Is there anything known yet about the experimental assets in workers coming out soon? I see some act

S

Subham Saurav•7/26/24, 12:44 PM

can anyone tell me, how to get key's value from kv namespace in workers??

SSubham Saurav can anyone tell me, how to get key's value from kv namespace in workers??

I

ItsWendellOP•7/26/24, 12:51 PM

https://developers.cloudflare.com/kv/api/read-key-value-pairs/, or

env.NAMESPACE.get(key);

env.NAMESPACE.get(key);

env.NAMESPACE.get(key);

env.NAMESPACE.get(key); make sure you have the bindings in wrangler.toml for your KV namespace.

IItsWendell https://developers.cloudflare.com/kv/api/read-key-value-pairs/, or `env.NAMESPAC...

S

Subham Saurav•7/26/24, 1:07 PM

i am getting some weird issue

S

Subham Saurav•7/26/24, 1:12 PM

i previously created a key with name "hlp" then deleted it but still it showing me on calling list()

K

kev-ac•7/26/24, 1:13 PM

You're using local mode in wrangler, which uses a local KV store afaik.

S

Subham Saurav•7/26/24, 1:14 PM

but when i write new data with put() it also writes on dashboard's namespace

SSubham Saurav but when i write new data with put() it also writes on dashboard's namespace

K

kev-ac•7/26/24, 1:17 PM

Did you use

wrangler dev --remote

wrangler dev --remote

before? Then the code runs in a CF datacenter and has access to the remote KV store.

As documented it should use a local store to not affect prod data when using local mode (the default):
https://developers.cloudflare.com/kv/get-started/#4-access-your-kv-namespace-from-a-worker

Cloudflare Docs

Get started · Cloudflare Workers KV

Workers KV provides low-latency, high-throughput global storage to your Cloudflare Workers applications. Workers KV is ideal for storing user …

S

Subham Saurav•7/26/24, 1:18 PM

oh i understand my mistake, i don't know that remote flag exist

K

kev-ac•7/26/24, 1:19 PM

You can use

--remote

--remote

or press

l

l in your wrangler terminal window to switch between modes (indicated at the bottom)

S

Subham Saurav•7/26/24, 1:20 PM

now again getting some different issue, btw my namespace name 'cred'

K

kev-ac•7/26/24, 1:21 PM

Its a safeguard to not overwrite prod data by accident. In dev mode wranlger looks for a key "preview_id" in wrangler.toml. Either copy your KV id or create a new one as described in the error message (which is the recommended way)

K

kev-ac•7/26/24, 1:23 PM

So like this:

kv_namespaces = [
    { binding = "<YOUR_BINDING>", id = "<YOUR_ID>", preview_id = "<YOUR_ID>" }
]

kv_namespaces = [
    { binding = "<YOUR_BINDING>", id = "<YOUR_ID>", preview_id = "<YOUR_ID>" }
]

kv_namespaces = [
    { binding = "<YOUR_BINDING>", id = "<YOUR_ID>", preview_id = "<YOUR_ID>" }
]

kv_namespaces = [
    { binding = "<YOUR_BINDING>", id = "<YOUR_ID>", preview_id = "<YOUR_ID>" }
]

S

Subham Saurav•7/26/24, 1:24 PM

ok got it thanks for your time, effort and help

C

caz•7/26/24, 2:16 PM

hey wanted to run something past the group to see if anyone has tried something similar or if i'm on my own here. everything in me tells me this should be possible, maybe using service workers as the app/api proxies but i haven't worked with them before... ideally this would be possible without using redirects to different domains, maybe using some form of rewrites but i don't know honestly. i'll post results here after i play around with it a bit!

Ccaz hey wanted to run something past the group to see if anyone has tried something ...

H

Hard@Work•7/26/24, 2:17 PM

For the Vite SPA, is it a static project, or does it have SSR/dynamic content?

HHard@Work For the Vite SPA, is it a static project, or does it have SSR/dynamic content?

C

caz•7/26/24, 2:17 PM

completely static, well client rendered

Ccaz completely static, well client rendered

H

Hard@Work•7/26/24, 2:18 PM

Yeah, then all of this could be a single Pages Project

H

Hard@Work•7/26/24, 2:18 PM

No need for separate Workers even

HHard@Work No need for separate Workers even

C

caz•7/26/24, 2:21 PM

good point! i would say though one thing i was considering was when someone hits "/" i would like if the proxy could do an auth check and then the spa would take over if it passes, otherwise render the honox app. that's one reason i considered breaking it up into different workers at least

C

caz•7/26/24, 2:22 PM

currently everything is already hosted in it's own workers/projects at different domains which is fine, just more or less doing a thought experiment i guess?

Ccaz good point! i would say though one thing i was considering was when someone hits...

H

Hard@Work•7/26/24, 2:22 PM

That should be possible

HHard@Work That should be possible

C

caz•7/26/24, 2:26 PM

very interesting, gonna play around with this a bit today then

D

dave•7/26/24, 4:06 PM

if (env.MY_CERT) {
  globalThis.fetch = env.MY_CERT.fetch.bind(env.MY_CERT);
}

if (env.MY_CERT) {
  globalThis.fetch = env.MY_CERT.fetch.bind(env.MY_CERT);
}

if (env.MY_CERT) {
  globalThis.fetch = env.MY_CERT.fetch.bind(env.MY_CERT);
}

if (env.MY_CERT) {
  globalThis.fetch = env.MY_CERT.fetch.bind(env.MY_CERT);
}

Will this cause all my fetches, including those in modules, to use the mTLS binding's fetch?

J

João Castro•7/26/24, 4:16 PM

I'm trying to connect directly to a Digital Ocean managed database through workers, but it just does not work, unless i'm using Hyperdrive. I've found some people complaining about similar things while searching, but no solution. Anyone has any idea about what might be causing this?

Ddave ```typescript if (env.MY_CERT) { globalThis.fetch = env.MY_CERT.fetch.bind(env...

K

kian•7/26/24, 4:23 PM

I assume so, but also means you can't fetch proxied hostnames anymore

K

kian•7/26/24, 4:24 PM

(well, assuming the callout in the docs is still accurate)

Currently, mTLS for Workers cannot be used for requests made to a service that is a proxied zone on Cloudflare. If your Worker presents a client certificate to a service proxied by Cloudflare, Cloudflare will return a 520 error.

Kkian I assume so, but also means you can't fetch proxied hostnames anymore

D

dave•7/26/24, 5:01 PM

Is there any way I can tell if a

Response

Response

Response

Response used mTLS?

D

dave•7/26/24, 5:02 PM

the issue I'm having is I'm confused if mTLS is actually being used for these requests or not

D

dave•7/26/24, 5:02 PM

I do have instant logs

Kkian (well, assuming the callout in the docs is still accurate) > Currently, mTLS fo...

D

dave•7/26/24, 5:55 PM

oh, I think I figured out the error, it looks like prime256v1 keys are NOT supported, despite it being allowed to upload

D

dave•7/26/24, 6:00 PM

yep, that was the issue.

HHard@Work Yes, that is correct

A

alierdem•7/26/24, 8:20 PM

hello. its me again. i pushed it to live with a subdomain. but still getting the same error :/

Ddave oh, I think I figured out the error, it looks like prime256v1 keys are *NOT* sup...

D

dave•7/26/24, 8:50 PM

heh, after all of that, it looks like Google Cloud doesn't even do anything with the mTLS cert -.-

P

Picolé•7/27/24, 12:26 AM

anyone know if it is possible to get an audio file duration inside cloudflare workers?

J

Jay•7/27/24, 4:06 AM

Anyone know why Discord webhooks & CF workers don't seem to work together?
This community post was the only thing I could find and had no real answer

JJay Anyone know why Discord webhooks & CF workers don't seem to work together? [This...

J

Jay•7/27/24, 4:45 AM

My current usage is via cron triggers:
1. cron queues
2. scheduled handler picks it up
3. cron task runs

Except the request to the webhook just silently fails? CF reports it all as a success and Discord doesn't show anything

JJay My current usage is via cron triggers: 1. cron queues 2. scheduled handler picks...

J

Jay•7/27/24, 6:17 AM

Here's simple repro code:

export default {
    async scheduled(event, env, ctx) {
        ctx.waitUntil(task());
    },
};

function task() {
    fetch(
        'https://canary.discord.com/api/webhooks/:id/:token',
        {
            method: 'POST',
            headers: {
                'Content-Type': 'application/json',
            },
            body: JSON.stringify({
                content: 'test',
            }),
        }
    );
}

export default {
    async scheduled(event, env, ctx) {
        ctx.waitUntil(task());
    },
};

function task() {
    fetch(
        'https://canary.discord.com/api/webhooks/:id/:token',
        {
            method: 'POST',
            headers: {
                'Content-Type': 'application/json',
            },
            body: JSON.stringify({
                content: 'test',
            }),
        }
    );
}

export default {
    async scheduled(event, env, ctx) {
        ctx.waitUntil(task());
    },
};

function task() {
    fetch(
        'https://canary.discord.com/api/webhooks/:id/:token',
        {
            method: 'POST',
            headers: {
                'Content-Type': 'application/json',
            },
            body: JSON.stringify({
                content: 'test',
            }),
        }
    );
}

export default {
    async scheduled(event, env, ctx) {
        ctx.waitUntil(task());
    },
};

function task() {
    fetch(
        'https://canary.discord.com/api/webhooks/:id/:token',
        {
            method: 'POST',
            headers: {
                'Content-Type': 'application/json',
            },
            body: JSON.stringify({
                content: 'test',
            }),
        }
    );
}

wrangler.toml

wrangler.toml

:

name = ...
main = ...
compatibility_date = ...

[triggers]
crons = [ "* * * * *" ]

name = ...
main = ...
compatibility_date = ...

[triggers]
crons = [ "* * * * *" ]

name = ...
main = ...
compatibility_date = ...

[triggers]
crons = [ "* * * * *" ]

name = ...
main = ...
compatibility_date = ...

[triggers]
crons = [ "* * * * *" ]

J

Jay•7/27/24, 6:17 AM

Am I doing something wrong here or is this a Discord thing? Cause I've seen others say they can do this

T

tkpop777•7/27/24, 6:36 AM

Does it make sense to use workers to host a chat endpoint? Not a peer to peer chat but a chat between a human and a model (the actual model is hosted elsewhere and the worker would just make a request to open ai or whatever and relay the results back to the client).

I ask cause with a serverless endpoint like workers you’re constantly spinning up and shutting down resources with each message. Since there will be so many, does it make sense to use server-less here? Or should I I look into a persistent connection?

Also if workers isn’t the right tool for the job, is durable objects the one I should be using?