Also another question @skoogeer after you've covered the previous, in that article you include: > Ma

JJames Also another question @skoogeer after you've covered the previous, in that artic...

D

dave•6/21/24, 10:36 PM

that's a feature to not have! ;P

D

dave•6/21/24, 11:07 PM

do I need to set

version_metadata

version_metadata

for each of my envs, or is the key enough to have in the root just once?

D

dave•6/21/24, 11:09 PM

answer, I need it for each env

D

dave•6/21/24, 11:17 PM

@Unsmart general-discussions

I found a good solution for this (Backblaze B2 erroring with 503, 429, 500, etc):

Setup: I set up two buckets, one in us-east and one in us-west. Then, I set up two bidirectional cloud replication rules.

In my Worker, I try to upload to the us-east bucket first, and flip flop between us-west and us-east for repeated attempts until it works.

Eventually, both buckets will contain 1:1 of the same data, even when there's an outage too in one region.

K

kindaraffy•6/22/24, 10:24 AM

Hi, you guys might want to extend the cf docs on worker RPC to include the utility type

InstanceType

InstanceType

. It automatically types your RPC classes so you could use your RPC methods in other workers with type-safety.

https://developers.cloudflare.com/workers/runtime-apis/rpc/

An example:

import { RpcTarget, WorkerEntrypoint } from 'cloudflare:workers';

class Counter extends RpcTarget {
    #value = 0;

    increment(amount: number = 1) {
        this.#value += amount;
        return this.#value;
    }

    get value() {
        return this.#value;
    }
}

export class CounterService extends WorkerEntrypoint {
    async newCounter() {
        return new Counter();
    }
}

export type CounterServiceStub = InstanceType<typeof CounterService>;

import { RpcTarget, WorkerEntrypoint } from 'cloudflare:workers';

class Counter extends RpcTarget {
    #value = 0;

    increment(amount: number = 1) {
        this.#value += amount;
        return this.#value;
    }

    get value() {
        return this.#value;
    }
}

export class CounterService extends WorkerEntrypoint {
    async newCounter() {
        return new Counter();
    }
}

export type CounterServiceStub = InstanceType<typeof CounterService>;

import { RpcTarget, WorkerEntrypoint } from 'cloudflare:workers';

class Counter extends RpcTarget {
    #value = 0;

    increment(amount: number = 1) {
        this.#value += amount;
        return this.#value;
    }

    get value() {
        return this.#value;
    }
}

export class CounterService extends WorkerEntrypoint {
    async newCounter() {
        return new Counter();
    }
}

export type CounterServiceStub = InstanceType<typeof CounterService>;

import { RpcTarget, WorkerEntrypoint } from 'cloudflare:workers';

class Counter extends RpcTarget {
    #value = 0;

    increment(amount: number = 1) {
        this.#value += amount;
        return this.#value;
    }

    get value() {
        return this.#value;
    }
}

export class CounterService extends WorkerEntrypoint {
    async newCounter() {
        return new Counter();
    }
}

export type CounterServiceStub = InstanceType<typeof CounterService>;

Then, use the CounterServiceStub in your RPC client for type-checked method signatures for the downstream worker you are calling.

Cloudflare Docs

Remote-procedure call (RPC) · Cloudflare Workers docs

The built-in, JavaScript-native RPC system built into Workers and Durable Objects.

Kkindaraffy Hi, you guys might want to extend the cf docs on worker RPC to include the utili...

K

kian•6/22/24, 6:28 PM

That is what the

Service

Service

and

DurableObjectNamespace

DurableObjectNamespace

generic types do, isn't it?

https://developers.cloudflare.com/workers/runtime-apis/rpc/typescript/

J

johtso•6/22/24, 6:33 PM

What's up with not being able to deploy workers with

logpush = true

logpush = true

when you're a member of another account? https://github.com/cloudflare/workers-sdk/issues/2991

J

johtso•6/22/24, 6:33 PM

isn't this an issue that loads of people should be bumping into? i'm confused as to why the issue was closed

J

johtso•6/22/24, 6:33 PM

not sure how removing logpush from your config is a solution

K

kian•6/22/24, 6:37 PM

If you're having issues with Logpush causing API errors then I'd recommend opening a new issue and just reference that one since it's more than a year old. Unless your account isn't able to use Logpush, it shouldn't cause errors.

H

Harris•6/22/24, 6:39 PM

Has anyone encountered a situation where trying to listen to the live log stream of a durable object, just causes all websocket connections to fail?

I can't have my game clients connect to the durable object and listen to live logs at the same time and it makes debugging quite hard

H

Harris•6/22/24, 6:39 PM

I get

{
      "name": "Error",
      "message": "Subrequest depth limit exceeded. This request recursed through Workers too many times. This can happen e.g. if you have a Worker or Durable Object that calls other Workers or objects recursively.",
      "timestamp": 1719081484425
    }

{
      "name": "Error",
      "message": "Subrequest depth limit exceeded. This request recursed through Workers too many times. This can happen e.g. if you have a Worker or Durable Object that calls other Workers or objects recursively.",
      "timestamp": 1719081484425
    }

{
      "name": "Error",
      "message": "Subrequest depth limit exceeded. This request recursed through Workers too many times. This can happen e.g. if you have a Worker or Durable Object that calls other Workers or objects recursively.",
      "timestamp": 1719081484425
    }

{
      "name": "Error",
      "message": "Subrequest depth limit exceeded. This request recursed through Workers too many times. This can happen e.g. if you have a Worker or Durable Object that calls other Workers or objects recursively.",
      "timestamp": 1719081484425
    }

no matter what I do, but only when I try to listen to live logs

edit: https://community.cloudflare.com/t/worker-throwing-recursion-error-when-calling-a-durable-object-even-though-there-is/640749/4 i see, is an error with smart mode

Jjohtso What's up with not being able to deploy workers with `logpush = true` when you'r...

W

Walshy•6/22/24, 6:42 PM

What's the error you hit?

WWalshy What's the error you hit?

J

johtso•6/22/24, 6:42 PM

workers.api.error.unauthorized [code: 10023]

W

Walshy•6/22/24, 6:43 PM

Sounds like that account doesn't have workers paid

J

johtso•6/22/24, 6:43 PM

ooooo

J

johtso•6/22/24, 6:43 PM

J

johtso•6/22/24, 6:43 PM

friendly error messages would be pretty awesome

W

Walshy•6/22/24, 6:44 PM

Workers errors are not always the clearest today :blobsweats:
cc: @tanushree │ Workers PM it'd be good to do an error audit at some point

J

johtso•6/22/24, 7:05 PM

oof, "Can't read from request stream after response has been sent." only seemed to happen when deployed, not when running locally

J

johtso•6/22/24, 7:05 PM

I guess it's a wrangler inconsistency

J

johtso•6/22/24, 9:25 PM

does this look right for forking a request to two separate endpoints?

  const [body, body2] = request.body.tee()
  const result = await Promise.allSettled([
    fetch(url1, { method: request.method, headers: request.headers, body: body }),
    fetch(url2, { method: request.method, headers: request.headers, body: body2 })
  ])

  const [body, body2] = request.body.tee()
  const result = await Promise.allSettled([
    fetch(url1, { method: request.method, headers: request.headers, body: body }),
    fetch(url2, { method: request.method, headers: request.headers, body: body2 })
  ])

  const [body, body2] = request.body.tee()
  const result = await Promise.allSettled([
    fetch(url1, { method: request.method, headers: request.headers, body: body }),
    fetch(url2, { method: request.method, headers: request.headers, body: body2 })
  ])

  const [body, body2] = request.body.tee()
  const result = await Promise.allSettled([
    fetch(url1, { method: request.method, headers: request.headers, body: body }),
    fetch(url2, { method: request.method, headers: request.headers, body: body2 })
  ])

J

johtso•6/22/24, 9:34 PM

hmm strange, so I'm forwarding the request to another workers.dev url, but I'm getting a 404

J

johtso•6/22/24, 9:34 PM

it works fine if I make the request directly to that url

J

johtso•6/22/24, 9:42 PM

oh right.. it's a workers limitation

J

johtso•6/22/24, 10:06 PM

so I think this is what I need to do then

  const [body, body2] = request.body.tee()
  const result = await Promise.allSettled([
    env.MY_SERVICE.fetch(new Request(request, { body })),
    fetch(url2, { method: request.method, headers: request.headers, body: body2 })
  ])

  const [body, body2] = request.body.tee()
  const result = await Promise.allSettled([
    env.MY_SERVICE.fetch(new Request(request, { body })),
    fetch(url2, { method: request.method, headers: request.headers, body: body2 })
  ])

  const [body, body2] = request.body.tee()
  const result = await Promise.allSettled([
    env.MY_SERVICE.fetch(new Request(request, { body })),
    fetch(url2, { method: request.method, headers: request.headers, body: body2 })
  ])

  const [body, body2] = request.body.tee()
  const result = await Promise.allSettled([
    env.MY_SERVICE.fetch(new Request(request, { body })),
    fetch(url2, { method: request.method, headers: request.headers, body: body2 })
  ])

E

Elektra•6/23/24, 5:23 AM

Hi everyone... is anyone implementing vertex ai in their cloudflare worker?

E

Elektra•6/23/24, 5:24 AM

How are you working with authentication?

S

Silent•6/23/24, 8:48 AM

I need to store a simple kv value but i dont really want to use workers KV since its only going to be used internally. My main storage is R2 but if i use it for this i will be doubling my A/B operations. Any other solutions or should i use r2?

SSilent I need to store a simple kv value but i dont really want to use workers KV since...

J

johtso•6/23/24, 9:02 AM

Why do you not want to use kv?

J

johtso•6/23/24, 9:02 AM

I don't really understand your reason

S

Silent•6/23/24, 9:42 AM

Kv is eventually consistent

F

Furina de Fontaine!!!•6/23/24, 10:37 AM

Is 10ms cpu time enough for requests to apis (lets say discord)?

F

Furina de Fontaine!!!•6/23/24, 10:37 AM

For the free plan

SSilent Kv is eventually consistent

J

johtso•6/23/24, 12:12 PM

d1?

J

johtso•6/23/24, 12:13 PM

is there any reason why a netlify rewrite to a cloudflare worker might not work?

/* https://my-worker.my-account.workers.dev/:splat 200

J

johtso•6/23/24, 12:13 PM

doing the same thing but to something like webhook.site works fine

FFurina de Fontaine!!!Is 10ms cpu time enough for requests to apis (lets say discord)?

A

AngusMa•6/23/24, 1:48 PM

I think web requests waiting time aren’t counted in 10ms. In my case, I have two workers that make requests to web APIs and they work fine.

AAngusMa I think web requests waiting time aren’t counted in 10ms. In my case, I have two...

F

Furina de Fontaine!!!•6/23/24, 1:50 PM

Im on free tier. So to confirm api requests arent counted as cpu time right? Also could you tell me about your workers, like how frequently are they called and how long do they take, what they do (as generally)

J

johtso•6/23/24, 1:51 PM

it's cpu time, so based on the amount of work the cpu is doing

J

johtso•6/23/24, 1:51 PM

waiting for a web request doesn't use any cpu

J

johtso•6/23/24, 1:52 PM

there are no limits on how long the worker can run

Jjohtso waiting for a web request doesn't use any cpu

F

Furina de Fontaine!!!•6/23/24, 1:53 PM

I see, thanks. Lets say I invoke 10 web requests and do json parsing. Assuming each json object is around 1000 characters, will it take more time or stay within 10ms?

F

Furina de Fontaine!!!•6/23/24, 1:53 PM

The json parsing ofc

FFurina de Fontaine!!!I see, thanks. Lets say I invoke 10 web requests and do json parsing. Assuming e...

J

johtso•6/23/24, 1:57 PM

try it and see what the cpu time is in the analytics?

Jjohtso try it and see what the cpu time is in the analytics?

F

Furina de Fontaine!!!•6/23/24, 1:59 PM

Yeah thanks

A