Just to make sure, Worker -> Hyperdrive connections count towards the 6 open connections limit?

JJoão Castro Just to make sure, Worker -> Hyperdrive connections count towards the 6 open con...

A

AJR•4/16/25, 2:54 PM

Yes, they do.

J

João CastroOP•4/16/25, 2:54 PM

Thanks!

R

Ramtin•4/19/25, 5:38 AM

Hi all,
Do we need to create a new SQL connection for every request with Hyperdrive and close it after fetching the data, or should we create a single connection instance and reuse it? The latter approach seems more reasonable to me but it doesn't appear to work on Workers.

P

plurk•4/20/25, 9:37 PM

We are using hyperdrive with postgres and drizzle which is being called from pages worker functions. We encountered a nasty bug which took a few days to be narrowed to one API call.

API endpoint is beyond simple, does just 1 db query:

const map = await db.query.maps.findFirst({
  where: eq(maps.geoguessrId, params.geoguessrId)
});
if (!map) {
  return json({ mapFound: false }, { status: 404 });
}

return json({
  mapFound: true
});

const map = await db.query.maps.findFirst({
  where: eq(maps.geoguessrId, params.geoguessrId)
});
if (!map) {
  return json({ mapFound: false }, { status: 404 });
}

return json({
  mapFound: true
});

const map = await db.query.maps.findFirst({
  where: eq(maps.geoguessrId, params.geoguessrId)
});
if (!map) {
  return json({ mapFound: false }, { status: 404 });
}

return json({
  mapFound: true
});

const map = await db.query.maps.findFirst({
  where: eq(maps.geoguessrId, params.geoguessrId)
});
if (!map) {
  return json({ mapFound: false }, { status: 404 });
}

return json({
  mapFound: true
});

Clients were falsely receiving

mapFound: true

mapFound: true

mapFound: true

mapFound: true on occasion. Logging in pages functions is very limited, but we we were able to reproduce the bug by barraging the API with requests until it falsely returns

mapFound: true

mapFound: true

mapFound: true

mapFound: true.

The

maps

maps

maps

maps table doesn't change often - a few modifications a day at best. Seems like a hyperdrive bug to me, but I don't have any evidence to support it except for the info provided above.

P

plurk•4/20/25, 9:42 PM

The problem does not happen locally. We added

map.geoguessrId !== params.geoguessrId

map.geoguessrId !== params.geoguessrId

map.geoguessrId !== params.geoguessrId

map.geoguessrId !== params.geoguessrId as a bandaid fix in a desperate attempt to fix it and it seems to be doing the trick. No idea what map object is being returned, logging would really help

P

plurk•4/20/25, 9:42 PM

And ye, no caching in hyperdrive conf.

N

Niveth•4/21/25, 6:31 AM

Hi Hyperdrive team,

Around 2025-04-20 21:34:41:216 UTC, we received a

PostgresError: Internal error

PostgresError: Internal error

PostgresError: Internal error

PostgresError: Internal error from Hyperdrive.

We've reviewed the Hyperdrive troubleshooting documentation (https://developers.cloudflare.com/hyperdrive/observability/troubleshooting/#hyperdrive-specific-errors) and the Cloudflare status page, and haven't found any reported outages during that time.

We're trying to understand if this error is related to our usage pattern, which we might be able to address, or if simply retrying the request is the recommended action.

We can provide our account ID and Hyperdrive ID if that would be helpful for investigation.

Cloudflare Docs

Troubleshoot and debug · Hyperdrive docs

Troubleshoot and debug errors commonly associated with connecting to a database with Hyperdrive.

NNiveth Hi Hyperdrive team, Around 2025-04-20 21:34:41:216 UTC, we received a `Postgres...

A

AJR•4/21/25, 1:12 PM

Hello!

Yes, a Hyperdrive ID would be helpful so we can double check whether there's anything specific to your usage that might be good to adjust.

In general, we do recommend application-level retries. With a network in the mix, there will always be some queries that fail and it's good to build around that where possible. We do expect to continue to improve our own retry behavior over time, but there will always be some gap because it won't always be clear to Hyperdrive what queries are safe to retry so we are very conservative on that front.

Aalex Is there any recommended way to effectively use hyperdrive caching + postgres rl...

A

AJR•4/21/25, 1:27 PM

No, we don't have good guidance around that yet. Our architecture is pretty much built around "1 Hyperdrive = 1 user", and trying to use session state to end-run that without letting it corrupt other connections/queries is tricky.

CC @thomasgauvin this is the second time we've had someone bump into this. Might be worth thinking through options here for our upcoming roadmap?

A

AJR•4/21/25, 2:12 PM

This is something we'll need to tinker with a bit to get a good design that offers security without degrading perf. We have similar challenges around how to handle bypassing caching and potentially similar challenges around replica routing. @thomasgauvin we should see about folding some of that into our design for cache flexibility. It feels like it sits in a similar spot.

M

midnight•4/21/25, 4:34 PM

I am working on standing up a CF tunnel to connect to a local postgres DB for development tier. I have

cloudflared

cloudflared

running in a docker container and the postgres bitnami contianer running but keep receiving this error when creating my hyperdrive config via Pulumi with v6.1.0:

failed to make http request: POST "https://api.cloudflare.com/client/v4/accounts/<redacted>/hyperdrive/configs": 400 Bad Request {
  "result": null,
  "success": false,
  "errors": [
    {
      "code": 2011,
      "message": "Network connection to the provided database was refused. Please check the that host and port are correct and that the database allows connections from public IP addresses
."
    }
  ],
  "messages": null
}

failed to make http request: POST "https://api.cloudflare.com/client/v4/accounts/<redacted>/hyperdrive/configs": 400 Bad Request {
  "result": null,
  "success": false,
  "errors": [
    {
      "code": 2011,
      "message": "Network connection to the provided database was refused. Please check the that host and port are correct and that the database allows connections from public IP addresses
."
    }
  ],
  "messages": null
}

failed to make http request: POST "https://api.cloudflare.com/client/v4/accounts/<redacted>/hyperdrive/configs": 400 Bad Request {
  "result": null,
  "success": false,
  "errors": [
    {
      "code": 2011,
      "message": "Network connection to the provided database was refused. Please check the that host and port are correct and that the database allows connections from public IP addresses
."
    }
  ],
  "messages": null
}

failed to make http request: POST "https://api.cloudflare.com/client/v4/accounts/<redacted>/hyperdrive/configs": 400 Bad Request {
  "result": null,
  "success": false,
  "errors": [
    {
      "code": 2011,
      "message": "Network connection to the provided database was refused. Please check the that host and port are correct and that the database allows connections from public IP addresses
."
    }
  ],
  "messages": null
}

I can login to the local machine and run the psql connection to connect to the postgres container successfully, I even ran psql using a second container to connect.

AAJR Yeah, before we added support for prepared statements you couldn't use postgres....

A

alaister•4/22/25, 8:32 AM

Hey, I made the original post on the forum and just want to add some additional details:
I think my assumption that it wasn't caching at all was incorrect. The responses just seem to be a lot slower...

For example my worker with:

await sql`select 'hello' as "message";`

await sql`select 'hello' as "message";`

await sql`select 'hello' as "message";`

await sql`select 'hello' as "message";`

usually responds in 20-50ms whereas:

await sql`select ${'hello'} as "message";`

await sql`select ${'hello'} as "message";`

await sql`select ${'hello'} as "message";`

await sql`select ${'hello'} as "message";`

usually responds in 200-300ms.

The reason I say my assumption was incorrect is the cached percentage in my hyperdrive dashboard is now showing a high percentage, so it must be something to do with how the caching is working.

I've tried with both

prepare: false

prepare: false

and

prepare: true

prepare: true

and neither seem to make a big difference to the times...

Any help here would be appreciated! My app is able to cache a large percentage of the data, so I could see some serious performance gains if I can work out what's wrong.

O

oritwoen•4/24/25, 6:08 PM

Emoji not possible on mysql hyperdrive?

Ooritwoen Emoji not possible on mysql hyperdrive?

A

AJR•4/24/25, 6:11 PM

Currently supports only UTF8 in queries. Most emojii are in UTF8, but I imagine there are plenty that aren't?

AAJR Currently supports only UTF8 in queries. Most emojii are in UTF8, but I imagine ...

O

oritwoen•4/24/25, 6:17 PM

return ??? after query

Ooritwoen 🚀 return ??? after query

A

AJR•4/24/25, 6:17 PM

Sorry, I'm not sure what your question is.

O

oritwoen•4/24/25, 6:19 PM

This text:

"Hello world

"
in normal mysl query (mariadb, mysql) return without change (with emoji),

in hyperdrive, query return:

"Hello world ???"

A

AJR•4/24/25, 6:21 PM

Hmmmm. Ok, I'll write up a ticket to investigate. Thanks for letting us know.

こ

こた•4/25/25, 5:25 AM

Hi there!
I'm trying to connect to Google Cloud SQL using Hyperdrive, but I'm getting an Internal Server Error and not sure how to fix it.
I'm connecting via the public IP, and I've allowed all of Cloudflare's IPv4 addresses.
When I use the same connection string with psql, the connection works fine, so I suspect the issue is on Cloudflare's side.
Does anyone know what I should do in this case?

(Update)
After further investigation, I found that the error occurs when

--ca-certificate-id

--ca-certificate-id

--ca-certificate-id

--ca-certificate-id and

--mtls-certificate-id

--mtls-certificate-id

--mtls-certificate-id

--mtls-certificate-id are specified.
When I create the connection without those flags, it works successfully.
I also tried re-uploading the certificates, but the error still persists.

T

thomasgauvin•5/1/25, 3:11 PM

Hey folks! We're working on docs to better document how to use Drizzle (Postgres and MySQL) and Prisma (Postgres only) with Hyperdrive here: https://github.com/cloudflare/cloudflare-docs/pull/22125

You can access the preview pages here:

Drizzle Postgres: https://thomasgauvin-hyperdrive-drivers-and-orms.preview.developers.cloudflare.com/hyperdrive/examples/connect-to-postgres/postgres-drivers-and-libraries/drizzle-orm/
Prisma Postgres: https://thomasgauvin-hyperdrive-drivers-and-orms.preview.developers.cloudflare.com/hyperdrive/examples/connect-to-postgres/postgres-drivers-and-libraries/prisma-orm/
Drizzle MySQL: https://thomasgauvin-hyperdrive-drivers-and-orms.preview.developers.cloudflare.com/hyperdrive/examples/connect-to-mysql/mysql-drivers-and-libraries/drizzle-orm/

Try them and let us know if they work or there's anything that should be changed!

TTyveris So Drizzle tends to encourage the node-postgres driver while your docs use the p...

I

Isaac McFadyen•5/1/25, 4:53 PM

From a quick look at the reasons they provide for choosing

node-postgres

node-postgres

:

With node-postgres, you can install pg-native to boost the speed of both node-postgres and Drizzle by approximately 10%.

This won't matter on Workers because Workers doesn't support native extensions.

node-postgres supports providing type parsers on a per-query basis without globally patching things. For more details, see Types Docs.

Seems valid but probably somewhat of an advanced use-case?

postgres.js uses prepared statements by default, which you may need to opt out of. This could be a potential issue in AWS environments, among others, so please keep that in mind.

Hyperdrive fully supports prepared statements now, and they improve performance so this is probably a benefit of

postgres.js

postgres.js

actually

Tthomasgauvin Hey folks! We're working on docs to better document how to use Drizzle (Postgres...

I

Isaac McFadyen•5/1/25, 4:54 PM

Nice! Only thing I can see here is that it might be a good idea to set the connection limit on the Postgres connection so it doesn't exceed 6 and deadlock the Worker (as discovered by James)

I

Isaac McFadyen•5/1/25, 4:55 PM

// Clean up the connection
ctx.waitUntil(sql.end());

Also, AJR explained why this was a good idea a while back, but might be good to add a line claryfing this in the docs because people might generally assume it's bad practice to close connections on every request.

IIsaac McFadyen From a quick look at the reasons they provide for choosing `node-postgres`: > W...

T

thomasgauvin•5/1/25, 4:57 PM

We actually have had some conversations with the maintainer of Postgres.js. Since the node compat improvements on Workers, node-postgres is just as valid of a choice, and it's actually faster when you need to fetch the types from Postgres so we may change that recommendation https://developers.cloudflare.com/hyperdrive/examples/connect-to-postgres/#postgresjs

Tthomasgauvin We actually have had some conversations with the maintainer of Postgres.js. Sinc...

I

Isaac McFadyen•5/1/25, 4:58 PM

Makes sense! I also see that doc has the

max

max

set to 5, so just porting that to the new docs would be good

T

thomasgauvin•5/1/25, 4:59 PM

I'll add a note for Postgres.js vs node-postgres to the drizzle docs, and I'll look into the configurable max connections

Tthomasgauvin Hey folks! We're working on docs to better document how to use Drizzle (Postgres...

M

marsoleg•5/1/25, 5:47 PM

This can be useful in productive environments where developers do not have direct access to the database from their local computers:

name: Deploy App

on:
  push:
    branches: main

permissions:
  contents: read

jobs:
  deploy:
    runs-on: self-hosted
    steps:
      - name: Checkout repository
        uses: actions/checkout@v4
        with:
          fetch-depth: 1

      - name: Install dependencies
        run: npm ci

      - name: Check DB migrations
        run: |
          output=$(npx drizzle-kit generate)
          echo "$output"
          if ! echo "$output" | grep -q "No schema changes, nothing to migrate"; then
            echo; echo "Error: Schema changes detected. Please generate migrations before deploying."
            exit 1
          fi

      - name: Apply DB migrations
        env:
          DATABASE_URL: ${{ secrets.DATABASE_URL }}
        run: npx drizzle-kit migrate

      - name: Deploy application to Cloudflare
        env:
          CLOUDFLARE_API_TOKEN: ${{ secrets.CLOUDFLARE_API_TOKEN }}
          CLOUDFLARE_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}
        run: npx wrangler deploy

name: Deploy App

on:
  push:
    branches: main

permissions:
  contents: read

jobs:
  deploy:
    runs-on: self-hosted
    steps:
      - name: Checkout repository
        uses: actions/checkout@v4
        with:
          fetch-depth: 1

      - name: Install dependencies
        run: npm ci

      - name: Check DB migrations
        run: |
          output=$(npx drizzle-kit generate)
          echo "$output"
          if ! echo "$output" | grep -q "No schema changes, nothing to migrate"; then
            echo; echo "Error: Schema changes detected. Please generate migrations before deploying."
            exit 1
          fi

      - name: Apply DB migrations
        env:
          DATABASE_URL: ${{ secrets.DATABASE_URL }}
        run: npx drizzle-kit migrate

      - name: Deploy application to Cloudflare
        env:
          CLOUDFLARE_API_TOKEN: ${{ secrets.CLOUDFLARE_API_TOKEN }}
          CLOUDFLARE_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}
        run: npx wrangler deploy

name: Deploy App

on:
  push:
    branches: main

permissions:
  contents: read

jobs:
  deploy:
    runs-on: self-hosted
    steps:
      - name: Checkout repository
        uses: actions/checkout@v4
        with:
          fetch-depth: 1

      - name: Install dependencies
        run: npm ci

      - name: Check DB migrations
        run: |
          output=$(npx drizzle-kit generate)
          echo "$output"
          if ! echo "$output" | grep -q "No schema changes, nothing to migrate"; then
            echo; echo "Error: Schema changes detected. Please generate migrations before deploying."
            exit 1
          fi

      - name: Apply DB migrations
        env:
          DATABASE_URL: ${{ secrets.DATABASE_URL }}
        run: npx drizzle-kit migrate

      - name: Deploy application to Cloudflare
        env:
          CLOUDFLARE_API_TOKEN: ${{ secrets.CLOUDFLARE_API_TOKEN }}
          CLOUDFLARE_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}
        run: npx wrangler deploy

name: Deploy App

on:
  push:
    branches: main

permissions:
  contents: read

jobs:
  deploy:
    runs-on: self-hosted
    steps:
      - name: Checkout repository
        uses: actions/checkout@v4
        with:
          fetch-depth: 1

      - name: Install dependencies
        run: npm ci

      - name: Check DB migrations
        run: |
          output=$(npx drizzle-kit generate)
          echo "$output"
          if ! echo "$output" | grep -q "No schema changes, nothing to migrate"; then
            echo; echo "Error: Schema changes detected. Please generate migrations before deploying."
            exit 1
          fi

      - name: Apply DB migrations
        env:
          DATABASE_URL: ${{ secrets.DATABASE_URL }}
        run: npx drizzle-kit migrate

      - name: Deploy application to Cloudflare
        env:
          CLOUDFLARE_API_TOKEN: ${{ secrets.CLOUDFLARE_API_TOKEN }}
          CLOUDFLARE_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}
        run: npx wrangler deploy

Tthomasgauvin I'll add a note for Postgres.js vs node-postgres to the drizzle docs, and I'll l...

J

James•5/1/25, 10:14 PM

+1 on the max_connections recommendation, this is a mitigation for https://github.com/cloudflare/workerd/issues/3514, at least until https://github.com/porsager/postgres/issues/1023 is resolved. Otherwise the drizzle docs look great and are essentially what we use in production today

AAJR I don't believe anyone is doing that, no. We haven't tested it, though some bri...

D

darkpool•5/2/25, 6:22 PM

Hi AJR I see that hyperdrive now supports GoogleSQL and some distributed SQL DBs. Im assuming Spanner still wont work with hyperdrive since I don't see it on the list? Im also assuming it wont work bc spanner does not provide a traditional connection string with host and port name but just wanted to confirm.

Ddarkpool Hi AJR I see that hyperdrive now supports GoogleSQL and some distributed SQL DBs...

A

AJR•5/2/25, 6:27 PM

I'm afraid that's correct, yes.

AAJR I'm afraid that's correct, yes.

D

darkpool•5/2/25, 6:35 PM

Thank you. Is there a recommended way to connect to google cloud resources in workers? Right now Im just creating and signing JWTs using a service account. I found this article but it’s a few years old so just wondering if there's a better way.

Ddarkpool Thank you. Is there a recommended way to connect to google cloud resources in wo...

A

AJR•5/2/25, 6:55 PM

Do you mean specifically GCP SQL? There's this: https://developers.cloudflare.com/hyperdrive/examples/connect-to-postgres/google-cloud-sql/

edit: I do notice that that still has you allowlisting 0.0.0.0, when now you can instead allowlist just the Cloudflare IP ranges

Cloudflare Docs

Google Cloud SQL

Connect Hyperdrive to a Google Cloud SQL for Postgres database instance.

AAJR Do you mean specifically GCP SQL? There's this: https://developers.cloudflare.c...

D

darkpool•5/2/25, 7:02 PM

I don't think that will work with Spanner since step 2 of the link is asking for a connection string like:

postgres://USERNAME:PASSWORD@HOSTNAME_OR_IP_ADDRESS:PORT/database_name

postgres://USERNAME:PASSWORD@HOSTNAME_OR_IP_ADDRESS:PORT/database_name

postgres://USERNAME:PASSWORD@HOSTNAME_OR_IP_ADDRESS:PORT/database_name

postgres://USERNAME:PASSWORD@HOSTNAME_OR_IP_ADDRESS:PORT/database_name
which spanner does not provide. I was asking more so if there's a recommended way if hyperdrive does not work.

Ddarkpool I don't think that will work with Spanner since step 2 of the link is asking for...

A

AJR•5/2/25, 7:34 PM

Ah, I see. Not that I'm aware of, I'm afraid. You might have better success asking in a more general Workers channel.

AAJR Ah, I see. Not that I'm aware of, I'm afraid. You might have better success a...

D

darkpool•5/2/25, 7:39 PM

Thank you.

W

William•5/4/25, 12:01 PM

Hi, everyone
I have a question with Hyperdrive. On this page: https://developers.cloudflare.com/hyperdrive/platform/limits/

As many hosted databases have limits on the number of unique connections they can manage, Hyperdrive attempts to keep number of concurrent pooled connections to your origin database lower.

How does Hyperdrive know the connection limit of my database? What does keep number of concurrent pooled connections to your origin database lower. means? lower than what?

Cloudflare Docs

Limits

The following limits apply to Hyperdrive configuration, connections, and queries made to your configured origin databases.

WWilliam Hi, everyone I have a question with Hyperdrive. On this page: https://developers...

A

AJR•5/4/25, 1:37 PM

That is a bit of an odd way to phrase it, I suppose.

Hyperdrive is a tx-mode connection pooler. It will agressively reuse connections, only opening a new one if a query arrives needing one and none of the currently open ones are available.

We currently have a soft limit of 60 connections to a pool, though can adjust that on request.

Hope that helps, please feel free to ask if you have any other questions!

AAJR That is a bit of an odd way to phrase it, I suppose. Hyperdrive is a tx-mode co...

W

William•5/4/25, 2:24 PM

Assume that users of my website may come from around the world, including North America, South America, Europe, Africa, and Asia.
This means there could be 5 Hyperdrive connection pools active at the same time:

Pool1: North America
Pool2: South America
Pool3: Europe
Pool4: Africa
Pool5: Asia

I'm using a Postgres database, and let's say the maximum allowed number of connections to my database is 80.

I saw in the documentation that for the Paid Plan, the "Maximum potential origin database connections" is approximately ~100 connections.

So, with these 5 connection pools combined, will the total number of connections to my database be limited to below 80, or below 100?

WWilliam Assume that users of my website may come from around the world, including North ...

A

AJR•5/4/25, 2:28 PM

We refactored our architecture, there'll be only a single pool near wherever your database is hosted, capped to 60.

We prioritize availability over enforcing this limit, so if the system is stressed your traffic may spill to a second pool, which would be 120 total.

Edit: I see the bit I'm missing.

No, Hyperdrive does not currently consider the maximum connection GUC on your origin database.

WWilliam Assume that users of my website may come from around the world, including North ...

A

AJR•5/4/25, 2:29 PM

We go into some detail on this here, if you're curious: https://blog.cloudflare.com/it-it/how-hyperdrive-speeds-up-database-access/#better-faster-smarter-closer

Il blog di Cloudflare

Pools across the sea: how Hyperdrive speeds up access to databases ...

Hyperdrive, Cloudflare's global connection pooler, relies on some key innovations to make your database connections work. We go deep into the details of what that looks like, in celebration of announcing its availability for Free Plan customers.

AAJR We refactored our architecture, there'll be only a single pool near wherever you...

W

William•5/4/25, 2:58 PM

I got it, thanks very much!

S

shay•5/4/25, 6:08 PM

i'm suddenly seeing what i believe to be increased connections from hyperdrive to my postgres server (there are many pg processes each using very little cpu), which is now pegging my server to 100% cpu usage, and that manifests in numerous "PostgresError: Timed out while waiting for an open slot in the pool." as well as CONNECTION_CLOSED errors in my logs. i don't think my traffic has increased proportionally to cause this--i've never noticed sustained CPU usage like this before and my daily requests are about what they've been for the past few weeks. what could be causing this? do i just need to scale up my server?

while writing this i'm noticing a few dips down into the 30-70% range which is more reasonable but the sustained highs are still concerning to me

O

oritwoen•5/5/25, 1:28 PM

Is this features will be available in the future?

https://developers.cloudflare.com/hyperdrive/reference/supported-databases-and-features/#unsupported-mysql-features

Ooritwoen Is this features will be available in the future? https://developers.cloudflare...

A

AJR•5/5/25, 1:29 PM

Some of them might be, depending on user needs. Which ones in particular would you be interested in?

Just to make sure, Worker -> Hyperdrive connections count towards the 6 open connections limit?

Similar Threads

Similar Threads

Similar Threads