where in the chain would it be yeeted? does discord itself run on cf infra?
where in the chain would it be yeeted? does discord itself run on cf infra?
M(just curious not questioning)
IYes it does.
IOr well, it uses CF.
II believe it's a combination of GCP and CF, each for certain things.
Mhas discord put out a statement as to why they block workers from their gateway api? or is it just the scarce (hard to find) issue comments ive found so far with "yea we block it for security reasons"
II don't believe there's been an official statement aside from those Issue comments. There have been many requests to remove the block (or make it more fine-grained) but they haven't been very receptive thus far.
Mill see if someone replies to my necroposting on an old closed issue i guess
Mthanks for your replies isaac & leo :3
Rwhat happens if i run out of daily worker requests? (free plan)
it will block the request or just will forward the request to origin
it will block the request or just will forward the request to origin
IDepends on the failure mode you have set. https://developers.cloudflare.com/workers/platform/limits/#daily-request
Ris it possible to see
console.logCloudflare Docs
Supported methods of the console API in Cloudflare Workers
Mu can try
wrangler tailthere is logs in the dashboard as well if you cant use wrangler cli
Rim using the wrangler cli
was wondering about the dashboard
i can see the logs in the dashboard but not the console.log output, tail worker is paid, im still testing my application
was wondering about the dashboard
i can see the logs in the dashboard but not the console.log output, tail worker is paid, im still testing my application
Rtheres no channel to talk about cloudflare tunnel?
CI’d recommend #general-help
IThere is #cloudflare-one
BI just face one issue in my worker, when there is null value for any key in object and if i pass that object to any funciton or if I first stringify it and then parse it that key with null value got completely remove! why is that? it is not behaviour of JS, only in worker I face this issue!
Phttps://developers.cloudflare.com/pages/how-to/use-direct-upload-with-continuous-integration/ still points to the depreciated github actions lib
Cloudflare Docs
Cloudflare Pages supports directly uploading prebuilt assets, allowing you to use custom build steps for your applications and deploy to Pages with Wrangler. This guide will teach you how to deploy your application to Pages, using continuous integration.
BI am not able to use @google-cloud/speech package when i import it and just initialize class like below
Error
Error
UHI guys, sorry can anyone help me. I’m facing a critical issue with my Cloudflare Tunnel setup and need urgent assistance. Here’s the problem:
Error Details
SSL Certificate Mismatch:
When accessing my site, browsers throw:
"SSL Error: Certificate does not match the domain name. The certificate is issued for *.cfargotunnel.com, not [my site domain]."
Bad URL/404 Errors:
Without VPN: Requests return 400 Bad Request or 404 Not Found even though:
The tunnel is active and healthy (cloudflared tunnel list shows ACTIVE).
DNS records (CNAME for my domain → <TUNNEL_ID>.cfargotunnel.com) are correctly configured.
Works Only with VPN:
The site loads only when connected to a VPN. Disabling VPN immediately breaks access.
What I’ve Tried
Forced Certificate Renewal:
powershell
Copy
.\cloudflared.exe tunnel --force-renewal run
Tunnel Configuration:
yaml
Copy
ingress:
Tested with 1.1.1.1 DNS (no change).
Why does Cloudflare’s auto-generated certificate default to *.cfargotunnel.com instead of my custom domain?
Is this a DNS propagation issue, or are Cloudflare’s edge nodes failing to apply the correct certificate?
Why does a VPN bypass the problem? (Suggests ISP/geo-blocking, but the domain is not restricted.)
Error Details
SSL Certificate Mismatch:
When accessing my site, browsers throw:
"SSL Error: Certificate does not match the domain name. The certificate is issued for *.cfargotunnel.com, not [my site domain]."
Bad URL/404 Errors:
Without VPN: Requests return 400 Bad Request or 404 Not Found even though:
The tunnel is active and healthy (cloudflared tunnel list shows ACTIVE).
DNS records (CNAME for my domain → <TUNNEL_ID>.cfargotunnel.com) are correctly configured.
Works Only with VPN:
The site loads only when connected to a VPN. Disabling VPN immediately breaks access.
What I’ve Tried
Forced Certificate Renewal:
powershell
Copy
.\cloudflared.exe tunnel --force-renewal run
Tunnel Configuration:
yaml
Copy
ingress:
- hostname: my domain
service: http://localhost:80
originRequest:
httpHostHeader: "my domain"
noTLSVerify: true
DNS Checks:
Tested with 1.1.1.1 DNS (no change).
Why does Cloudflare’s auto-generated certificate default to *.cfargotunnel.com instead of my custom domain?
Is this a DNS propagation issue, or are Cloudflare’s edge nodes failing to apply the correct certificate?
Why does a VPN bypass the problem? (Suggests ISP/geo-blocking, but the domain is not restricted.)
TRelatively new to workers though also a relatively senior dev (rails). I have a couple questions that itch at the back of my mind and I just wanna make sure I'm understanding things right.
- Static Assets can nominally support queue consumers but I'm a little unclear on what that implies, does it mean that the main worker can define a single push consumer? I'd assume that kind of has to be it, right, it doesn't seem possible to have multiple push consumers since the consume method doesn't declare anything about from which queue it consumes. It, like me on a good costco sample day, just consumes. Is the idea that, if you want to build a consolidated full stack app with multiple queues, you'd use this strategy for one consumer processing multiple queues?: https://developers.cloudflare.com/queues/reference/how-queues-works/#consumers
- Conceptually, when using a framework with Static Assets, are SSR endpoints and SSR components de facto micro deployments of individual workers? I guess this is as much a question about how the adapters work. For example, if I have an Astro endpoint file that handlers multiple HTTP verbs, are those functionally translated into separate workers, or maybe more likely a single worker that happens to have Astro-specific syntax to handle the different request types? Is this analogous to how Pages Functions worked? How literal is this, like, were PFs ultimately run on individual isolates? My apologies if none of this makes sense, I'm asking out of curiosity from a sort of half formed shower thought involving leveraging the ephemeral nature of global scope in workers to emulate some architectural stuff that is laid out in the blog posts about how Waiting Room works.
Cloudflare Docs
Cloudflare Queues is a flexible messaging queue that allows you to queue messages for asynchronous processing. Message queues are great at decoupling components of applications, like the checkout and order fulfillment services for an e-commerce site. Decoupled services are easier to reason about, deploy, and implement, allowing you to ship featu...
HFrameworks bundle to a single output Worker. This is also the same for Pages, as all routes are bundled into a single output "Functions" bundle
TAwesome, good to know, thank you!
JIs it true that the cache api may not store the content-length header if I'm caching a stream?
BHello everyone, I have what I think is a pretty basic question about workers and environment variables / secrets. I've had a couple instances where my runtime environment variables and secrets disappear in the dashboard and I have to re-add them. I'm adding my runtime environment variables (not secrets) to my wrangler.jsonc as vars and my assumption was that once I add my secrets in the dashboard they just stay there until I delete or rotate them. Is my understanding correct? Are there any configuration things that could be causing this behavior? I currently have my workers set to auto-build after a merge in my git branch (not deploying from IDE). Any info helps, thanks!
TActually, I'm sorry, one follow up to cement my understanding, does this mean that framework-based static assets workers are also boiled down to one worker that is composed of all the static assets, the SSR calls, AND one conventional worker definition deployed as one bundle? I guess this is a roundabout way of asking how static asset routing works in that case since my vague mental model is that SSR elements in a framework are neither asset routes nor calls to the main worker definition. Like are those routes treated as, to coin a terrible confusing phrase, static dynamic assets?
IStatic assets aren't included in the bundle. Basically:
- Bundle: Functions (for Pages), or your Worker + any deps in the Worker + anything the Worker directly imports
- Deploy alongside: everything else (images, audio, html, whatever that is not imported into your Worker)
IThe "deploy alongside" actually uses KV but it's all handled for your by the platform (either Pages or Workers Assets)
TI think I get it, and I apologize if this is annoyingly repatative, into which of those two buckets does an SSR endpoint fall? Or more practically: Is anything that's not a freely served asset counted/billed as a worker invocation?
ISSR would be part of the bundle and which is billed as a Worker request because the Worker is doing some work (to SSR) the files.
IFor example, if you rendered a page and served it with SSR, the request for that page would be billed, and then if the page was then sent to the browser and then an image on that page was loaded, that would be an asset request not a Worker request.
JAnyone got any advice? I'm streaming audio data through since it's large and I KNOW the content-length, but I can't send a streaming response and set content-length headers. The header gets removed. Is there a way to get around that?
IThe Cache API should be able to store streaming objects but there's a limit of 512MB for the Cache API which you may be running into?
IHow big are your objects?
DCan my Worker call a path served by a Snippet? Or is it like my Worker calling a Worker, where I need a binding?
小
小Make a joke
小:stare:
AWhy did I get a connection error? It only happened once.
JCloudflare it’s time to raise the 1 MB message size limit on Websockets. I beg you to make this change for developer’s week.
JI am able to store the object (they are under 512mb, though I wasn't aware of that limit). The issue I'm referring to is when i send a response object, and set the content-length if it's a streaming response, the content-length is ignored and it winds up blank at the client.
IPer the official spec, if the response is streamed it's not allowed to have a Content-Length (https://datatracker.ietf.org/doc/html/rfc9112#section-6.2) so I expect that's why Cloudflare's stripping it before sending it to the client.
IMost clients should handle streaming responses properly though, what client is this?
IHmmm that's odd. Do you see the full content being downloaded in the network tab of the browser?
森first thing came to my mind is Next.js support, I know you could deploy open next js; and there is indeed guides to use opennextjs on to existing nextjs apps, but it isn't clear if it's fully compatible with current nextjs version (that is deployed to cloudflare pages) or not.
くCan I use libraries that rely on native C++ bindings with Cloudflare Pages?
INo, because Pages uses Workers in the background and Workers only support JS, WASM, or anything that can be transpiled to JS or WASM (TS, Rust).
