I tested with Bruno and it looks like it has no http keepalive so each request is brand new, hitting

I tested with Bruno and it looks like it has no http keepalive so each request is brand new, hitting different metal/machine and potentially different colo (in the case of lax which has multiple colos in one, would all show as lax)

Walshy•5/29/24, 3:32 AM

We're working to cut that time down (from this specific request, it should half)

Walshy•5/29/24, 3:32 AM

and we'll continue improvements there

Walshy•5/29/24, 3:32 AM

but again, I wouldn't index on cold reads. Hot reads are what you will actually see

CChaika I tested with Bruno and it looks like it has no http keepalive so each request i...

Walshy•5/29/24, 3:34 AM

makes sense, this was my suspicion

Aalgads I had this issue while doing some debugging. It appeared that if I did not cons...

avogadro•5/29/24, 11:27 PM

I did some more debugging. It turns out if I return back an object like {foo: 'bar'}{foo: 'bar'} from a method on an RpcTarget, then the call to that rpc target shows up as "cancelled". (I call a method that returns an RpcTarget/RpcStub, and then call a method on that RpcTarget class instance)
but if i return a simple string like "foo bar""foo bar" then the target shows up as "OK".

I did the above in controlled environment, with the only variable being the value being returned from the rpc method.

Aavogadro I did some more debugging. It turns out if I return back an object like `{foo: '...

Isaac McFadyen•5/30/24, 2:09 AM

This is Workers for Platforms, correct?

Isaac McFadyen•5/30/24, 2:09 AM

Or is it DO RPC?

IIsaac McFadyen This is Workers for Platforms, correct?

avogadro•5/30/24, 3:08 AM

I first noticed it with WfP but then I tried the above with normal worker to worker service bindings, and saw the same behavior.

javeed•5/31/24, 6:25 AM

Hi Folks,

I am implementing A/B split test on workers, i would like to know where can i store the data of percentage and retrieve for split percentage and show in workers

Puliczek•6/3/24, 2:47 PM

Hi, are there any startups that are using "Workers for Platforms"?

rob•6/3/24, 4:07 PM

so to confirm, the only way to publish a user's Frontend code via pages is how? Is there a REST api i can use for that or?

rob•6/3/24, 4:07 PM

do i link a github action to it or something, how can i programatiically deploy on Pages then?

PPuliczek Hi, are there any startups that are using "Workers for Platforms"?

Kai•6/3/24, 8:28 PM

https://tenor.com/view/oh-me-saturday-night-live-snl-uso-performance-thats-me-call-my-name-gif-19743416

Tenor

rob•6/3/24, 11:37 PM

same @Puliczek

Natalie•6/11/24, 2:04 PM

Don't know if this is the right time place for this, but if I want to create a storage bucket per organisation would it make sense to use WfP?

NNatalie Don't know if this is the right time place for this, but if I want to create a s...

Beny•6/11/24, 3:11 PM

WfP is to allow your customers/users to create their own workers. Those user workers are mostly for extending the capabilities of your platform or hosting the user's custom scripts.
For file storage you would use #r2.
If your customers operate on the R2 bucket directly (like via a WfP user worker) it's recommended to create a bucket for each, otherwise use 1 bucket with a folder for each customer.

BBeny WfP is to allow your customers/users to create their own workers. Those user wor...

Natalie•6/11/24, 3:12 PM

Can you query usage per folder?

NNatalie Can you query usage per folder?

Beny•6/11/24, 3:15 PM

I don't believe so, unless you add the size of each object in it together

Natalie•6/11/24, 3:26 PM

Want to give different pricing tiers in my app different storage quotas

Natalie•6/11/24, 3:29 PM

As textures and models and so on can get quite large

FFilip John I do not currently have access to workers for platforms, but it should be in pub...

ChaikaOP•6/12/24, 7:51 PM

under Workers & Pages in the dash you should be able to subscribe to it now

ChaikaOP•6/12/24, 7:51 PM

magic link: https://dash.cloudflare.com/?to=/:account/workers-for-platforms

FFilip John I just get a 404 page when clicking the magic link

ChaikaOP•6/12/24, 7:59 PM

Are you a super admin in the account?

ChaikaOP•6/12/24, 8:00 PM

if you are, might be: workers-for-platforms

javeed•6/13/24, 7:22 AM

Hi folks,

I would like to use caches.default in workers for platforms,

Or please suggest me how to cache the content which is served under workers for platforms

Jjaveed Hi folks, I would like to use caches.default in workers for platforms, Or ple...

Beny•6/15/24, 12:37 PM

cache.default doesn't work but you can use a custom cache name, for example:

let myCache = await caches.open('custom:cache');
await myCache.match(request);

let myCache = await caches.open('custom:cache');
await myCache.match(request);

See the documentation for more info.

BBeny cache.default doesn't work but you can use a custom cache name, for example: ```...

javeed•6/15/24, 4:30 PM

thank you @Beny let me try this and get back to you

Kai•6/16/24, 2:54 PM

Does every user worker get 128 MB of memory of their own?

KKai Does every user worker get 128 MB of memory of their own?

Hello, I’m Allie!•6/16/24, 2:56 PM

Every isolate does. If you have multiple requests to the same server, it may start two Worker instances in the same isolate, and thus it will share the 128 MB

HHello, I’m Allie!Every isolate does. If you have multiple requests to the same server, it may sta...

Kai•6/16/24, 3:02 PM

Oh yeah, makes sense. User workers are just workers. Thanks!

Kai•6/16/24, 4:12 PM

Just realized when looking through the cf bill that I'm not subscribed to the plan to access wfp (got access before it was released outside of Enterprise), I've submitted a support ticket but currently only on the workers paid plan so i was put at the end of the queue.

Is there somebody I can directly contact to get this resolved?

KKai https://tenor.com/view/oh-me-saturday-night-live-snl-uso-performance-thats-me-ca...

Puliczek•6/19/24, 7:57 AM

@Kai @rob Can you share landing page? Really qurious how you describe integration process

PPuliczek @Kai @rob Can you share landing page? Really qurious how you describe integratio...

Kai•6/19/24, 8:10 AM

Sure https://tryflot.com.

The process of integrating WFP was very smooth in my opinion, the only problem i faced was passing state that's attached to the session without polluting the incoming request. I ended up going with putting that state into the params_object parameter and attaching a outbound worker to grab the state from the user worker.

Flot

Flot - A workflow automation platform

A workflow automation platform that allows you to quickly connect your favorite apps and services together.

KKai Just realized when looking through the cf bill that I'm not subscribed to the pl...

rob•6/19/24, 11:57 AM

@tanushree │ Workers PM should be able
To help you tbh ,always very helpful

Andreas H•6/20/24, 2:04 PM

Hey, is there a way to get hands on an error code (e.g. https://developers.cloudflare.com/workers/observability/errors/#identify-errors-workers-metrics) when invoking a user worker from a dispatch worker fails? The only "useful" thing that seems to be available on the exception that is being thrown is a name property that probably loosely maps to those code but I'd really prefer something more explicit.

Cloudflare Docs

Errors and exceptions · Cloudflare Workers docs

Review Workers errors and exceptions.

DORSHT•6/21/24, 12:31 PM

Hi, question regarding pricing and plans.
If i purchased WFP on my account, do i still need to purchase Workers Paid for anything? Also, how does my domain plan comes into to play here? will the plan (free / pro / business) of the domain i attached to my dispatch worker affect it?

Thanks.

OOle Melhus Can we upload source maps with the wfp user script?

Walshy•6/21/24, 1:40 PM

you should be able to yes

DDORSHT Hi, question regarding pricing and plans. If i purchased WFP on my account, do i...

Walshy•6/21/24, 1:43 PM

You probably don't need it but I'd recommend having it
as for the zone plan, doesn't play a factor in Workers

Erisa•6/22/24, 8:09 AM

If you're on a startup plan then contact support, it likely needs manually enabling

I tested with Bruno and it looks like it has no http keepalive so each request is brand new, hitting

Similar Threads