Cloudflare Developers•2mo ago

When you refresh, what does it show? Might just be some UI bug where the input get reset on button c

When you refresh, what does it show? Might just be some UI bug where the input get reset on button click

Ttexan When you refresh, what does it show? Might just be some UI bug where the input g...

danawoodman•10/29/25, 3:27 AM

same behavior even after browser restart

danawoodman•10/29/25, 3:28 AM

have you given it a shot? does it work for you? i'm on Brave fwiw (with shields off)

danawoodman•10/29/25, 3:28 AM

it will even overwrite my text while I'm typing

1984 Ford Laser•10/29/25, 5:00 AM

Is the 10 maximum transforms for the Images binding something that can be increased?

welz•10/29/25, 7:29 AM

What is powering rate limiting under the hood?

Doan | Full-Stack Developer•10/29/25, 7:29 AM

Hello everyone.

Doan | Full-Stack Developer•10/29/25, 7:29 AM

How are you doing?

SODaLC•10/29/25, 1:02 PM

I'm not sure if this is a bug or not, but I have a request logger for my workers for analytics purposes and sends the fetch back to the worker, which is connected to multiple workers and works on CF. I spun up a new project that would use it and used remote in wrangler config to connect to it but every request fails because "Cannot serialize value: [object Response]" but if i shift to the local version of it, it just works as intended. Is this intended between local and remote or a bug?

ALECKS•10/29/25, 1:33 PM

When using RPC method invocation between workers/durable objects, do I need to build some security layer around this myself, or does the CF architecture give me some guarantees? For ex, in a normal multi-service architecture, it's quite common to have service-to-service calls be authenitcated, where each service has some kind of token/cert/etc...
It's not quite clear if I need to do the same when doing worker-to-worker / worker-to-durableobject RPC calls. (not fetch).

1984 Ford Laser•10/29/25, 8:29 PM

Has anyone noticed wrangler deployments taking longer the last few weeks? Used to be able to deploy a new version within a couple seconds, but now its 20+ seconds, even for a 1.4KB worker

Kris•10/29/25, 10:19 PM

I've got a geeky question about RPC reentrancy. Ran into a realworld issue and worked around it, but curious.

Worker A uses the Dynamic Workers Loader to load Worker B, and awaits B.c().
B.c() initializes resources, including using a callback provided by A to create Object D within Worker A that stores Callback E (an RpcStub to Worker B).
Then B.c() awaits D.f(), which ends up awaiting E(), which fails with "RPC stub used after being disposed."

This surprises me, since this is all within the same request. The workaround is to return some information from E() that tells Worker B to call the function directly, which works, but is less clean, and wouldn't work if E() needed the return value.

I realize this sounds fairly convoluted, but the pattern is required to support self-registering dynamic code.

My best theory is there is some limit on reentrancy, or Worker B awaiting an RPC on Worker A prevents it from receiving a previously saved RPC from Worker A. Just curious to understand the limits to design around them.

11984 Ford Laser Is the 10 maximum transforms for the Images binding something that can be increa...

1984 Ford Laser•10/30/25, 12:08 AM

Have requested an increase to this limit, will be interesting to see if I hear back at all

PhoenixSpace•10/30/25, 1:05 AM

Does Argo routing do anything when using service bindings? Like if I have a worker with bindings to r2, kv, D1, etc. will argo smart routing optimize the connection from the worker to that resource, or is it just from user to cloudflare?

EducatedSuddenBucket•10/30/25, 2:52 PM

What is difference between not yet supported and partially supported (non functional) ?
In this page
https://developers.cloudflare.com/workers/runtime-apis/nodejs/
I want to use datagram

Cloudflare Docs

Node.js compatibility

Node.js APIs available in Cloudflare Workers

EEducatedSuddenBucket What is difference between not yet supported and partially supported (non functi...

Peps•10/30/25, 2:55 PM

Cloudflare Workers does not support UDP at the moment. I'm guessing 'partially supported' means the functions are available in the scope, but that they are essentially a no-op to not break packages who depend on it's existence

Ttexan When you refresh, what does it show? Might just be some UI bug where the input g...

danawoodman•10/30/25, 3:45 PM

were you able to try and repro this?

PPhoenixSpace Does Argo routing do anything when using service bindings? Like if I have a work...

PhoenixSpace•10/30/25, 4:30 PM

In case anyone searches later, from manual testing looks like the answer is “no”

PPhoenixSpace In case anyone searches later, from manual testing looks like the answer is “no”

Chaika•10/30/25, 4:47 PM

Those should already go across their internal networking/be as quick as possible durable-objects (D1, R2, use Durable Objects)

Chaika•10/30/25, 4:47 PM

Argo does optimize the connection to origins though, that's sort of the main point whereas the ingress optimization is kind of just another benefit. Fetch within worker should also benefit from that

Chaika•10/30/25, 4:48 PM

Fetching content is one of the most common Worker tasks. fetch() has always leveraged powerful Cloudflare features like Argo and Load Balancing.

https://blog.cloudflare.com/introducing-the-workers-cache-api-giving-you-control-over-how-your-content-is-cached/
old confirmation

CChaika Those should already go across their internal networking/be as quick as possible...

PhoenixSpace•10/30/25, 4:49 PM

Thanks for the response! Good to know, I’m glad you get that backbone for “free” :)

Sorry unrelated but

d1, r2 are durable objects

D1 I knew, but r2? That’s really cool

PPhoenixSpace Thanks for the response! Good to know, I’m glad you get that backbone for “free”...

Hard@Work•10/30/25, 4:50 PM

Note, R2 utilizes DOs, but R2 isn't a DO, unlike with D1

Chaika•10/30/25, 4:50 PM

oops yea

use

use

not areare, the metadata is on there but not the objects

Chaika•10/30/25, 4:50 PM

there's more info https://developers.cloudflare.com/r2/how-r2-works/

PhoenixSpace•10/30/25, 4:51 PM

That is incredibly cool

PhoenixSpace•10/30/25, 4:51 PM

I knew durable objects were powerful but wow

trololo•10/31/25, 7:15 AM

Hi 2 all
What is the best solution for storing audit logs for customers(organizations) from Workers+DO ?

trololo•10/31/25, 7:20 AM

my use case is writing logs for customers for critical (and not only) actions with their data in my web apps. Then customers could see this logs (at least last 3m) or download for 7 years

Ttrololo my use case is writing logs for customers for critical (and not only) actions wi...

1984 Ford Laser•10/31/25, 7:52 AM

What sort of scale are you thinking? You could use Workers Analytics Engine to store the data then export it as files per 30 days or something in R2 for longer term archive. I think WAE is the cheapest way to write logs on platform

11984 Ford Laser What sort of scale are you thinking? You could use Workers Analytics Engine to s...

trololo•10/31/25, 8:10 AM

Oh thank you. I've thought about this solution, but maybe there is another way

Scaling - like 2-3 web apps, but want to make it scalable from the start. WAE is better and free(included in workers paid) from the start. On R2 will store aggregated data each hour (with scale will be more frequently)

1984 Ford Laser•10/31/25, 8:12 AM

How many events per hour are you thinking of storing? On average

11984 Ford Laser How many events per hour are you thinking of storing? On average

trololo•10/31/25, 8:18 AM

at start 1k. But I think will be more in time

BigFungus•10/31/25, 9:24 AM

We are trying to setup a 24/7 realtime dashboard with workers that receive ~5 messages per minute. Does anyone have cost estimates using websockets? We are considering using SSE for reduced costs. Does that make a difference?

Ttrololo my use case is writing logs for customers for critical (and not only) actions wi...

dave•10/31/25, 4:54 PM

IMO use any S3 API provider you like, and https://github.com/mhart/aws4fetch in your Worker. Backblaze B2 would be my suggestion. It's $0.006/GB, or $0.012/GB if you want to do high availability by mirroring two buckets in different regions.

https://gist.github.com/Manouchehri/733e6235457e60de24fdbb15046fba7f

Add a lifecycle policy to the bucket that deletes files after 7 years.

Put each customer's files into like s3_bucket_name/customer_id_here/2025/10/31/23/epoch_time_here_uuid-of-event-here.jsons3_bucket_name/customer_id_here/2025/10/31/23/epoch_time_here_uuid-of-event-here.json. Then, when a customer wants say, the last two days of logs, you can do a ListObjectsV2 request to s3_bucket_name/customer_id_here/2025/10/30/s3_bucket_name/customer_id_here/2025/10/30/ and s3_bucket_name/customer_id_here/2025/10/31/s3_bucket_name/customer_id_here/2025/10/31/ to get all the JSON files of their events, which you can then use aws4fetch again to give the user presigned S3 URLs to download their events.

dave•10/31/25, 4:57 PM

Backblaze B2 also doesn't charge for uploading objects, so it's fine if you're uploading millions of small files to the same bucket. https://www.backblaze.com/cloud-storage/transaction-pricing

dave•10/31/25, 4:58 PM

And lastly, definitely use gzip for compressing JSON events! The user's browser (or the Worker) will automatically decompress it as long as you set the content-encoding: gzipcontent-encoding: gzip when uploading.

geg•10/31/25, 7:51 PM

Am I correct to assume that MessageChannel API is safe way to send events between different requests the worker is handling? For example letting 2 websocket clients that hit the same worker isolate talk to eachother is now possible? I kind of did something like this before but by polling a global value

(circumventing the "Cannot perform I/O on behalf of a different request." error)

BBigFungus We are trying to setup a 24/7 realtime dashboard with workers that receive ~5 me...

1984 Ford Laser•10/31/25, 7:56 PM

SSE will be more expensive than Durable Objects with Hibernating WebSockets

Ggeg Am I correct to assume that MessageChannel API is safe way to send events betwee...

Isaac McFadyen•10/31/25, 7:58 PM

I believe it's request-scoped.

IIsaac McFadyen I believe it's request-scoped.

geg•10/31/25, 7:58 PM

I just tried it out and seems to work out, just asking here to be sure it's okay for future

Ggeg I just tried it out and seems to work out, just asking here to be sure it's okay...

Isaac McFadyen•10/31/25, 7:59 PM

Oh I see, if it works then it's probably fine to rely on

Isaac McFadyen•10/31/25, 7:59 PM

Keep in mind that there is absolutely 0 guarantees you'll hit the same datacenter or even that there won't be multiple copies of a Worker running in a datacenter.

Tito•11/1/25, 1:26 AM

is there a place for suggestions? if not, can we please allow more than 50 api keys or have it not on by default when creating a worker :3

TTito is there a place for suggestions? if not, can we please allow more than 50 api k...

texanOP•11/1/25, 2:19 AM

#workers-and-pages-feature-suggestions

Ttexan #workers-and-pages-feature-suggestions

Tito•11/1/25, 2:27 AM

ty will :CoolCat: also surprised you ain't on here (https://who-to-bother-at.com/cloudflare) same w/ Walshy

unless you not on X

texanOP•11/1/25, 2:28 AM

I’m definitely hit and miss on activeness in discord. But yeah also no twitter for me, and not even in the same ballpark as many of the CF folks here!

11984 Ford Laser SSE will be more expensive than Durable Objects with Hibernating WebSockets

BigFungus•11/1/25, 5:00 PM

Oh right I forgot about hibernating websockets thanks you saved us a lot of Dev time