are worker "frameworks" replacing pages? - Cloudflare Developers

© 2026 Hedgehog Software, LLC

More

Communities Docs About Terms Privacy

Kevin

KKevin any thoughts on sub request just closign networ connection lost after 100 second...

Chaika

C

Chaika•3/25/25, 1:31 AM

The Cloudflare wide Proxy Read Timeout is 100 seconds https://developers.cloudflare.com/fundamentals/reference/connection-limits/, if your backend is that slow you'll need to speed it up, have it use a job style where it returns that it's running and to poll for status, etc. For SSE specifically, I haven't tried it but there's a few posts mentioning it kind of works and that 524's only trigger if nothing is sent within 100s

Tobias

TTobias Is there any way I to get custom domains without a zone on cloudflare working on...

Chaika

C

Chaika•3/25/25, 1:33 AM

Not if you have no zones at all. If you have at least one, even just a throwaway/random ones, can use CF For SaaS: https://developers.cloudflare.com/cloudflare-for-platforms/cloudflare-for-saas/start/advanced-settings/worker-as-origin/ and have your real custom domain CNAME to it, which then would run your worker. More meant for full SaaS cases though. It's the same thing Pages uses under the hood

wot

Wwot are worker "frameworks" replacing pages?

Chaika

C

Chaika•3/25/25, 1:35 AM

Frameworks is just their list of guides in the docs, "Workers Static Assets" is the underlying replacement, allowing Workers to support static assets just like Pages. CF has said before they plan on offering more migration steps in the future to Workers Static Assets from Pages, but eventually they will converge

TDat

T

TDat•3/25/25, 3:05 AM

Unable to preview your Worker

matt

M

matt•3/25/25, 4:00 AM

if a worker connects to a cloudflare tunnel endpoint, and there's a WAF rule to block any non-worker traffic, eg.

BLOCK cf.worker.upstream_zone != "my.domain"

BLOCK cf.worker.upstream_zone != "my.domain"

how much latency should both the WAF rule & cloudflare tunnel be adding?

Chaika

C

Chaika•3/25/25, 4:29 AM

waf should be too small to measure, tunnel is more complex. I'd make sure you're connecting to a nearby PoP ( example.com/cdn-cgi/traceexample.com/cdn-cgi/trace on your tunnel hostname) and in your tunnel logs, that your tunnel is connecting to a nearby location as well, not too related to Workers though

masd4

M

masd4•3/25/25, 5:14 AM

if the worker memory limit is 128mb, wouldn't I run into that if I use the kv store .list() operation and retrieve 1280 keyvalue items with a size of 100kb? I mean those string values get loaded into memory if I .get() them all one by one in a for loop? I guess the GC would be able to start running if that's how it works in a loop with a memory limit set.

Chaika

CChaika Not if you have no zones at all. If you have at least one, even just a throwaway...

Tobias

T

Tobias•3/25/25, 10:23 AM

thanks so much! this worked well

lisekilis

L

lisekilis•3/25/25, 12:04 PM

lil help please?
https://canary.discord.com/channels/595317990191398933/1354022352987619349

Justin Schroeder

J

Justin Schroeder•3/25/25, 3:53 PM

Cloudflare Team: Does this https://discord.com/channels/595317990191398933/1040420029080018945/1354117859671805964 announcement mean that ctx.waitUntil()ctx.waitUntil() can have a longer execution time too. Its been 30s of wall-clock duration, and I could really use an increase time limit there.

zerozerozeroone

Z

zerozerozeroone•3/25/25, 11:17 PM

Chaps when connecting github, I can only see repositories created by the account that was connected. It doesn't show repos shared with me. Is there anyway repos shared with me can be used?

Justin Schroeder

JJustin Schroeder Cloudflare Team: Does this https://discord.com/channels/595317990191398933/10404...

Walshy

W

Walshy•3/25/25, 11:20 PM

no, waitUntil hsa not changed

Walshy

WWalshy no, waitUntil hsa not changed

Justin Schroeder

J

Justin Schroeder•3/25/25, 11:21 PM

Cool cool cool. Any chance they might happen in the near future?

matt

M

matt•3/26/25, 12:31 AM

in this worker code im measuring time to hit origin. it starts off ~350 and then after a few refreshes, becomes a mixture of ~350 and ~100

is this TCP or HTTP keepalive at play? are there any viewable stats, or any ways that I can get the ~100 to occur more frequently?

const start = performance.now()
await fetch('https://unweb.io/healthcheck').then(x => x.text())
return new Response(String(performance.now() - start));

const start = performance.now()
await fetch('https://unweb.io/healthcheck').then(x => x.text())
return new Response(String(performance.now() - start));

divby0

D

divby0•3/26/25, 10:23 AM

I read something somewhere (a while ago) about workers being favored in the future in place of pages. Did I read that wrong? Can't find anything on pages getting deprecated or something.

divby0

Ddivby0 I read something somewhere (a while ago) about workers being favored in the futu...

Vero

V

Vero•3/26/25, 12:05 PM

Hey. Pages is still supported but longer term we want to get to a point where every new application can be built on Workers

Vero

V

Vero•3/26/25, 12:06 PM

If there are gaps preventing you from using Workers today (and instead choosing Pages), we’d love to hear about that

Tobias

T

Tobias•3/26/25, 2:49 PM

Is it possible to setup a subdomain for cloudflare workers without either

Setup the apex domain within cloudflare
Pay 250 dollar a month to setup a subdomain zone

Currently this is possible for cloudflare pages (i.e. just point the subdomain to cloudflare pages).

If not, is it planned in order to have feature parity between cloudflare workers & pages?

Tobias

TTobias Is it possible to setup a subdomain for cloudflare workers without either - Setu...

Hello, I’m Allie!

H

Hello, I’m Allie!•3/26/25, 2:51 PM

Use SSL for SaaS yourself, like Pages does?

Hello, I’m Allie!

HHello, I’m Allie!Use SSL for SaaS yourself, like Pages does?

Tobias

T

Tobias•3/26/25, 2:52 PM

But then I would need to create a zone within cloudflare with a domain, or I might have missed something?

Tobias

TTobias But then I would need to create a zone within cloudflare with a domain, or I mig...

Hello, I’m Allie!

H

Hello, I’m Allie!•3/26/25, 2:52 PM

Yeah, you would need one zone at minimum on Cloudflare

jsonperl

J

jsonperl•3/26/25, 2:54 PM

Hi there! I'm trying to build a worker that does realtime audio "splicing" from component mp3s. Essentially based on an audio file layout, grab the files from r2, remove header info and splice the mp3 audio data together and stream.

I've succeeded in building this. But now I'm running into a problem that seems maybe unsolvable: clients expect to be able to issue range requests in a header asking for specific bytes in the audio (I'd like to listen starting at 10:00 in...).

What I'd ideally like is that an initial request will be cached entirely, and then the cdn would take over to respond to the range request with a proper 206 response and the bits. There doesn't seem to be a way to do this.

If I respond with a 206 from the worker, it can't be cached
If I respond with a 200 with the whole file, the client doesn't get what it asked for

Any ideas?

jsonperl

Jjsonperl Hi there! I'm trying to build a worker that does realtime audio "splicing" from ...

Hello, I’m Allie!

H

Hello, I’m Allie!•3/26/25, 2:56 PM

You would need to implement this cache slicing logic in your Worker. Worker responses can’t be cached, so requests will always be billed. Note however that they can interact with the Cache API as part of their handling of a request

Hello, I’m Allie!

HHello, I’m Allie!You would need to implement this cache slicing logic in your Worker. Worker resp...

jsonperl

J

jsonperl•3/26/25, 3:00 PM

Is there no way to rely on range request handling from standard cloudflare cdn? Like front a worker with the standard cdn?

Hello, I’m Allie!

HHello, I’m Allie!You would need to implement this cache slicing logic in your Worker. Worker resp...

jsonperl

J

jsonperl•3/26/25, 3:01 PM

The other huge gotcha here is that anything besides directly streaming from r2 out to the client will quickly surpass the memory limits of a worker. Audio is large, but the streaming solution built works well as it just streams it through without holding much in memory.

Any caching of the whole thing (or writing out to r2 for a "tempfile") is going to need it loaded in memory, which cannot work.

jsonperl

Jjsonperl Is there no way to rely on range request handling from standard cloudflare cdn? ...

Hello, I’m Allie!

H

Hello, I’m Allie!•3/26/25, 4:43 PM

Workers are a layer of the CDN on top of the cache. They can interact with it, but responses from a Worker will not be cached themselves

jsonperl

Jjsonperl The other huge gotcha here is that anything besides directly streaming from r2 o...

Hello, I’m Allie!

H

Hello, I’m Allie!•3/26/25, 4:44 PM

You don’t need to buffer the whole thing in memory, you can stream directly from R2 into the Cache

Vero

VVero Hey. Pages is still supported but longer term we want to get to a point where ev...

divby0

D

divby0•3/26/25, 8:02 PM

thanks a lot

No shortcomings, just curiosity because this recommendations are not really easily findable

Mia

M

Mia•3/26/25, 10:41 PM

RIP i just wasted an entire day going down the websockets rabbit hole only to realise discord has a cloudflare ban so i can't use workers with DO for their gateway api :(:(

Mia

M

Mia•3/26/25, 10:42 PM

they just 401 when a new connection attempt originates from a worker ip

Mia

M

Mia•3/26/25, 10:42 PM

and from what ive read the only way around this is with Aegis but that's enterprise only

Mia

M

Mia•3/26/25, 10:43 PM

proxy policies would be cool on a per-worker basis

Mia

M

Mia•3/26/25, 10:43 PM

im fine hosting a proxy on my own server just to get around the discord cloudflare ban :\

Mia

M

Mia•3/26/25, 10:44 PM

although that will add even more layers of complexity and latency for the traffic so not optimal

Mia

M

Mia•3/26/25, 10:44 PM

i just dont wanna host the entire bot on my own machine xd

Mia

M

Mia•3/26/25, 10:52 PM

I guess it’s not a huge waste of time cause I did learn a ton about workers, hono, durable objects and websockets… it just sucks i can’t do the thing i set out to do after all that time

roma

R

roma•3/27/25, 1:53 AM

https://developers.cloudflare.com/workers/platform/limits/#simultaneous-open-connections

Simultaneous open connections

You can open up to six connections simultaneously, for each invocation of your Worker. The connections opened by the following API calls all count toward this limit:

- the fetch() method of the Fetch API.

Simultaneous open connections

You can open up to six connections simultaneously, for each invocation of your Worker. The connections opened by the following API calls all count toward this limit:

- the fetch() method of the Fetch API.

I do not quite understand this part, it said that fetch() calls all count toward the concurrent request limit.

So, if I have a code below

const results = await Promise.all([
  fetch(url1),
  fetch(url2),
  fetch(url3),
  fetch(url4),
  fetch(url5),
  fetch(url6),
  fetch(url7), // Exceeds the limit of 6
]);

const results = await Promise.all([
  fetch(url1),
  fetch(url2),
  fetch(url3),
  fetch(url4),
  fetch(url5),
  fetch(url6),
  fetch(url7), // Exceeds the limit of 6
]);

Is the seventh call is going to be delayed (?) or something? I'm not a native english speaker so 99% I have misunderstood the docs here. Just want to confirm here

Peps

P

Peps•3/27/25, 1:53 AM

correct, the seventh call will be slightly delayed until one of those other 6 fetches are done

roma

R

roma•3/27/25, 1:54 AM

Ahh wow interesting, Thank you! I think I'm going to make batch (of six) call then just to make sure everything okay. Thanks!

S

s•3/27/25, 8:10 AM

Just going to put this out there: Adding Gemma 27B w/ LORA support would probably put Worker AI at the front of a lot of people's minds

Gemma is shaping up to be one of the best finetune targets in a while, and people are noticing :Eyes:

Velipaw

V

Velipaw•3/27/25, 9:09 AM

Hi, I am getting an issue when deploying my next project. Who can help me?

2025-03-27_10-59-42.snagx159.17KB

C

cj•3/27/25, 10:40 AM

I would like to run @vercel/og@vercel/og inside a worker. But @cloudflare/pages-plugin-vercel-og@cloudflare/pages-plugin-vercel-og uses an older version.

Anyone here who can merge pull requests and update the plugins?

https://github.com/cloudflare/pages-plugins/pull/57

GitHub

Update @vercel/og to 0.6.2 by mastermakrela · Pull Request #57 · ...

I was playing around with @vercel/og to understand why I can't use it directly in a worker.
During this, I've noticed that after updating @vercel/og to current version, the patch ca...

ruan

R

ruan•3/27/25, 1:54 PM

export default {
    async fetch(req, env): Promise<Response>
    {
        // https://developers.cloudflare.com/workers/runtime-apis/bindings/rate-limit/

        const ipAddress = req.headers.get("cf-connecting-ip") || "";
        const { success } = await env.MY_RATE_LIMITER.limit({ key: ipAddress })
        if (!success)
            return new Response(`429 Failure – rate limit exceeded for ${ipAddress}`, { status: 429 });

export default {
    async fetch(req, env): Promise<Response>
    {
        // https://developers.cloudflare.com/workers/runtime-apis/bindings/rate-limit/

        const ipAddress = req.headers.get("cf-connecting-ip") || "";
        const { success } = await env.MY_RATE_LIMITER.limit({ key: ipAddress })
        if (!success)
            return new Response(`429 Failure – rate limit exceeded for ${ipAddress}`, { status: 429 });

if i return "false" like this EXAMPLE, it would still count this request on my worker limits?

ruan

Rruan ```js export default { async fetch(req, env): Promise<Response> { ...

Chaika

C

Chaika•3/27/25, 1:56 PM

yes, worker still ran/executed

Chaika

CChaika yes, worker still ran/executed

ruan

R

ruan•3/27/25, 1:57 PM

do you know if on the workers paid plan is possible to you set any kind of expending limit in any way?
like if it get > than a $ cap stop working

Chaika

C

Chaika•3/27/25, 1:57 PM

most you can do is set usage alerts for specific amount of requests/cpu time

Simultaneous open connections

You can open up to six connections simultaneously, for each invocation of your Worker. The connections opened by the following API calls all count toward this limit:

- the fetch() method of the Fetch API.

Simultaneous open connections

You can open up to six connections simultaneously, for each invocation of your Worker. The connections opened by the following API calls all count toward this limit:

- the fetch() method of the Fetch API.

export default {
    async fetch(req, env): Promise<Response>
    {
        // https://developers.cloudflare.com/workers/runtime-apis/bindings/rate-limit/

        const ipAddress = req.headers.get("cf-connecting-ip") || "";
        const { success } = await env.MY_RATE_LIMITER.limit({ key: ipAddress })
        if (!success)
            return new Response(`429 Failure – rate limit exceeded for ${ipAddress}`, { status: 429 });

export default {
    async fetch(req, env): Promise<Response>
    {
        // https://developers.cloudflare.com/workers/runtime-apis/bindings/rate-limit/

        const ipAddress = req.headers.get("cf-connecting-ip") || "";
        const { success } = await env.MY_RATE_LIMITER.limit({ key: ipAddress })
        if (!success)
            return new Response(`429 Failure – rate limit exceeded for ${ipAddress}`, { status: 429 });