Cloudflare Developers•2y ago

Its 1 minute cache so yeah wouldnt be a problem but KV is more expensive so might as well use DO ins

Its 1 minute cache so yeah wouldnt be a problem but KV is more expensive so might as well use DO instead

pippin_mole•4/28/24, 6:48 PM

Ok thanks for the help

UUnsmart Its 1 minute cache so yeah wouldnt be a problem but KV is more expensive so migh...

pippin_mole•4/28/24, 6:53 PM

DO is only available on paid plans - KV it is

rootalien•4/29/24, 12:06 AM

Hello, I have the paid version on the workers platform. My CPU selection is 30,000 so last. Does this show me that it won't crash no matter how many visitors are online?

Rrootalien Hello, I have the paid version on the workers platform. My CPU selection is 30,0...

Beny•4/29/24, 2:27 AM

It means your script can use 30s of CPU time for each request to the worker, which is more than enough for a regular website.
CPU time is time spent processing stuff (example parsing JSON files), not waiting for a response from external services (such as fetching from a database).
You also need to take into consideration the memory limit which is 128 MB. It's more than enough for a normal site. You'll only run into issues if you are processing huge amounts of data like large file uploads in a bad way.

pippin_mole•4/29/24, 12:36 PM

Can I use Cloudflare KV outside of workers?

Original message was deleted

pippin_mole•4/29/24, 2:08 PM

I have an app which would benefit from a simple KV solution

pippin_mole•4/29/24, 2:09 PM

But the actual python project is too big to sit in a worker - and also isnt compatible with webassembly

Ppippin_mole I have an app which would benefit from a simple KV solution

Chaika•4/29/24, 2:10 PM

your options are API or Worker Proxy

API is easier, but has a global rate limit 1200 requests per 5 minutes, shared with your account including dashboard, and all request have to go back to core/no edge caching

Worker is harder and you pay per request but you get edge caching

CChaika your options are API or Worker Proxy API is easier, but has a global rate limit...

pippin_mole•4/29/24, 2:10 PM

Is this for KV or the actual python project

Ppippin_mole Is this for KV or the actual python project

Chaika•4/29/24, 2:12 PM

what do you mean?

Chaika•4/29/24, 2:12 PM

Your python project would either use the CF API directly, or you use a CF Worker to proxy requests to the KV binding, bound to that worker

pippin_mole•4/29/24, 2:13 PM

I'm confused as to what you're talking about regarding 'your options are API or Worker Proxy' - do you mean as a replacement for KV or where I should host my Python project

CChaika Your python project would either use the CF API directly, or you use a CF Worker...

pippin_mole•4/29/24, 2:14 PM

Again, the CF Workers require a build to webassembly - some of the packages I'm using cannot compile to webassembly

Ppippin_mole I'm confused as to what you're talking about regarding 'your options are API or ...

Chaika•4/29/24, 2:14 PM

I mean as a way to access Cloudflare KV

Chaika•4/29/24, 2:14 PM

you host your python project wherever, and either you call the CF API to use KV, or you call a worker you make to access KV, both have different benefits/downsides

pippin_mole•4/29/24, 2:15 PM

Am I not able to just use something like cfkv

PyPI

cfkv

A Python wrapper for the Cloudflare KV Store API

Chaika•4/29/24, 2:15 PM

That's using the CF API, so you have those downsides as noted above

global rate limit 1200 requests per 5 minutes, shared with your account including dashboard, and all request have to go back to core/no edge caching

pippin_mole•4/29/24, 2:15 PM

I dont care so much about edge compatibility - it's only ever going to run on a single instance

pippin_mole•4/29/24, 2:16 PM

The python project is only ever going to run once per 10-15minutes

Chaika•4/29/24, 2:16 PM

then sounds like that could work for you

pippin_mole•4/29/24, 2:16 PM

I dont really see how workers can ever be used for anything more than returning 'hello world'

pippin_mole•4/29/24, 2:16 PM

Literally every package I've tried to use requires me to 'request availability from cloudflare team'

Ppippin_mole Literally every package I've tried to use requires me to 'request availability f...

Chaika•4/29/24, 2:17 PM

talking about python workers? Yea they're still very much in beta, JS workers are recommended and probably will always be the most supported

pippin_mole•4/29/24, 2:17 PM

Well yeah I tried using Rust workers, which tbh is in even more of a beta

pippin_mole•4/29/24, 2:18 PM

Both Rust + Python gave me a bad experience

Chaika•4/29/24, 2:18 PM

Yea, JS/TS is just best for workers. I mean it all runs on V8 JS Engine - same as Chrome, Both Rust and Python are just running as WASM and binding back to javascript methods/it's all still JS at the start and finish

BBeny It means your script can use 30s of CPU time for each request to the worker, whi...

rootalien•4/29/24, 4:33 PM

I have max 10mb files but what is the bad usage?

Original message was deleted

Isaac McFadyen•4/29/24, 4:38 PM

?channel-crossposting

Flare•4/29/24, 4:38 PM

Please do not post your question in multiple channels per the rules at #

welcome-and-rules. It creates confusion for people trying to help you and doesn't get your issue or question solved any faster.

Original message was deleted

rootalien•4/29/24, 4:38 PM

#stream-and-calls

Isaac McFadyen•4/29/24, 4:38 PM

You probably want to move to #stream-and-calls yeah.

Isaac McFadyen•4/29/24, 4:38 PM

I'll delete your message here.

EvanPiro•4/29/24, 4:39 PM

Is sending multi part form data with files supported in workers? I can't seem to get a trivial example working:

      const fileContent = "This is a simple text file.";
      const file = new File([fileContent], "simple_text_file.txt", {
        type: "text/plain",
      });

      formData.set("file", file);
      formData.append("purpose", "fine-tune")
      const fetchOptions: RequestInit = {
        method: "POST",
        headers: {
          Authorization: `Bearer ${apiKey}`,
        },
        body: formData,
      };

      // Make the POST request to OpenAI
      const response = await fetch("https://api.openai.com/v1/files", fetchOptions);

      const fileContent = "This is a simple text file.";
      const file = new File([fileContent], "simple_text_file.txt", {
        type: "text/plain",
      });

      formData.set("file", file);
      formData.append("purpose", "fine-tune")
      const fetchOptions: RequestInit = {
        method: "POST",
        headers: {
          Authorization: `Bearer ${apiKey}`,
        },
        body: formData,
      };

      // Make the POST request to OpenAI
      const response = await fetch("https://api.openai.com/v1/files", fetchOptions);

Louis•4/29/24, 4:46 PM

I'm working on a project where I need to upload videos to Cloudflare using JavaScript. I found a code that works fine for normal-sized videos, but I've run into an issue when trying to upload videos larger than 200 MB.

I've tried several solutions, but I haven't yet found an effective way to handle the uploading of large videos. Does anyone have an example or can you guide me on what I might be missing or how I can approach this issue?

ERROR CONNECTION

rootalien•4/29/24, 4:53 PM

#r2

an•4/29/24, 5:47 PM

Is there any way to measure cpu time? Since Date.now() doesn't work as it mentioned in this wiki https://developers.cloudflare.com/workers/runtime-apis/performance
My worker is run heavily on cpu with wasm like processing image , measure only database fetches or api fetches is not enough for request tracing, didn't give me enough info to visual the request timeline

Cloudflare Docs

Performance and timers · Cloudflare Workers docs

Measure timing, performance, and timing of subrequests and other operations.

hob8yte•4/29/24, 10:50 PM

Anyone know of resources/projects running ffmpeg.wasm on workers? I want to transcribe some pretty short videos (i think under the 10MB limit around workers size) using workers ai and know very little about both ffmpeg and wasm

Rrootalien I have max 10mb files but what is the bad usage?

Beny•4/30/24, 12:57 AM

For example loading the entire file into memory just to send it somewhere else, using streams is better for that

tonytheonly•4/30/24, 1:56 AM

I have a worker that I want to call an external api on, but I want to return a response without awaiting the the external api call. (the external api call is to just save time writing to the db on another service).

        const promise = fetch('http://localhost:3000/impression', {
            method: "POST",
            body: JSON.stringify(data),
            headers: {
                "content-type": "application/json",
            }
        });
        c.executionCtx.waitUntil(promise.then())

        flow.params = { campaignId: campaignId }
        return flow;

        const promise = fetch('http://localhost:3000/impression', {
            method: "POST",
            body: JSON.stringify(data),
            headers: {
                "content-type": "application/json",
            }
        });
        c.executionCtx.waitUntil(promise.then())

        flow.params = { campaignId: campaignId }
        return flow;

tonytheonly•4/30/24, 1:57 AM

however when I am testing it local and it is not hitting the service

tonytheonly•4/30/24, 1:58 AM

I am using hono https://hono.dev/api/context#executionctx

Context - Hono

Ultrafast web framework for Cloudflare Workers, Fastly Compute, Deno, Bun, Vercel, Node.js, and others. Fast, but not only fast.

tonytheonly•4/30/24, 2:00 AM

the worker is on localhost:8787 and the write service for the impression is on localhost:3000 I am able to create the impression successully calling it directly but the waitUntil is not hitting it when calling the localhost:8787 endpoint

tonytheonly•4/30/24, 12:15 PM

If not waitUntil what method would I use? I believe I still have to await a durable object call correct?

const promise = fetch('http://localhost:3000/impression', { method: "POST", body: JSON.stringify(data), headers: { "content-type": "application/json", } }); c.executionCtx.waitUntil(promise.then()) flow.params = { campaignId: campaignId } return flow;

Its 1 minute cache so yeah wouldnt be a problem but KV is more expensive so might as well use DO ins

Similar Threads

Similar Threads

Similar Threads