But either way you should see your updated KV value everywhere after about 60 seconds. If you were hitting a non-existent key before you created it, that's equivalent to an update, the empty value itself will be cached in colos. So it depends what your test intended to... test, often an app won't even know what key to try and fetch until it has been created and shared via some other means, and in that case it should be instantly available (since it won't have been cached as a miss anywhere)

Yea 60sec for us is way too long. Maybe if there was a way to instruct a browser page to always hit the same colo?

Are you actually updating existing keys containing these videos in your app? Or are they stable once created?

Im pretty sure that violate the TOS 2.8;

If not i will glad to be corrected.

Yea right it's against rules i thought so too

I tested on updates on a totally unrelated project.

In this project I would not be updating existing keys

I wanted to do this for one of my projects. But even on Enterprise plan i could not get a good deal on high bandwidth video traffic. But you can get good bandwidth fees using some EU providers, my Workers redirect the traffic to those points using 307 redirects when video files.

I want make full use of cloudflare but those "american" bandwidth fees is not for our budget;

Anyway.... Im testing DO for the past month, except the problem with traffic, on the technical side i have few questions: -- All wrangler args for DO like --new-class, --delete-class are a pain in the ass for our CI/CD; Its should be make implicity using the information on the wrangler.tom. The old bash script using rest API was better on that; -- On KV we can just look for the data stored using the panel, wrangler, or any other worker. But for DO its hard do debug the data on storage api. Will be nice have a way to query that storage using wrangler tail,/dev/whatever without make a custom route for that. -- Is the schedule on DO working? i make a regular worker with schedule just to call a DO because of that. -- Theres any plan to expose the DO without a regular worker in front? -- For worker to DO call: Any plan to call using request/import like any regular JS class without the fetch? Thats make everything less susceptible an error and can be internally optimized on when both are at same colo.

how do you use cron triggers with modules? do we just export a ScheduledEvent handler along with the fetch handler? if so, what arguments does it need?

I'm assuming you can use just a

scheduled(request: ScheduledEvent, env)

handler if that module only handles crons. https://github.com/cloudflare/workers-types/blob/master/index.d.ts says you should expect > interface ScheduledEvent { > /** > * The type of event. This will always return

"scheduled"

. > */ > type: string; > /** > * The time the

ScheduledEvent

was scheduled to be executed in > * milliseconds since January 1, 1970, UTC. > * It can be parsed as

new Date(event.scheduledTime)

> */ > scheduledTime: number; > /** > * Use this method to notify the runtime to wait for asynchronous tasks > * (e.g. logging, analytics to third-party services, streaming and caching). > * The first

event.waitUntil

to fail will be observed and recorded as the > * status in the Cron Trigger Past Events table. Otherwise, it will be > * reported as a Success. > */ > waitUntil(promise: Promise): void; > }

awesome. thank you! is there anyway to distinguish between cron triggers in code? like have one every minute that does a small amount of work and another every day that does some big task?

Huh, good question. I can't see anything that distinguishes crons from each other, and you can't pass in a url or per-cron env that I can see... I can't see anything in the docs so I'd say no, but you should definitely ask someone who actually knows what they're talking about 😄

There should be string of the cron schedule available

so you can use that string to id the job then i guess. you just can't have 2 triggers with the same schedule and do that

Does the environment parameter have a typescript definition in worker-types? Or is it just a plain object with no methods and only the fields you define?

i just use "any" for that

Hey guys, anyone have any idea why I can't get Websockets to connect? I'm using the CF chat example. Here's the code Durable Object side:

let url = new URL(request.url);
            
            switch (url.pathname) {
                case '/connect-websocket':
                    if (request.headers.get('Upgrade') !== 'websocket') {
                        return new Response('expected websocket', {status: 400});
                    }
                    
                    let ip = request.headers.get('CF-Connecting-IP');
                    let pair = new WebSocketPair();
                    
                    await this.handleSession(pair[1], ip);
                    
                    return new Response(null, {status: 101, webSocket: pair[0]})
                    break;
                case '/init-data':
                    this.data = await request.json();
                    
                    return new Response(JSON.stringify({message: 'Data initialized'}), {headers: corsHeaders});
                    break;
                default:
                    return new Response('Not found', {status: 404});
            }

Client side:

this.websocketConnection = new WebSocket('wss://secretdomain/connect-websocket?userId=20');
        this.websocketConnection.addEventListener('error', (err) => {
            console.log(err);
        });

        this.websocketConnection.addEventListener('open', (event) => {
            console.log(event);
        });

The only error I get is WebSocket connection to blabla failed Btw I can make an HTTP request just fine. Have tried to debug this for several hours 😄

Handle session is like this:

async handleSession(webSocket, ip) {
        webSocket.accept();
        
        webSocket.addEventListener('message', async (message) => {
            try {
                this.siteData = message.data;
            } catch (e) {
                console.log(e);
            }
        })
    }

Okay found the issue, don't mind the post above 😄 🤦‍♂️

Don't think it has anything else but your bindings. But I have never actually checked.. Only ever used it for that..

for clarification, will lookups using IDs derived from

idFromName

always be slower than lookups using IDs created with

newUniqueId

?

I understand that this is the case upon creating a DO for the first time, however, I'm still unsure about later instantiations of DOs

This is nice, DO limits documented, inc. WebSockets and CPU limit ❤️ https://developers.cloudflare.com/workers/platform/limits#durable-objects

@User how does that 500ms per request CPU time work if the request sets off a setTimeout?

How does one handle crons with the new ES6 class syntax

export default {
  async fetch(request, env) { ... }
  async scheduled(scheduledevent, env) { .. }
}

?

how can I have scheduled and fetch in same export?