers it in memory or not shouldn't have any impact, it can still be streamed as normal

TheKaiser•11/20/23, 12:59 AM

does anyone have advice on how to do typing when using itty-router-openapi? I've been using zod's infer() to create a type but I'd love to have typing for the "data" argument in handle, all the documentation just sets it to "any"

Colin•11/20/23, 1:21 AM

I'm encountering some confusing Worker runtime behavior, would love if anyone more seasoned could chime in

basically:

We have a WebSocket opened to a Worker (not a DO, just plain worker)

When the client closes the WebSocket first, the Worker emits a

WebSocket disconnected without sending Close frame.

WebSocket disconnected without sending Close frame.

The Worker emits a

Network connection lost.

Network connection lost.

This is counted as a "Script Threw Exception" from Cloudflare's dashboard/api/logpush

My understanding was that the WebSocket disconnect + network connection lost just indicate the client going away + the isolate terminating, which is expected behavior. However, why is this counted as an exception instead of a "Client disconnected" (what I'd expect + what standard HTTP requests do when the client disappears) or anything else?

(Why do we care in the first place? We have some monitoring + logpush configured only for errors, and these "false errors" create noise in those systems)

kian•11/20/23, 1:28 AM

The

exception

exception

outcome is for any unhandled exception, which

Network connection lost

Network connection lost

would be. The errors around WebSockets aren't ideal, and https://github.com/cloudflare/workerd/pull/804 hopes to improve them.

Kkian The `exception` outcome is for any unhandled exception, which `Network connectio...

Colin•11/20/23, 1:53 AM

I guess I'm confused how to prevent

Network connection lost

Network connection lost

here - I assume that's what error messages generated by that PR should tell us?

It looks like that PR has been untouched for 3 months-ish, do you know what the status is or should I leave a comment?

nEEk•11/20/23, 8:35 AM

I'm building a ticketing system and need puppeteer for pdf screenshots. Any eta on when @cloudflare/puppeteer will be available to use in workers? Super keen for this since it will mean I can migrate to Cloudflare from vercel, render.com and aws.

.gwapes•11/20/23, 12:03 PM

Hey, I’m using my own express server hosted on my vps to handle request for database updating, and then making the requests on my worker to update things in my database (mongodb)

What is a good way to authorise this? To make sure no one can make random PATCH requests for example? Atm, I’m just using an authorisation header with a random string and then checkin that on my express server. Is there a better way?

.gwapes•11/20/23, 12:04 PM

It’s only for something small so it doesn’t matter really but it’s more of a learning thing

..gwapes Hey, I’m using my own express server hosted on my vps to handle request for data...

Hello, I’m Allie!•11/20/23, 12:05 PM

Create a DNS record for it, then put the Worker on top.

Hello, I’m Allie!•11/20/23, 12:06 PM

Then all requests hitting your origin have to pass through your Worker first, where you are free to accept/deny

.gwapes•11/20/23, 12:07 PM

So I’d check if it was a patch, and the route was the specific route that would be on my express server?

..gwapes So I’d check if it was a patch, and the route was the specific route that would ...

Hello, I’m Allie!•11/20/23, 12:08 PM

That, or just never pass a request back to your origin unless it is you making a DB request

.gwapes•11/20/23, 12:08 PM

And how would I make sure it was my worker

.gwapes•11/20/23, 12:09 PM

Also, is it not very optimal to just have an authorisation header?

..gwapes And how would I make sure it was my worker

Hello, I’m Allie!•11/20/23, 12:10 PM

You could use a Tunnel, and then not expose the server at all

..gwapes Also, is it not very optimal to just have an authorisation header?

Hello, I’m Allie!•11/20/23, 12:11 PM

You can, but with this method, you literally can't hit the origin at all(barring any issues on CF's side), while with the Auth Header method, a bad actor can still hammer your DB with requests, even if they aren't authorized

.gwapes•11/20/23, 12:12 PM

I’m a bit lost with the last part there. I only actually make the db update if the auth headers value is that “random string”, so it won’t do anything if it’s not right

HHello, I’m Allie!You could use a Tunnel, and then not expose the server at all

.gwapes•11/20/23, 12:12 PM

Is there any docs/examples on tunnelling?

..gwapes I’m a bit lost with the last part there. I only actually make the db update if t...

Hello, I’m Allie!•11/20/23, 12:13 PM

Yes, but in this case, if I want to take your DB out of commision, I can just send random requests to your server. The changes won't be able to be applied, but because your server is busy handling my requests, it won't have time to handle your valid changes

..gwapes Is there any docs/examples on tunnelling?

Hello, I’m Allie!•11/20/23, 12:13 PM

https://developers.cloudflare.com/cloudflare-one/connections/connect-networks/get-started/create-remote-tunnel/

Create a remotely-managed tunnel (dashboard) · Cloudflare Zero Trus...

Follow this step-by-step guide to get your first tunnel up and running using Zero Trust.

HHello, I’m Allie!Yes, but in this case, if I want to take your DB out of commision, I can just se...

.gwapes•11/20/23, 12:13 PM

But that would be an express like thing, right? Rather then a db thing. Because it’s just busy not doing anything but handling the wrong requests

.gwapes•11/20/23, 12:14 PM

So it would be better to return wrong requests from my cf worker where everything goes through anyway?

Thanks. Based on the docs and info I found I assumed the same when I changed it a few weeks ago. Tho

Similar Threads

Similar Threads

Similar Threads