For Hyperdrive questions I recommend asking in #hyperdrive-beta where the team is active, individual

For Hyperdrive questions I recommend asking in #hyperdrive-beta where the team is active, individual teams for products don't always monitor channels like this or #workers-help

EErisa For Hyperdrive questions I recommend asking in #hyperdrive-beta where the team i...

Mitya•12/14/23, 11:50 AM

Aha sorry - I asked here because I figured it was more of a Wrangler/TOML question, but will ask in Hyperdrive.

EErisa And the compatibility flag to unlock the `node:` stuff is just ```toml compatibi...

bun•12/14/23, 12:55 PM

deploying another worker, this didn't work smide

bun•12/14/23, 12:56 PM

returned the same node_compat error, only fixed when i put node_compat = truenode_compat = true in

kian•12/14/23, 12:56 PM

nodejs_compat

nodejs_compat

only covers a subset of modules and they must be imported with the node:node: prefix -

node_compat

node_compat

is a set of polyfills added at build-time which covers a lot more modules.

bun•12/14/23, 12:56 PM

bun•12/14/23, 12:56 PM

bit confusing KEKW

kian•12/14/23, 12:58 PM

nodejs_compat

nodejs_compat

is a new-ish push towards providing NodeJS built-ins at runtime whereas

node_compat

node_compat

is fairly old - once

nodejs_compat

nodejs_compat

provides enough coverage to remove the need for

node_compat

node_compat

then

node_compat

node_compat

would be removed.

kian•12/14/23, 12:59 PM

One of the big issues w/

nodejs_compat

nodejs_compat

is the requirement of a node:node: prefix when importing the module, but there is work to have Wrangler handle that automatically for you.

Antonio32A•12/14/23, 1:56 PM

What's the maximum duration a TCP socket can be kept alive? 30s?

sathoro•12/14/23, 2:04 PM

just started getting this all of a sudden for a particular domain when calling

fetch

fetch

. I've got internet, am able to access it in my browser, have restarted the dev server, reset my internet, etc

only seems to be one particular domain

sathoro•12/14/23, 2:14 PM

seems like the server has started closing the connection early. its bizarre though, because it is happening no matter what my IP is but there are no issues when deployed in prod

rob•12/14/23, 2:44 PM

What's Hyperdrive do really ?

kian•12/14/23, 2:48 PM

Connection pooling and caching for Workers.

kian•12/14/23, 2:48 PM

Workers are short-lived instances so you can't really do connection pooling with them - Hyperdrive maintains regional pools of connections.

Unsmart•12/14/23, 4:32 PM

You could maybe proxy all requests through #durable-objects and manage the throughput there but probably more effort and costly than just doing backoffs instead.

Salakar•12/14/23, 6:04 PM

anyone else seeing their freshly deployed workers (new worker + links not shared anywhere publicly) getting bot scraped for vulnerabilities? sure helps keep the request counts nice and toasty mildpanicintensifies

kian•12/14/23, 6:04 PM

Every Worker gets a certificate issued and bots just watch certificate transparency logs.

Kkian Every Worker gets a certificate issued and bots just watch certificate transpare...

Salakar•12/14/23, 6:05 PM

true, but are these requests not something that CF can/should block automatically, since bots/bad-actors?

SSalakar true, but are these requests not something that CF can/should block automaticall...

Chaika•12/14/23, 6:10 PM

If you have the Managed WAF Rulesets (Pro or higher), it'll block some of them on custom domains.
Otherwise not really, you block anything based off paths or request IPs/etc and you'll end up with some false positives.

Cyb3r-Jak3•12/14/23, 6:21 PM

I just disable all custom domains and workers dev for like 20-30 minutes after creation to cause those spam requests to fail

CCarsonPoole how much GPU ram should we expect for WebGPU in durable objects?

zegevlier•12/14/23, 6:23 PM

I don't think that is known yet, but since the main use case is AI stuff, I would expect a good amount.

Jacob Wright•12/14/23, 10:33 PM

Is it possible to use unstable_dev and connect using a websocket for integration tests on websockets?

ajgeiss0702•12/14/23, 11:09 PM

where should I set platform: 'node'platform: 'node'?

ajgeiss0702•12/14/23, 11:17 PM

actually it seems like its throwing an error because it sees a dynamic import that wont actually be called since webcrypto exists on cf. Is there some way to disable that?
please ping me if you reply

wesleyyue•12/15/23, 2:27 AM

Is it possible to transform the originorigin header using just cloudflare transform rules, without writing a worker? When I try, it doesn't seem to work. Alternatively, is there a way to look at logs of transform rules running?

Rrvk Can you put cloudflare access in front of a workers.dev domain? I was allowed to...

Cyb3r-Jak3•12/15/23, 2:39 AM

No you don't have the ability to have access in front of .workers.dev.workers.dev domain. You have to use a custom domain for your worker.

�

😈 Donkey 💫•12/15/23, 9:15 AM

using create-cloudflare version 2.8.2
with option Typescript but output is always .js. I believe this is a bug

Eusebio Trigo•12/15/23, 11:16 AM

Hi! I'm using wrangler 3.20.0 and when I stop a wrangler with "x", the process gets orphaned and depending on PID 1, and I have to look for the process and kill it, as I cannot start it again as it throws a "port already in use". I'm on macOS sonoma 14.3 (Yes, I know, it is a beta).

Tetrax-10•12/15/23, 12:10 PM

Hi am new to cloudflare workers. I am trying to create an api with hono js. where hono gets an id and responds with a value for that id. The ids and values are stored in a CSV file it has like 1M ids. what is the best way to quickly get the value form it. Like should I use a database or things like kv. What will you guys recommend?

�😈 Donkey 💫using create-cloudflare version 2.8.2 with option Typescript but output is alway...

James•12/15/23, 1:25 PM

Yeah this is a bug: https://github.com/cloudflare/workers-sdk/issues/4599

I would recommend cloudsparkcloudspark: https://github.com/Cloudflare-Community/cloudspark

GitHub

GitHub - Cloudflare-Community/cloudspark: Community-created CLI for...

Community-created CLI for the Cloudflare Developer Platform. - GitHub - Cloudflare-Community/cloudspark: Community-created CLI for the Cloudflare Developer Platform.

Pato•12/15/23, 4:44 PM

Hey I use a service binding in one worker and have it defined like this in my wrangler toml but everytime I push new code it unbinds

services = [
  { binding = "SERVICE", service = "service-worker", environment = "production" }
]

services = [
  { binding = "SERVICE", service = "service-worker", environment = "production" }
]

but w

Screenshot_2023-12-15_at_10.44.00_a.m..png

Pato•12/15/23, 4:44 PM

Its kinda of anoying

kian•12/15/23, 4:45 PM

What's your full Wrangler.toml?

kian•12/15/23, 4:45 PM

If that's below any tables (surrounded by [][]) then it's probably being inherited into the wrong field.

Pato•12/15/23, 4:46 PM

Okay i see, so it should be before this:

[build]
command = "npm install && npm run build"

[build.upload]
format = "service-worker"

[build]
command = "npm install && npm run build"

[build.upload]
format = "service-worker"

Pato•12/15/23, 4:48 PM

it works now

, thanks!

kian•12/15/23, 4:52 PM

Yep. Alternatively

kian•12/15/23, 4:52 PM

[[services]]
binding = "SERVICE"
service = "service-worker"
environment = "production"

[[services]]
binding = "SERVICE"
service = "service-worker"
environment = "production"

kian•12/15/23, 4:52 PM

That's the exact same, but isn't affected by other tables.

Unsmart•12/15/23, 4:57 PM

This is the its not dns its always dns but with toml mhm

sanetram•12/15/23, 9:44 PM

Hey, I'm implementing user sessions and honestly I wasn't around coding when JWTs started taking over. In terms of Workers, should I use JWTs and incur additional CPU cost for crypto or would it be more economical to use KV, session cookies, and D1?

For Hyperdrive questions I recommend asking in #hyperdrive-beta where the team is active, individual

Similar Threads