durable objects + rpc are great. Must keep mining this vein
durable objects + rpc are great. Must keep mining this vein
Aitty-durable
HI'm wondering, what does
itty-durable
Zif it's not using rpc I wouldn't use it
Za thin abstraction over rpc, do, streams, etc. could be fun if it made some super clever pattern easy
Zthere are many ways to do actors and events
Zwe could call it Streaming Hashable Interface TTY
AI haven't tried the official, to me it's a fully typed magic store like Pinia for Vue but for workers
OHello everyone, Any one is using cloudflare workers as a there full serverless backend, like workers as a microserverless and the rest of services and would like to share his experiance, I'm thinking about it but I couldn't fine a clear answers about the pros and cons
KHit me up
KI like it for the most part. Would recommend.
MHello! I am building a proxy on Workers. If the proxied api is slow, will this cost me GB-s? So optimizing my api will save me money on workers? Or is only real computing power billed, like heavy calculations?
CIf you're using Workers Standard (which everyone should be outside of Enterprises, and Durable Objects), you're only billed for CPU time (the time the worker is actively doing work) and not duration (GB-s), so the api being slower shouldn't cost you anymore, as the cpu is idle while waiting
CCorrect, there was no need for a seperate channel for it
Ifully supported. Nearly every example in the docs shows both JS and TS.
ECan someone explain the difference between workers for platforms and standard? I couldnt unterstand it pretty well on internet. Thanks
AI have a worker that runs every minute, checks if there are posts newer than the last post saved in KV, and forwards them if so. On the past few days, I've noticed posts being forwarded twice, which indicates that KV is returning outdated info a minute later. I've never seen this happen before; was there a recent change that would make KV take longer to update? Is there anything I can do to avoid this happening, other than changing the interval the worker runs on? (please ping with response)
WWorkers for platforms is so you can turn workers into a SaaS product.
Docs give a good explanation: https://developers.cloudflare.com/cloudflare-for-platforms/workers-for-platforms/
Docs give a good explanation: https://developers.cloudflare.com/cloudflare-for-platforms/workers-for-platforms/
ECan i use them as normal workers? I just need the cf properties and better limits
WWonder if this is what we're seeing in #releases
I haven't had time yet to investigate.
There shouldn't have been any changes which could cause this but of you can confirm that behaviour I'd recommend posting in #kv
I haven't had time yet to investigate.
There shouldn't have been any changes which could cause this but of you can confirm that behaviour I'd recommend posting in #kv
WWfP doesn't give you more limits (except more script count) and the cf properties is available in all workers
WWhat are you hitting exactly?
EWell i have my own basic dashboard for cloudflare API and manage few hundreds of domains. Each of them should have its own worker and each worker should save headers etc from each visitor. Would this be possible with normal workers? - also there is a limit of 500 workers per account right? Can i have more workers with WFP
WI'd recommend doing that in a single Worker but err yeah, you can have unlimited with WfP so if you really wanted to go that route it'd work
EWhat would you recommend for saving the information of the visitors. Should i use the hypersrive, d1… or maybe something else
WDepends on the information
WIf it's for analytics and you just want to save like request count - Analytics Engine
WIf you need an actual DB (e.g. auth) then yeah D1 or your own DB with Hyperdrive
EOk thank you very much i appreciate it.
Ayeah looks the same - I think I first noticed it on 5/20. should I just repost that same message in #kv?
KIs an event handler (like fetch()) is always required even if you only use that worker in a service binding via RPC? I can't seem to deploy without specifying fetch.
Kyou don't need a fetch handler, but you do need a default export
Kif you're not exporting your WorkerEntrypoint class as the default, Wrangler will get upset
THello there!
I built a multiplayer game server with node.js that communicates with the clients via websockets and I was wondering whether cloudflare workers could be used to host it (currently I'm using a traditional vps):exitedCat:
I built a multiplayer game server with node.js that communicates with the clients via websockets and I was wondering whether cloudflare workers could be used to host it (currently I'm using a traditional vps):exitedCat:
DHey, workers (and more specifically, Durable Objects with their websocket hibernation api) would be perfect, assuming you can adapt your data storage solution. #durable-objects is the best channel to ask if you get stuck, the docs don't contain enough examples to understand everything
Tthanks for the quick answer :) I'll make same research regarding durable objects then
Jhow can i clear everything i have cached using the workers cache api cache.put()?? i tried a curl {purge_everything:true} using the api but the cached response is not being deleted
KI'm looking for an example of a successfully built workers-sdk / workerd stack --- I'm fixing a bug in workerd myself and would like to compile it end-to-end. Anybody have any examples / resources?
PQuestion regarding durable objects (DO). If I have one DO class but let’s say 100 instances. Will I be billed 100 x 128mb or only once 128mb if they all do in practice not occupy more than that? It is not explained very well in the documentation
Cloudflare Docs
Durable Objects are only available on the Workers Paid plan.
CThose footnotes go over it pretty well:
4 Duration is billed in wall-clock time as long as the Object is active, but is shared across all requests active on an Object at once. Once your Object finishes responding to all requests, it will stop incurring duration charges. Calling accept() on a WebSocket in an Object will incur duration charges for the entire time the WebSocket is connected. If you prefer, use state.acceptWebSocket() instead, which will stop incurring duration charges once all event handlers finish running.
5 Duration billing charges for the 128 MB of memory your Durable Object is allocated, regardless of actual usage. If your account creates many instances of a single Durable Object class, Durable Objects may run in the same isolate on the same physical machine and share the 128 MB of memory. These Durable Objects are still billed as if they are allocated a full 128 MB of memory.
Cif they were all constantly active, you would get billed 100 x 128 mb. Otherwise it's just a static billing for 128 mb x the time that instance is active
PSo I could read it as if they are still billed individually?
CWhat do you mean by
if they all do in practice not occupy more than that?
PLike. If each instance uses 24/7 1mb. Do I get billed 100 x 128mb?
CCorrect, as the footnote says
Duration billing charges for the 128 MB of memory your Durable Object is allocated, regardless of actual usage
PJust to confirm. The footnote means each DO instance and not DO class?
