Thanks. Yes, I tried this before; I got the same error. Another issue I noticed is I can't access th
Thanks. Yes, I tried this before; I got the same error. Another issue I noticed is I can't access the "cf" request object in user workers.
Rreally? So their workers don't image access to the cf request object from the context ?
RIs that by design
BYeah the API seems to have a few issues unfortunately.
As for the cf props, something similar happens on DOs too - see the explanation of why here: durable-objects
If your project doesn't have a strict requirement of using the "fetch" entry point you might be able to use RPC to pass it down to the user worker (I've not tested this though)
As for the cf props, something similar happens on DOs too - see the explanation of why here: durable-objects
If your project doesn't have a strict requirement of using the "fetch" entry point you might be able to use RPC to pass it down to the user worker (I've not tested this though)
BI will check this. I have passed "cf" fields as request headers - as a quick fix.
BThere's a workaround I just figured out for deleting secrets, you can use the 'Patch Script Settings' API with a new array of bindings and it'll replace all the existing bindings with the ones specified. If you have other secrets you'd have to define them again which is a bit annoying
BI think patching does not have effect on secrets that are not included in the bindings array.
BIt did when I just tried it
BAh yeah you are right, it's actually still there but doesn't show as there in the API, strange
BThey even remain if you delete the script & upload it again, so I guess they're impossible to remove for now
AHey Folks,
Can a user worker be a consumer of a queue? if so how can I configure it?
using this configures the user worker as a producer, but not a consumer.
Should I use the dispatcher as the consumer and then route those messages to the user worker from there?
The issue there would be they will change to be fetch events instead of queue events as there's only
Can a user worker be a consumer of a queue? if so how can I configure it?
using this configures the user worker as a producer, but not a consumer.
Should I use the dispatcher as the consumer and then route those messages to the user worker from there?
The issue there would be they will change to be fetch events instead of queue events as there's only
userWorker.fetch method.
BYou asked this one before, and the answer is still the same unfortunately.
The issue is that the API to create a consumer on a queue doesn't support referencing to dispatch namespace workers. Consumers are configured on the queue rather than the worker.
The only way for now is to create a regular worker that dispatches it to the user worker. You can use RPC to pass it to a handler other than fetch
The issue is that the API to create a consumer on a queue doesn't support referencing to dispatch namespace workers. Consumers are configured on the queue rather than the worker.
The only way for now is to create a regular worker that dispatches it to the user worker. You can use RPC to pass it to a handler other than fetch
Mcomplete noob question, but I wasn't getting smart out of the docs: do user workers support scheduling?
UI assume you mean cron triggers? I would assume they do but have not tested myself.
UAnd if by scheduling you meant very specific time instead then you can use durable objects to make alarms to support invoking at a specific time
DHey, beginner question that maybe someone here can help me out with.
I'm going through the basic "Getting Started" docs of WFP, got a namespace, a user worker and a dispatcher.
So far everything works. Now, i've updated my dispatcher to be a dynamic dispatcher which by the docs is supposed to receive calls to
It doesn't say in this part of the docs how i should have configured my dispatcher's domains / routes for this to work, can someone help me understand what i'm missing?
Thanks.
I'm going through the basic "Getting Started" docs of WFP, got a namespace, a user worker and a dispatcher.
So far everything works. Now, i've updated my dispatcher to be a dynamic dispatcher which by the docs is supposed to receive calls to
<USER_WORKER_NAME>.<MY-DOMAIN>.com/* and from there extract the user worker name and call it. But, calls i'm making to *.<MY-DOMAIN>.com are not getting to my dispatcher worker (only calls to <MY-DOMAIN>.com do).It doesn't say in this part of the docs how i should have configured my dispatcher's domains / routes for this to work, can someone help me understand what i'm missing?
Thanks.
BWhich route did you put on the dispatch worker? Should be something like
*.mydomain.com/*
Mhowdy folks, sorry for the delay, but this error when deleting secrets from namespaced workers should be fixed
BWhen using #rate-limiting-beta with Workers for Platforms is there anything special implemented for the
namespace_id or is that ID still shared among all scripts?Uif you use the same id it would use the same namespace if thats what you mean
BYeah, that's how I imagine it works. Was just wondering if there had been any considerations for Workers for Platforms, such as them being scoped to each dispatch namespace.
It might become a challenge to maintain a database of namespace ids assigned to each user (to prevent them affecting each other) when there are a lot of workers across various dispatch namespaces and deployment systems.
I'm sure it wouldn't be an issue at the moment with the maximum rate limit period being 60 seconds, but that's sure to increase in the future and might affect other workers if they both end up using the same namespace id.
Perhaps an option that disables it being scoped across all workers would be useful in situations like this, where that behaviour becomes an obstacle rather than something useful.
It might become a challenge to maintain a database of namespace ids assigned to each user (to prevent them affecting each other) when there are a lot of workers across various dispatch namespaces and deployment systems.
I'm sure it wouldn't be an issue at the moment with the maximum rate limit period being 60 seconds, but that's sure to increase in the future and might affect other workers if they both end up using the same namespace id.
Perhaps an option that disables it being scoped across all workers would be useful in situations like this, where that behaviour becomes an obstacle rather than something useful.
UNot sure why it would be hard to maintain a db?
BWell in my situation that would involve creating a central database among 3 different deployment systems, with clashing user ids & script names. Of couse it is achievable, but not an easy task.
BAll that work could be avoided by a single additional option in the binding
UI highly doubt they would do that as it completely defeats the point of namespaces
BHmm, what if namespace ids could be any string and were converted to a number behind the scenes?
If we could use a dynamic string that would solve everything, such as
Knowing which script correlates to which namespace number is the tricky bit
If we could use a dynamic string that would solve everything, such as
namespace_id = "stores/[user-id]/1"Knowing which script correlates to which namespace number is the tricky bit
BYeah agreed, the option to disable it would remove the purpose of namespaces, but that might be useful for some people.
Instead of being at the account level, if they were at the dispatch namespace level for user workers this would also be helpful. It would be easier to maintain a database of ids for user workers in each dispatch namespace than across the whole account.
The option to set a scope level, like account/dispatch_namespace/self might be better
Instead of being at the account level, if they were at the dispatch namespace level for user workers this would also be helpful. It would be easier to maintain a database of ids for user workers in each dispatch namespace than across the whole account.
The option to set a scope level, like account/dispatch_namespace/self might be better
WA namespace is expected to contains thousands or millions of scripts from different customers so I don't think that'll add much more than account level
Wbut yes, I understand the need here. The rate limiting binding is a bit special here in how it works. cc: @tanushree │ Workers PM (for Monday)
YHello,
Is it possible to start tailing a WFP user worker, in the same way as:
https://developers.cloudflare.com/api/operations/worker-tail-logs-start-tail
?
Is it possible to start tailing a WFP user worker, in the same way as:
https://developers.cloudflare.com/api/operations/worker-tail-logs-start-tail
?
Interact with Cloudflare's products and services via the Cloudflare API
BThe API for that doesn't work in WFP. The recommended ways to debug a user worker are with Workers Logpush and Trace Workers.
There was a Google Docs we all got shared in the beta with a walkthrough on how to do it, but some of that information hasn't been added to the Cloudflare site yet. It's not marked as confidential so I assume I can share it here.
The "Trace Workers" in there look identical to Tail Workers, perhaps they are the same thing (renamed from trace to tail?)
There was a Google Docs we all got shared in the beta with a walkthrough on how to do it, but some of that information hasn't been added to the Cloudflare site yet. It's not marked as confidential so I assume I can share it here.
The "Trace Workers" in there look identical to Tail Workers, perhaps they are the same thing (renamed from trace to tail?)
YThanks 
YIs there a way to get a list of active traces?
returns 404:
even though I have some traces, for example calling
returns 404:
even though I have some traces, for example calling
/workers/traces/4ee25b110c9748fd9344790259c7ce73 I get 200 OK with empty body.BDoing a GET to /workers/traces works for me, it returns an array of all the traces set up
Bhttps://api.cloudflare.com/client/v4/accounts/[account_id]/workers/tracesYyep, that's the call I'm making:
I get:
I get:
BStrange, mine is working as expected, and that's not even the error message you get if the trace is not found. It is the error you get if you try to delete a trace that doesn't exist though 
Rhttps://developers.cloudflare.com/api/operations/namespace-worker-put-script-content Is this what we should be using to programattically upload a user's code to a worker on WfP?
Interact with Cloudflare's products and services via the Cloudflare API
Uyes
RInteract with Cloudflare's products and services via the Cloudflare API
BThe first API link you sent is for updating the code only
The second is the one you should use for the initial deployment and if you are changing any bindings
The second is the one you should use for the initial deployment and if you are changing any bindings
Rah i see.. and whats platform_module then ?
RWe ever get an answer for this?
BThat's the name of the file for the example, such as index.js, but theirs is platform_module.mjs
BIt lets a user script import another file uploaded by the platform instead of the user
import { platformThing } from "./platform_module.mjs";Rohh ok,. got it situated. thank you
KAs in the shape of the object for the bindings property in your metadata JSON?
K[[dispatch_namespaces]] in TOML maps to an array of these objects in JSON:KCreating a dispatch namespace is https://developers.cloudflare.com/api/operations/namespace-worker-create
