are there any blogs that go over winning combinations of cloudflare workers products for different u

T

TrojanOP•12/31/24, 2:59 PM

with so many options sometimes it’s hard to choose

TTrojan are there any blogs that go over winning combinations of cloudflare workers prod...

C

Chaika•12/31/24, 3:07 PM

like this? https://developers.cloudflare.com/workers/platform/storage-options/

Cloudflare Docs

Choosing a data or storage product. · Cloudflare Workers docs

Storage and database options available on Cloudflare's developer platform.

T

TrojanOP•12/31/24, 3:07 PM

yooooo thanks!!

J

James•12/31/24, 4:03 PM

I'd go with something external like Neon for PG which also has a generous free tier, or Turso for SQLite.

E

Ephemeros•12/31/24, 4:28 PM

I'd also recommend PG, I use supabase but am considering moving to neon (haven't decided yet)

Original message was deleted

J

James•12/31/24, 4:31 PM

Direct database connections are absolutely possible with workers nowadays

J

James•12/31/24, 4:32 PM

You can use Postgres.js just fine, for example. It uses cloudflare:sockets under the hood

J

James•12/31/24, 4:58 PM

A package needs to be built for it, but there are many common ones that already use

cloudflare:sockets

cloudflare:sockets

cloudflare:sockets

cloudflare:sockets to talk to DBs today, like

postgres

postgres

E

Ephemeros•12/31/24, 5:10 PM

i use hyperdrive myself, works very well

E

Ephemeros•12/31/24, 5:10 PM

via drizzle

E

Ephemeros•12/31/24, 5:14 PM

export const getDbString = (env: Env) => {
    // Check for override first
    if (env.OVERRIDE_CONNECTION_STRING) {
        return env.OVERRIDE_CONNECTION_STRING;
    }

    if (env.HYPERDRIVE?.connectionString) {
        return env.HYPERDRIVE.connectionString;
    }

    return env.SUPABASE_DATABASE_URL;
};

export const getDbString = (env: Env) => {
    // Check for override first
    if (env.OVERRIDE_CONNECTION_STRING) {
        return env.OVERRIDE_CONNECTION_STRING;
    }

    if (env.HYPERDRIVE?.connectionString) {
        return env.HYPERDRIVE.connectionString;
    }

    return env.SUPABASE_DATABASE_URL;
};

export const getDbString = (env: Env) => {
    // Check for override first
    if (env.OVERRIDE_CONNECTION_STRING) {
        return env.OVERRIDE_CONNECTION_STRING;
    }

    if (env.HYPERDRIVE?.connectionString) {
        return env.HYPERDRIVE.connectionString;
    }

    return env.SUPABASE_DATABASE_URL;
};

export const getDbString = (env: Env) => {
    // Check for override first
    if (env.OVERRIDE_CONNECTION_STRING) {
        return env.OVERRIDE_CONNECTION_STRING;
    }

    if (env.HYPERDRIVE?.connectionString) {
        return env.HYPERDRIVE.connectionString;
    }

    return env.SUPABASE_DATABASE_URL;
};

import postgres from "postgres";

// ...

const postgresInstance = postgres(url, {
        max: 10,
        idle_timeout: 20,
        prepare: false,
        onclose: (connId) => {
            numConnections--;
            if (debug) {
                console.log(
                    "Postgres connection closed, connId:",
                    connId,
                    "numConnections:",
                    numConnections,
                    // stack,
                );
            }
        },
    });

return drizzle(postgresInstance, { schema });

import postgres from "postgres";

// ...

const postgresInstance = postgres(url, {
        max: 10,
        idle_timeout: 20,
        prepare: false,
        onclose: (connId) => {
            numConnections--;
            if (debug) {
                console.log(
                    "Postgres connection closed, connId:",
                    connId,
                    "numConnections:",
                    numConnections,
                    // stack,
                );
            }
        },
    });

return drizzle(postgresInstance, { schema });

import postgres from "postgres";

// ...

const postgresInstance = postgres(url, {
        max: 10,
        idle_timeout: 20,
        prepare: false,
        onclose: (connId) => {
            numConnections--;
            if (debug) {
                console.log(
                    "Postgres connection closed, connId:",
                    connId,
                    "numConnections:",
                    numConnections,
                    // stack,
                );
            }
        },
    });

return drizzle(postgresInstance, { schema });

import postgres from "postgres";

// ...

const postgresInstance = postgres(url, {
        max: 10,
        idle_timeout: 20,
        prepare: false,
        onclose: (connId) => {
            numConnections--;
            if (debug) {
                console.log(
                    "Postgres connection closed, connId:",
                    connId,
                    "numConnections:",
                    numConnections,
                    // stack,
                );
            }
        },
    });

return drizzle(postgresInstance, { schema });

J

James•12/31/24, 5:16 PM

I'd drop the

prepare: false

prepare: false

prepare: false

prepare: false these days - hyperdrive is well optimised for prepared statements, but yes exactly. Drizzle + postgres + hyperdrive on Workers is awesome.

E

Ephemeros•12/31/24, 5:25 PM

cool, will try updating that! i had been having some errors with prepare that's why I disabled it before

J

James•12/31/24, 5:26 PM

There were definitely issues some months ago, but that's long since been fixed afaik. I don't use

prepare: false

prepare: false

prepare: false

prepare: false in my stuff and haven't seen any issues yet.

J

James•12/31/24, 9:59 PM

Drizzle can, yes. The example above from Ephemeros is showing exactly that

J

James•12/31/24, 9:59 PM

I also run Drizzle with Postgres in many Cloudflare Workers

CChaika like this? https://developers.cloudflare.com/workers/platform/storage-options/

T

TrojanOP•1/1/25, 2:38 AM

i need more ): it didn’t go into architecture designs like i was hoping it would

JJames There were definitely issues some months ago, but that's long since been fixed a...

I

Isaac McFadyen•1/1/25, 3:21 AM

^ What James said, this was confirmed by an engineer on the Hyperdrive team a few days ago when it came up, so yes, you should be good to use

prepare: true

prepare: true

prepare: true

prepare: true or omit it because it defaults to

true

true

.

Original message was deleted

J

James•1/1/25, 1:51 PM

Neon’s serverless driver was useful before Workers could connect to DBs directly. Nowadays I’d recommend just the

postgres

postgres

library paired with Hyperdrive, and then you can use your ORM library of choice like drizzle or prisms

J

James•1/1/25, 1:51 PM

Don’t bother with the “integrations” in the dashboard

J

James•1/1/25, 3:48 PM

Yep - just

postgres

postgres

Neon pooler URL would be my recommendation then

H

Hakemo•1/1/25, 4:12 PM

is it against cloudflare worker terms of service

to have a singnalling server like https://github.com/gfodor/p2pcf and have multiple members of a community host a version on their cloudflare account

the web app will look for one of these cloudflare workers that is active and connect using it

GitHub

GitHub - gfodor/p2pcf: Low cost, low effort P2P WebRTC serverless s...

Low cost, low effort P2P WebRTC serverless signalling using Cloudflare Workers - gfodor/p2pcf

H

Hakemo•1/1/25, 4:13 PM

i am afraid it might be abusing the free tier so i had to make sure

J

James•1/1/25, 4:37 PM

I would forego using

@neondatabase/serverless

@neondatabase/serverless

@neondatabase/serverless

@neondatabase/serverless and just use

postgres

postgres

if you can - it doesn't provide much benefits nowadays.

J

James•1/1/25, 4:37 PM

But generally no, you will not be able to define and connect to the DB outside of your fetch function

J

James•1/1/25, 4:38 PM

A common approach to this is something like a

useDB

useDB

useDB

useDB function that you pass in your

env

env

, and it returns a DB client, and then you can also re-use them across different endpoints, etc. if you're running

hono

hono

hono

hono or some other router where your logic is split into multiple files

J

James•1/1/25, 4:38 PM

Another approach would be to throw it into AsyncLocalStorage, but that's a litle more complex: https://developers.cloudflare.com/workers/runtime-apis/nodejs/asynclocalstorage/

S

Shados•1/2/25, 4:45 AM

Does anyone have any insight on the perf characteristics of workers when you have a larger (a few megabyte) codebase? Since its serverless, how does it get loaded? Do the workers get "prewarmed" and ready to handle their request, or is it just in time (so if there's a lot of javascript to load to serve a single request, your workers are gonna be slow?). Its not obvious to me exactly how they work.

K

kian•1/2/25, 5:28 AM

Prewarmed during TLS handshake, but otherwise loaded upon request and hang around for awhile for other requests till evicted (i.e resources needed for other Workers or runtime update)

https://blog.cloudflare.com/eliminating-cold-starts-with-cloudflare-workers/

The Cloudflare Blog

Eliminating cold starts with Cloudflare Workers

A “cold start” is the time it takes to load and execute a new copy of a serverless function for the first time. It’s a problem that’s both complicated to solve and costly to fix.

EErisa Click to see attachment

Q

qwertymaniac•1/2/25, 3:08 PM

thanks. I'll check

CCobra I'm running into a problem similar to the one Bradley described (https://discord...

B

Bradley•1/2/25, 3:33 PM

Curious if you were able to figure this out? Still facing the general issue of not being able to set the old headless mode.

P

P•1/2/25, 4:26 PM

Sending of emails via MailChannels from my worker had stopped to work. (I have correct settings of TXT records). Did anything change in the integration https://blog.cloudflare.com/sending-email-from-workers-with-mailchannels ? I get the 401 Authorization Required response when I want to send an email.

The Cloudflare Blog

Send email using Workers with MailChannels

MailChannels has created an email sending service specifically for Cloudflare Workers that removes all the friction associated with sending emails.

PP Sending of emails via MailChannels from my worker had stopped to work. (I have c...

J

James•1/2/25, 4:27 PM

It's no longer a supported thing: https://community.cloudflare.com/t/mailchannels-end-of-life-notice-migrate-now/658638?u=cherryjimbo

P

P•1/2/25, 4:27 PM

Is there any alternative to send emails?

P

P•1/2/25, 4:28 PM

Ok I will read the migration tutorial.

J

James•1/2/25, 4:45 PM

There's a few alternatives I linked in that post, yeah. Postmark, SendGrid, Resend, etc.

S

sam•1/2/25, 6:20 PM

i think this is the case, but just asking to confirm: are stack traces only de-obfuscated using the sourcemap for uncaught exceptions? (docs seem to hint at this, but im not certain)

is there a way to get printed stack traces (in a specific structured logging field, for instance) to be remapped back to original line numbers using the uploaded source maps? thanks in advance!

B

Brett Willis•1/2/25, 9:41 PM

With Tail Workers, is there a way to stream events while the primary Worker's request is still running?
E.g. the request might last many minutes (wall/clock time) so I want to log console messages (to an external service) in a streaming fashion... not just when the request finally finishes 30 minutes later.

M

me_app•1/3/25, 12:00 PM

Is it possible to use the Workers with the bolt.new terminal?

Mme_app Is it possible to use the Workers with the bolt.new terminal?

H

Hello, I’m Allie!•1/3/25, 12:20 PM

For development? No, since workerd(the Workers runtime) doesn’t have a WASM build

L

Lancelot™•1/3/25, 2:34 PM

I deployed a nextjs app on cloudflare workers. Now where can I access/edit the worker code? I want to check the env object

LLancelot™I deployed a nextjs app on cloudflare workers. Now where can I access/edit the w...

B

boop•1/3/25, 2:41 PM

https://dash.cloudflare.com/{userId}/workers/services/view/{project}/production

https://dash.cloudflare.com/{userId}/workers/services/view/{project}/production

https://dash.cloudflare.com/{userId}/workers/services/view/{project}/production

https://dash.cloudflare.com/{userId}/workers/services/view/{project}/production

Workers & Pages > select your application > in the top right there's an "Edit Code" button

Bboop `https://dash.cloudflare.com/{userId}/workers/services/view/{project}/production...

L

Lancelot™•1/3/25, 2:42 PM

I am getting this

LLancelot™I am getting this

B

boop•1/3/25, 2:44 PM

Even if you would be able to use the editor, the way your framework packages it would make it unusable anyway.

Here's an example of a framework deployment to workers

O

Ondrej•1/3/25, 3:30 PM

Quict question someone maybe will know. How many requests does it take to enable "Smart Placement". I've just checked the status with this curl and getting

        "placement_status": "INSUFFICIENT_INVOCATIONS",

        "placement_status": "INSUFFICIENT_INVOCATIONS",

        "placement_status": "INSUFFICIENT_INVOCATIONS",

        "placement_status": "INSUFFICIENT_INVOCATIONS",

Cloudflare Docs

Smart Placement · Cloudflare Workers docs

Speed up your Worker application by automatically placing your workloads in an optimal location that minimizes latency.

J

James•1/3/25, 3:49 PM

There's no specific number of requests that's public. It just requires consistent traffic from multiple locations around the world, before it can make a decision.

JJames There's no specific number of requests that's public. It just requires consisten...

O

Ondrej•1/3/25, 4:05 PM

Hmm... any idea how I can "mock it" on my hobby project?

J

James•1/3/25, 4:06 PM

No, but I wouldn't worry about it. It'll kick in if it provides a benefit, or it just won't.

JJames No, but I wouldn't worry about it. It'll kick in if it provides a benefit, or it...

O

Ondrej•1/3/25, 4:09 PM

I was just playing with the demo and the "smart placement improved" metrics hyped me to dive deeper into it. I have a managed postgress sitting at East US (North Virginia) and moving the worker closer would help!

O

Ondrej•1/3/25, 4:10 PM

FYI I'm on a vacation in Vietnam, so it is not "usual scenario"