I am currently working on a project involving the deployment of various workers / pages in a Turborepo. Ideally, I'm aiming to trigger deployments only for the applications that have been updated, rather than deploying all applications with each change. This would certainly help streamline the process and optimize resources. Does anyone have a reference or guide that provides a step-by-step procedure for achieving this? Any insights, suggestions, or resources would be much appreciated!

Hey so my problem is that my 10 minutes cron trigger run and execute twice but i only setup one and handle one, that happend randomly yesterday without any changes of my side. I switched the timer now to 8 minutes and 15 minutes and all works fine, only one execute but if i switch back to 10 minutes it runs again twice all 10 minutes My Scheduled Handler: my wrangler.toml cron settings: past cron logs: you can see twice execution cron dashboard settings:

Hello! I've just received an email saying that I reached 50% of the daily limit for KV for a project that I haven't modified in months and for which the route is disabled. So I checked the cron events logs and I noticed they are called 3 times every time 🤔 I don't understand why and today is the first time it happens.

Hello! im trying to allow a domain through CORS using https://developers.cloudflare.com/workers/examples/cors-header-proxy I have a frontend sitting at domainA.com that calls an api at domainB.com for some reason I get this whenever a request fails

workerd/server/server.c++:2347: error: Uncaught exception: kj/async-io-win32.c++:385: failed: WSARecv(): #64 The specified network name is no longer available.
stack: 7ff680e88950 0 0 0 0

doesnt seem like its a workers issue since stack overflow wasnt helpful but also I may be doing this all wrong (exact code https://pastebin.com/dT4G1xnE)

I have a Worker deployed that handles redirections for users based on their experiment grouping for an A/B test. The Worker handles 4 cases: * If the user has an existing analytics cookie, and the user is in: - the control group - the variant group * If the user has no analytics cookie yet (i.e. is a new user), and the user is in: - the variant group - the control group For the variant group cases, the worker redirects users to the experimental URL:

const statusCode = 302
const resp = new Response(null, {
  status: statusCode,
  headers: new Headers({ Location: EXPERIMENTATION_URL }),
  })
return resp

For the control group, we send the user through to the original site they requested:

let resp = await fetch(request)
resp = new Response(resp.body, resp)
resp.headers.append("Set-Cookie", newCookie) // only if they have no cookie yet
return resp

However, for some reason, in about 50% of the cases that we hit the case where the user has no analytics cookie yet, and is in the control group, the request in the Worker logs shows up as canceled. A few seconds after the canceled request, there is usually a successful request, which seems to be a retry, but this results in the user being assigned to a variant group. This case is the last case to get handled in the Worker code. However, I am a bit baffled as this is the only case that fails. Is there any way to check why these requests are being canceled? Or are there any common reasons why requests get canceled, especially if the requests only get canceled in a specific code path ?

Hi there, I'm trying to add an R2 binding to my (free) Worker, but I get the error message:

Workers Unbound has not yet been enabled for this account. (Code: 10075)

I don't have any other variables yet, so I assume this would be allowed. I can't seem to find much information on this. Does anybody know how to resolve it other than by credit card? 🙂

Hello everyone, before upgrading to wrangler@3 (and now using the fully local development mode) the port was normally on

8787

. Now each start of the environment seems to have a random port (at the time of trying its

64278

). As the Port is not always the same and a new restart creates a new port, it is a pain to update the development frontend env or my api testing client (insomnia) to always be at the correct port. Is this a bug on macOS or is there a way to force the port to be at 8787 ? Thanks in advance, David

Does html-rewriter have support for css or is there something equivalent?

it seems like a lot of clients are being disconnected. how can i debug / be able to properly handle that? can i respond with something right before its disconnected?

When I static import some wasm libraries, I often get the error

Error: Script startup exceeded CPU time limit. [code: 10021] when using static import

when deploying. It is resolved when I use dynamic import (

import('...

), but dynamic import apparently doesn't work with wasm. Any suggestions?

In the past two days my cron trigger has been triggering twice. I've set only one trigger at 8:00 UTC but it is triggering twice within the minute. I've searched in this forum and I've found someone also encountered the same situation but seems like it resolved itself without any of their action.

Hello does anybody uses hono v3.0.0 or superior with html rewriters it seems not working for us. Context.res seem immutable...

Hi there, I'm trying to use cloudflare worker to change the instances of hostname of a sitemap that is generated by Webflow. We have to use this method as our webflow is reverse-proxied but Webflow customization is a bit lacking and won't use the global canonical URL in the auto-generated sitemap.xml. Gist of what I'm doing is fetching the actual webflow sitemap.xml, replacing all instances of the hostname with what we want to be user-facing and SEO-facing, serving the modified sitemap.xml. I'm having an issue where cloudflare worker is getting no response when fetching. The sitemap resource is public and works fine in postman GET.

async function handleSitemap() {
  const sitemapUrl = 'https://tempsite.com/sitemap.xml'
  
  const headers = {
    'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/89.0.4389.82 Safari/537.36',
    'Accept-Language': 'en-US,en;q=0.9'
  }

  let response = await fetch(sitemapUrl, {method: 'GET', headers: headers})
  console.log(response.body)

  if (!response.ok) {
    throw new Error('No response!')
  }
  
  let data = await response.text()
  const contentType = response.headers.get('content-type')

  // Replace 'tempsite.com' with 'mysite.com'
  data = data.replace(/tempsite\.com/g, 'mysite.com')

  return new Response(data, {
    headers: {
      'content-type': contentType
    }
  })
}

Hey all! I am working on UGC file uploading in my app. This goes through a CF Worker. I am wondering how I would go about piping the media from the worker to one API (for checking for illegal content pre upload) and afterward piping that to an s3 bucket via aws4fetch. Anyone know how I would go about it?

Hi there, how do I detect whether my worker was invoked via a HTTP Route or via a service binding from another worker? I would like to skip auth + implement custom logic in the latter scenario

i have a really weird bug when working with cloudflare workers (compiling from typescript) when I set some data to a Map from command.ts, all entries disappear in callback.ts code: https://tsplay.dev/NVRlqN

After updating to wrangler 3, I am getting this error

workerd/jsg/util.c++:276: error: e = kj/compat/tls.c++:215: failed: TLS peer's certificate is not trusted; reason = unable to get local issuer certificate
stack: 7ff6229c8a6f 7ff622a0c9dd 0 0 0 0 0 0 0 0 0 0 0 0 7ff6229d6234; sentryErrorContext = jsgInternalError

Even when i explicitly pass the --local-protocol http

npx wrangler dev src/index.ts --local-protocol http

My workers script with d1 binding, and '/api/view/customers_data' on GET request is succesful but sending POST request with body to '/api/add/customer' giving error Fetch request

const response = await fetch(
        "https://xxxx.xxxxx.workers.dev/api/add/customer",
        {
          method: "POST",
          headers: {
            "Content-Type": "application/json",
          },
          body: JSON.stringify({
        id: xxx,
        cabin: xxxx,
        name: xxxx
      },)
        }
      );

// Workers Script

export default {
  async fetch(request, env, ctx) {
    const corsHeaders = {
      'Access-Control-Allow-Origin': '*',
      'Access-Control-Allow-Headers': '*',
      'Access-Control-Allow-Methods': 'GET, POST',
    };
    const { pathname } = new URL(request.url);
    try {
      if (pathname === "/api/view/customers_data") {
        const results = await env.DB.prepare(
          "SELECT * FROM customers_data "
        ).all();
        return new Response(JSON.stringify(results), {
          headers: corsHeaders,
        });
      }else if (pathname === "/api/add/customer") {
        const body = await request.json();
        const results = await env.DB.prepare(
          "INSERT INTO customers_data (cabin, name, email, phone, gender) VALUES(?1, ?2, ?3, ?4, ?5)"
        ).bind(body.cabin, body.name, body.email, body.phone, body.gender).run();
return new Response(JSON.stringify(results), {
          headers: corsHeaders,
        });      }
    } catch (e) {
      return Response.json(e);
    }
   
    });  
  }
};

//Error Access to fetch at 'https://xxx.xxxx.workers.dev/api/add/customer' from origin 'http://localhost:3001' has been blocked by CORS policy: Response to preflight request doesn't pass access control check: No 'Access-Control-Allow-Origin' header is present on the requested resource. If an opaque response serves your needs, set the request's mode to 'no-cors' to fetch the resource with CORS disabled.

What gives?

I have a Rust worker bound to a queue. It gets

Unknown Event - Exceeded CPU Limit

errors without doing very much, e.g. just deserializing the message batch exactly as in the

worker-rs

example for queues. This happen with a single message. I tried setting

usage_model=unbound

in wrangler.toml and I see it picked up in the UI, but the error persists. Am I missing something obvious?

Has anyone found a way to skip a module dependancy when bundling? I'm using

googleapis

, which has a

require('http2')

while

http2

isn't available in the runtime. That might be fine -- I think I can pass runtime options to avoid using it. I just need to find a way to get

esbuild

to ignore the missing dependancy.

Hi all, We're trying to use a workers sites bucket defined in our API wrangler to host some (temporary) image assets in to show in our app, but once we attempt to fetch these we see that they are suffixed with hashes. What is the recommended way to continue here so we can access assets in these buckets through e.g.

/public/some-temp-image.png

? We use getAssetFromKV:

return await getAssetFromKV(
      {
        request: ctx.req.raw,
        waitUntil: ctx.executionCtx.waitUntil.bind(ctx.executionCtx),
      },
      {
        ASSET_NAMESPACE: ctx.env.__STATIC_CONTENT,
        mapRequestToAsset: (req) => {
          const url = new URL(req.url);
          url.pathname = url.pathname.replace(/^\/public/, "");
          return new Request(url.toString(), req);
        },
      },
    );
  } catch (e) {
    if (e instanceof Error) {
      console.error(e);
      if ("status" in e) {
        if (e.status === 404) {
          return ctx.notFound();
        }
      }
    }
    throw e;
  }

Can we skip workers sites from adding these hashes?

Running into a situation where Google search is unable to index the site running as pages. Robots.txt renders as index.html for some reason. I did create explicit robots.txt in my repo and I'm pretty sure it's propagating to pages - anyone know what might be happening?

I'm converting my worker to module but the part where it gets posted data request.json() is no longer working. What am I doing wrong? I verified the payload is being sent. async fetch(request, env, ctx) { request.json() // return null }

Hi there. When I tried to use Quick Edit while the browser language is Chinese Simplified/Traditional/Hong Kong, the Quick Edit will just white out, while other languages won't (even those minor languages can still be loaded, the Quick Edit will just be presented in English). I guess this is a localization issue. Please fix it. Thanks a lot.

I'm wrapping up development of my api on top of Cloudflare's workers and am about to start setting up rate-limitting. Initially wanted to do this via serverles redis, but it looks like WAF https://developers.cloudflare.com/waf/ achieves the same. But I cant manage to find this setting under my workers page. I believe it is only available for domains? Am I correct in assuming that in order to limit my workers routes, I can only do so when custom domain is set up for it?

I created a few workers directly using the dashboard. Is there a way PULL a current worker into a folder via wrangler? The online "quick edit" isn't working right now, and I'd like to start using it via vscode.

I love developing with workers, however I continue to get errors about the startup limits (CPU time). They seem quite random and hard to debug. Pushing multiple times sometimes works. Coming from AWS Lambda, the global scope is no bad place to cache certain data. How should we do that in Cloudflare workers? E.g. Reusing Database Connections etc. Let´s take this code, it wont deploy, however I am not doing anything in the global scope at coldstart:

js
import {
  AICache,
  authorize,
  CorsResponse,
  OpenAIStreamHandler,
} from "@tm9657/backend-worker";
import GPT3Tokenizer from "gpt3-tokenizer";
import { Configuration, OpenAIApi } from "openai";

export interface Env {
  DATABASE_URL: string;
  OPENAI_KEY: string;
}

type AIRequestBody = {
  prompt: string;
  system?: string;
  top_p?: number;
  frequency_penalty?: number;
  max_tokens?: number;
};

let cache: AICache | null = null;

export default {
  async fetch(
    request: Request,
    env: Env,
    ctx: ExecutionContext
  ): Promise<Response> {
    const auth = await authorize(request);
    if (!auth) return new CorsResponse("Unauthorized", 401).finalize(request);

    const body: AIRequestBody = await request.json();
    if (!body || !body?.prompt)
      return new CorsResponse("Bad Request", 400).finalize(request);
    if (!cache) cache = new AICache();
    cache.init(env.DATABASE_URL);
    let openai: OpenAIApi = new OpenAIApi(
      new Configuration({
        apiKey: env.OPENAI_KEY,
      })
    );

    if (!openai)
      return new CorsResponse("Internal Server Error", 500).finalize(request);

    const usage = await cache.getUsage(auth.sub);
    if (!usage.getHardLimit())
      return new CorsResponse(
        "Monthly API limit hit, please upgrade your subscription!",
        429
      ).finalize(request);

    const tokenizer = new GPT3Tokenizer({ type: "gpt3" });
    const { readable, writable } = new TransformStream();

    const openAIRequest = await openai.createChatCompletion(
      {
        model: "gpt-3.5-turbo",
        messages: [
          { role: "system", content: body.system || "" },
          { role: "user", content: `${body.prompt}` },
        ],
        top_p: body.top_p || 0.05,
        max_tokens:
          4096 - tokenizer.encode(`${body.prompt} ${body.system}`).bpe.length,
        user: auth.sub,
        frequency_penalty: body.frequency_penalty || 1.0,
        stream: true,
      },
      { responseType: "stream" }
    );

    const writableStream = writable.getWriter();

    let total = "";
    const handler = new OpenAIStreamHandler(
      openAIRequest,
      async (message) => {
        total += message;
        await writableStream.write({ total, message });
      },
      async () => {
        await writableStream.close();
        await cache?.updateUsage(
          auth.sub,
          "chatGPT",
          tokenizer.encode(`${body.prompt} ${body.system} ${total}`).bpe.length
        );
      }
    ).promise();

    return new CorsResponse(readable).finalize(request);
  },
};

I'm trying to publish adding durable object to my worker. I have a new file in the project folder called rate-limiter.js. I'm getting this error, when running wrangler publish. * Cannot create binding for class in script rate-limiter that does not exist [code: 10061]*

// rate-limiter.js
export class APIRequestCounter {}

}

// index.js
import { APIRequestCounter } from './rate-limiter.js';
export default {
  async fetch(request, env, ctx) {
  }
}

TRY 1: WRANGLER.toml

[durable_objects]
bindings = [{name = "API_RATE_LIMITER", class_name = "APIRequestCounter", script_name = "rate-limiter.js"}]

TRY 2: WRANGLER

[durable_objects]
bindings = [{name = "API_RATE_LIMITER", class_name = "APIRequestCounter", script_name = "rate-limiter"}]

TRY 3: WRANGLER

[durable_objects]
bindings = [{name = "API_RATE_LIMITER", class_name = "APIRequestCounter", script_name = "./rate-limiter.js"}]

Using workers as a reverse proxy for .example.com to return example.com/id. I've checked the network sources, and the assets are in deed coming through and not 404-ing. But there seems to be parsing error, somewhere.

ts

export default {
  async fetch(request: {
    url: string | URL; headers: { get: (arg0: string) => string } 
}, env: any, ctx: any) {
    try {
      const subdomain = request.headers.get('host').split('.')[0];
      // ... map subdomain to id
      const destinationURL = new URL(`http://localhost:3001/${id}`);
      const response = await fetch(destinationURL);
      const proxyResponse = new Response(response.body, { headers: response.headers, status: response.status, statusText: response.statusText})
      return proxyResponse
    } catch (error) {
      return new Response(null, { status: 404, statusText: "Oops can't find that page" });
    }
  },
};

What am I missing here?