I just found out that websites with workers + assets tend to be _slower_ than pages on sequent navig

I just found out that websites with workers + assets tend to be slower than pages on sequent navigations, since assets don't have cache with max-age

Will there be a way to set that assets return with specific cache-control value? It's weird that there isn't an option to add it directly, considering that most projects with frameworks build assets with max-age headers in mind.

Kenneth Solberg•4/30/25, 6:39 AM

After switching from Pages to Workers for my Astro site I can run deploy from local dev and everything works. However when I connect it to the GitHub repo the build fails with this message: "You have to fill in all your /.env environmental variables". Not 100% sure what's missing?

Kenneth Solberg•4/30/25, 6:58 AM

I have public vars in .toml and secrets configured in dashboard. Do i have to configure them in build vars in dashboard as well?

Kenneth Solberg•4/30/25, 8:19 AM

resolved by adding to build vars, case closed

Rabid•4/30/25, 1:14 PM

How can I add nodejs_compat to a new pages project? I need to change the underlying source github repo, and this doesn't seem to be an option in the UI anymore

RRabid How can I add nodejs_compat to a new pages project? I need to change the underl...

Silvan•4/30/25, 2:39 PM

From unrelated issues i have seen, you need to put it in your wrangler config

SSilvan From unrelated issues i have seen, you need to put it in your wrangler config

Rabid•4/30/25, 2:39 PM

I got it to work, I just opened the network tab, copied the request to add some unrelated flag, and then changed it to nodejs_compat and just ran the curl lol. Shouldn't be this hard to do.

RRabid I got it to work, I just opened the network tab, copied the request to add some ...

Silvan•4/30/25, 2:40 PM

If its a new project might also just consider using asset workers instead

SSilvan If its a new project might also just consider using asset workers instead 👍

Rabid•4/30/25, 2:40 PM

All my stuff are on pages, do asset workers support auto CICD from github?

Silvan•4/30/25, 2:42 PM

mostly
https://developers.cloudflare.com/workers/static-assets/migrate-from-pages/#compatibility-matrix

Cloudflare Docs

Migrate from Pages

A guide for migrating from Cloudflare Pages to Cloudflare Workers. Includes a compatibility matrix for comparing the features of Cloudflare Workers and Pages.

daz•4/30/25, 2:50 PM

Afternoon,
I have a workers paid plan, for a pro site.
I'd like to enable Workers Logpush for my worker.
If I try to enable it via wrangler.toml I get:

You do not have access to use Logpush. Please ensure it is enabled. If you are an Enterprise user, reach out to your account team. [code: 10023]

You do not have access to use Logpush. Please ensure it is enabled. If you are an Enterprise user, reach out to your account team. [code: 10023]

If i try to enable it with a CURL request I get
{"code": 1004, "message": "creating a new job is not allowed: exceeded max jobs allowed"}{"code": 1004, "message": "creating a new job is not allowed: exceeded max jobs allowed"}
When i try to list all log push jobs I get:
{"errors": [], "messages": [], "result": [], "success": true}{"errors": [], "messages": [], "result": [], "success": true}
I have no Logpush entry in my administrative UI
I am a super administrator.
Has anyone else experienced issues enabling workers logpush for a non-enterprise account.
So far searching cloudflare knowledge base and discord has only yielded suggestion that we could be on an older 'paid' worker plan somehow.
Many thanks in advance if you have any pointers
Apologies if I'm asking in the wrong channel

Cloudflare Docs

Workers Logpush

Send Workers Trace Event Logs to a supported third party, such as a storage or logging provider.

kaelys.js•4/30/25, 3:34 PM

I managed for POC and curiosity to:
- Use a LB with Geosteering to UK and a 100% weighted pool
- Tie the pool endpoint to an “inner” worker say “lhr.domain.com”
- have another worker say “api.domain.com” do a sub request to that worker, the result is that the colo of that worker ends up being “LHC” — most of the time. So the inner worker is executed out of that colo, where without the LB setup it would be executed at least for me generally in SEA or YVR depending ofc.

Should that even work? It’s more of a poc than anything.

The goal here is to utilize workers and the browser rendering api to spider out customers site to scan for multi region compliance (with their permission ofc). We currently use aws serverless for this but a CF solution is in my experience always simpler and more maintainable and also more cost effective.

But unless there’s a decent way outside enterprise we may stick with what we have.

We also heavily use analytics, 30k customers recording page view / consent events. Works great.

The central problem is that we sacrifice full control (such as regional execution of worker, and deciding where analytics data is housed, say EU for GDPR compliance for starters) for less engineering efforts overall.

It’s a tough sell to move up to enterprise (discussions already happened back in Dec) for some on these reasons.

BBenjamín Vicente 🇨🇱I just found out that websites with workers + assets tend to be _slower_ than pa...

Benjamín Vicente 🇨🇱OP•4/30/25, 7:46 PM

ok, adding a public/_headerspublic/_headers solves this, but it isn't clear that this file applies to all and only to static assets

Benjamín Vicente 🇨🇱OP•4/30/25, 8:05 PM

vite has a assetsDirassetsDir option, ideally cloudflare (via the vite plugin) should add a cache header to all files under that path.

Natan Parmigiano•4/30/25, 8:25 PM

Hey CF Team!

Just got done reading up on the "Simple, scalable, and global: Containers are coming to Cloudflare Workers"

Im truly amazed at the progress and rate of new feature releases!

One question that came up when reading was:

Will there be a maximum wall-clock limit on containers execution for statefull workloads?

We have some jobs that take a while, and for this reason had to be executed outside the CF platform (currently fly.io), the question is, would this new feature be suitable to run a container that, for example, runs some automation steps that take aproximatelly 40 minutes before exiting?

I know its probably early to ask about platform limits, but i'd love to know more

Wrekah•4/30/25, 11:03 PM

Why they removed jq from cloudflage pages build image? its availabe in v1, not available in v2. Any workarounds besides moving away from building it there?

Avi•5/1/25, 7:08 AM

should we check worker-configuration.d.tsworker-configuration.d.ts into git?

yupyupppp•5/1/25, 9:21 AM

Cloudflare workers GBS billing only counts for durable objects?

ShafSpecs•5/1/25, 11:58 AM

Hello

ShafSpecs•5/1/25, 11:59 AM

Question, how can one upgrade upload limit for workers? Thx

Walshy•5/1/25, 12:11 PM

https://developers.cloudflare.com/cache/concepts/default-cache-behavior/#upload-limits

ShafSpecs•5/1/25, 12:13 PM

Yeah, saw that. Didn't phrase my question well. I wanted to know if it was possble for worker.dev subdomains (non-custom), but just saw it was not possible. Thx

rufi•5/1/25, 12:55 PM

hello!
It seems there is a limit in the number of characters for tagging a worker version (We are using gradual deployments)
is that by design? we append the environment naming in the tag and this create artificial issues where we have to create shorter tags. Where is this documented? Can we suggest to revise this limit?

Walshy•5/1/25, 1:01 PM

it is by design. what is the ideal limit you're looking for?

Walshy•5/1/25, 1:02 PM

You could also use a message to put the env name in, we limit that to 100 chars

LordSilver•5/1/25, 6:37 PM

what's the diff between https://github.com/cloudflare/playwright-mcp and https://github.com/cloudflare/mcp-server-cloudflare/tree/main/apps/browser-rendering ?

mr.niko.la•5/1/25, 7:43 PM

on paid plan, CPUexecution time is 30s?

i have a audio processing worker that uses 3rd party API to scribe 4-15s and i pipe the response to another llm api call that can take 8s all in one worker.

The dead times waiting for api are not counted only the cpu execution time?

how do i see workers total cpu usage time and memory usage time for each call.
I want to make sure my worker is not going to run out of time if i add more llm api cools on the same worker
thx

Ellie Rider•5/1/25, 8:03 PM

Question, is it possible to direct a path to a different host?

i.e. www.example.com proxies mywebsite.sample.com
but www.example.com/foo proxies foo.sample.com

Ellie Rider•5/1/25, 8:12 PM

OK thanks, I thought workers would have to be the way but I was hearing some conflicting ideas

mr.niko.la•5/1/25, 8:13 PM

maybe page rules could do it as well. i would ask grok/chatgpt

mr.niko.la•5/1/25, 8:14 PM

or dns settings if ur domain is with cf or whichever provider has dns

liam | helius.dev•5/1/25, 8:27 PM

Is the only way to get reliable websocket connections via DO's? We have lots of cases where websockets stop transmitting data with no error and support suggested setting up DO's. This is a problem though because DO's are not very scalable

Lliam | helius.dev Is the only way to get reliable websocket connections via DO's? We have lots of ...

Hard@Work•5/1/25, 9:06 PM

DOs are the way to make WebSockets survive an isolate restart, yeah. To make sure I understand, which part of DOs are you speaking about when you mention they aren't scalable?

Senti3ntB3ing•5/1/25, 9:11 PM

how do i call upon the secrets i set in the dashboard? docs only show how to set/delete them, not how to call upon them

SSenti3ntB3ing how do i call upon the secrets i set in the dashboard? docs only show how to set...

Hard@Work•5/1/25, 9:14 PM

What do you mean "call upon them"? Secrets are just present, like

export default class extends WorkerEntrypoint<Env> {
  fetch() {
    const mySecret = this.env.SOME_SECRET;
  }
}

export default class extends WorkerEntrypoint<Env> {
  fetch() {
    const mySecret = this.env.SOME_SECRET;
  }
}

Senti3ntB3ing•5/1/25, 9:22 PM

i don't have that section (or i couldn't find it) when i first started the project with wrangler

Senti3ntB3ing•5/1/25, 9:22 PM

i just had the export default{.. fetch(req,env,ctx)export default{.. fetch(req,env,ctx)

Senti3ntB3ing•5/1/25, 9:26 PM

import { Database } from "./database";

const TWITCH_APP_ID = await Database.get("twitch_app_id");
const TWITCH_APP_SECRET = await Database.get("twitch_app_secret");
const TWITCH_REFRESH = await Database.get("twitch_refresh_token");

const FIREBASE_SECRET = env.FIREBASE_SECRET

import { Database } from "./database";

const TWITCH_APP_ID = await Database.get("twitch_app_id");
const TWITCH_APP_SECRET = await Database.get("twitch_app_secret");
const TWITCH_REFRESH = await Database.get("twitch_refresh_token");

const FIREBASE_SECRET = env.FIREBASE_SECRET

i dont suppose this is how i would be calling upon the secret in question (FIREBASE_SECRET)

Senti3ntB3ing•5/1/25, 9:26 PM

or would i need to extend that default class?

SSenti3ntB3ing ```js import { Database } from "./database"; const TWITCH_APP_ID = await Datab...

Hard@Work•5/1/25, 9:26 PM

Yeah, env.FIREBASE_SECRETenv.FIREBASE_SECRET should work

Senti3ntB3ing•5/1/25, 9:27 PM

thank you, it didn't generate a .env file and i know other runtimes have their own ways of calling secrets

HHard@Work DOs are the way to make WebSockets survive an isolate restart, yeah. To make sur...

liam | helius.dev•5/1/25, 9:58 PM

Why is it that workers hold zombie connections? It doesn't really make sense

DO requests/s is too low and creation of new DO is too slow

Lliam | helius.dev Why is it that workers hold zombie connections? It doesn't really make sense DO...

Hard@Work•5/1/25, 10:03 PM

Not sure I understand what you mean on the first part?

HHard@Work Not sure I understand what you mean on the first part?

liam | helius.dev•5/1/25, 10:05 PM

we frequently see that a websocket stays open, but data stops flowing. Something internal to CF is causing this and our support tickets have kept telling us to use DO's instead (which is also confusing because it requires so many extra layers/room for error when all we want is to proxy)

We've tested this extensively, removing CF solves the problem

Senti3ntB3ing•5/1/25, 10:23 PM

okay so i did some reading on the fetch handler - if i have a class object i have created to handle a whole bunch of API calls, and it uses fetch within itself, how can i make all of those fetch requests valid?

Senti3ntB3ing•5/1/25, 10:24 PM

would i wrap my whole class in

export default{
  async fetch(request){
    export class SomeClass{...

export default{
  async fetch(request){
    export class SomeClass{...

Senti3ntB3ing•5/1/25, 10:26 PM

so that every instance of the class calling

fetch

fetch

will be within the cloudflare fetch handler?

Senti3ntB3ing•5/1/25, 10:26 PM

or would i take each instance of me calling fetch and wrap the handler around those?

WWalshy it is by design. what is the ideal limit you're looking for?

rufi•5/2/25, 3:10 AM

Currently when I deploy and tag it seems limited to 25 characters .
100 characters would be great.

Silvan•5/2/25, 9:27 AM

So trying to pass Headers from my sveltekit app to a rpc function i sadly get
DevalueError: Cannot stringify arbitrary non-POJOsDevalueError: Cannot stringify arbitrary non-POJOs is there a recommended way to serialize/deserialize them?

kerohije•5/2/25, 10:11 AM

Can this route setting be configured in wrangler's config file?

I just found out that websites with workers + assets tend to be _slower_ than pages on sequent navig

Similar Threads

Similar Threads

Similar Threads