Do containers run as close to the edge as workers? I'm trying to figure out if I'm actually getting

F

FrozenFireOP•7/18/25, 12:58 AM

I managed to get my latency-bound workload running in a container, and its performance is about the same as we get in AWS Fargate. Still better than Fargate since we could conceivably get better geoavailability with less lift.

But, I'm considering now whether doing the work to compile to a WASM target might be worth the few ms of latency we shave off.

FFrozenFire Do containers run as close to the edge as workers? I'm trying to figure out if I...

E

Elliot Hesp•7/18/25, 8:58 AM

I don't think so - workers are edge, DOs themselves are not (still global, but not as "edge" as workers), and I'm not sure containers run in the same location as the DO (at the moment) either

E

Elliot Hesp•7/18/25, 8:58 AM

Thats my take from the docs

F

flof.fly•7/18/25, 9:57 AM

Once in a while, I cannot deploy container changes with Wrangler because: Authentication error [code: 10000]

✘ [ERROR] A request to the Cloudflare API

Am i the only one experiencing this?

I try to deploy again, and it works, is non-deterministic

FFrozenFire I managed to get my latency-bound workload running in a container, and its perfo...

つ

つんでれd•7/18/25, 10:21 AM

Durable Objects is only deployed to some data centers and may not be the same from Workers. You can see them at https://where.durableobjects.live/

I'm not sure about if Containers or WebAssembly is fast, but if your application is lightweight as enough to not exceed the bundle size limit, it can be worth trying for compiling into WebAssembly.

Tracking where Durable Objects are created, wherever you are in the world.

SStoicGameDev#290 Is it possible to run a node.js runtime (backend with logic) on Cloudflare?

H

Hard@Work•7/18/25, 5:03 PM

With containers, yeah. For some Node applications, you can get them to run on Workers

Y

yeahmatt•7/19/25, 2:08 AM

How do containers know when they are idle and start the sleepAfter countdown? The docs mention that sleepAfter stops the instance if requests not sent for X minutes, does a request include the full request-response cycle? For example, if sleepAfter="5m", and one request is made that takes up to 15m to respond (triggers a long-running task in the container), will the container go to sleep before it returns the response?

Yyeahmatt How do containers know when they are idle and start the sleepAfter countdown? Th...

C

Chaika•7/19/25, 2:29 AM

this is part of the open source container lib
https://github.com/cloudflare/containers/blob/185903e33472aa93a6e11aff2cefb6c62c13d67b/src/lib/container.ts#L717
It's called on receiving a request, and a few spots on container startup, but not after the request.

X

xz•7/19/25, 10:41 PM

hm

F

FrozenFireOP•7/21/25, 4:42 PM

Is WebTransport possible with containers, since persistent connections are?

P

peterp•7/21/25, 9:21 PM

During development has anyone been able to get

EXPOSE

EXPOSE

working for multiple ports?

A

Aster•7/21/25, 11:18 PM

Hey hey, I am trying to run k3 in a container, with the following:

# syntax=docker/dockerfile:1
FROM rancher/k3s:latest

# Expose the required ports for K3s
EXPOSE 6443
EXPOSE 443
EXPOSE 80

# Set the default command to run K3s server
CMD ["server", "--disable", "traefik"]

# syntax=docker/dockerfile:1
FROM rancher/k3s:latest

# Expose the required ports for K3s
EXPOSE 6443
EXPOSE 443
EXPOSE 80

# Set the default command to run K3s server
CMD ["server", "--disable", "traefik"]

# syntax=docker/dockerfile:1
FROM rancher/k3s:latest

# Expose the required ports for K3s
EXPOSE 6443
EXPOSE 443
EXPOSE 80

# Set the default command to run K3s server
CMD ["server", "--disable", "traefik"]

# syntax=docker/dockerfile:1
FROM rancher/k3s:latest

# Expose the required ports for K3s
EXPOSE 6443
EXPOSE 443
EXPOSE 80

# Set the default command to run K3s server
CMD ["server", "--disable", "traefik"]

And this is my code:

export class MyContainer extends Container {
  defaultPort = 6443;
  requiredPorts = [6443, 443, 80];
  // ...
}

app.get("/readyz", async (c) => {
  const container = getContainer(c.env.MY_CONTAINER);
  return await container.fetch(c.req.raw);
});
// ...

export class MyContainer extends Container {
  defaultPort = 6443;
  requiredPorts = [6443, 443, 80];
  // ...
}

app.get("/readyz", async (c) => {
  const container = getContainer(c.env.MY_CONTAINER);
  return await container.fetch(c.req.raw);
});
// ...

export class MyContainer extends Container {
  defaultPort = 6443;
  requiredPorts = [6443, 443, 80];
  // ...
}

app.get("/readyz", async (c) => {
  const container = getContainer(c.env.MY_CONTAINER);
  return await container.fetch(c.req.raw);
});
// ...

export class MyContainer extends Container {
  defaultPort = 6443;
  requiredPorts = [6443, 443, 80];
  // ...
}

app.get("/readyz", async (c) => {
  const container = getContainer(c.env.MY_CONTAINER);
  return await container.fetch(c.req.raw);
});
// ...

I get the following:

⎔ Container image(s) ready
⎔ Starting local server...
✘ [ERROR] Error checking if container is ready: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.
✘ [ERROR] Error checking if container is ready: internal error; reference = r48029i07cer86vit2h90t6i
✘ [ERROR] Error checking if container is ready: internal error; reference = df2ceaqdffenqqhpp6ka9igg
✘ [ERROR] Error checking if container is ready: internal error; reference = gist59tegar22c3tp1iaj7ac
✘ [ERROR] Error checking if container is ready: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.
Container error: [Error: Container exited with unexpected exit code: 1] { exitCode: 1 }
▲ [WARNING] workerd/io/actor-sqlite.c++:647: warning: NOSENTRY SQLite alarm handler canceled with requestScheduledAlarm.; scheduledTime = 1753139731.815s; localAlarmState.orDefault(kj::UNIX_EPOCH) = 1753139733.074s; actorId = b60d370ee54d17e273624ceb809e666d47bc8cb445b274c648f7e06520b5a613
[wrangler:info] GET /readyz 500 Internal Server Error

⎔ Container image(s) ready
⎔ Starting local server...
✘ [ERROR] Error checking if container is ready: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.
✘ [ERROR] Error checking if container is ready: internal error; reference = r48029i07cer86vit2h90t6i
✘ [ERROR] Error checking if container is ready: internal error; reference = df2ceaqdffenqqhpp6ka9igg
✘ [ERROR] Error checking if container is ready: internal error; reference = gist59tegar22c3tp1iaj7ac
✘ [ERROR] Error checking if container is ready: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.
Container error: [Error: Container exited with unexpected exit code: 1] { exitCode: 1 }
▲ [WARNING] workerd/io/actor-sqlite.c++:647: warning: NOSENTRY SQLite alarm handler canceled with requestScheduledAlarm.; scheduledTime = 1753139731.815s; localAlarmState.orDefault(kj::UNIX_EPOCH) = 1753139733.074s; actorId = b60d370ee54d17e273624ceb809e666d47bc8cb445b274c648f7e06520b5a613
[wrangler:info] GET /readyz 500 Internal Server Error

⎔ Container image(s) ready
⎔ Starting local server...
✘ [ERROR] Error checking if container is ready: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.
✘ [ERROR] Error checking if container is ready: internal error; reference = r48029i07cer86vit2h90t6i
✘ [ERROR] Error checking if container is ready: internal error; reference = df2ceaqdffenqqhpp6ka9igg
✘ [ERROR] Error checking if container is ready: internal error; reference = gist59tegar22c3tp1iaj7ac
✘ [ERROR] Error checking if container is ready: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.
Container error: [Error: Container exited with unexpected exit code: 1] { exitCode: 1 }
▲ [WARNING] workerd/io/actor-sqlite.c++:647: warning: NOSENTRY SQLite alarm handler canceled with requestScheduledAlarm.; scheduledTime = 1753139731.815s; localAlarmState.orDefault(kj::UNIX_EPOCH) = 1753139733.074s; actorId = b60d370ee54d17e273624ceb809e666d47bc8cb445b274c648f7e06520b5a613
[wrangler:info] GET /readyz 500 Internal Server Error

⎔ Container image(s) ready
⎔ Starting local server...
✘ [ERROR] Error checking if container is ready: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.
✘ [ERROR] Error checking if container is ready: internal error; reference = r48029i07cer86vit2h90t6i
✘ [ERROR] Error checking if container is ready: internal error; reference = df2ceaqdffenqqhpp6ka9igg
✘ [ERROR] Error checking if container is ready: internal error; reference = gist59tegar22c3tp1iaj7ac
✘ [ERROR] Error checking if container is ready: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.
Container error: [Error: Container exited with unexpected exit code: 1] { exitCode: 1 }
▲ [WARNING] workerd/io/actor-sqlite.c++:647: warning: NOSENTRY SQLite alarm handler canceled with requestScheduledAlarm.; scheduledTime = 1753139731.815s; localAlarmState.orDefault(kj::UNIX_EPOCH) = 1753139733.074s; actorId = b60d370ee54d17e273624ceb809e666d47bc8cb445b274c648f7e06520b5a613
[wrangler:info] GET /readyz 500 Internal Server Error

Any idea what could be the issue?

M

Manuel Ogomigo•7/22/25, 9:21 AM

Hi is it possible to connect wrangler binding like R2 into the cotainer?

I have a ffmpeg running with rust, and I want to connect my R2 to it, so I can just add it once it's done but doesn't seem to work.

Is there any way to make it work?

MManuel Ogomigo Hi is it possible to connect wrangler binding like R2 into the cotainer? I have...

1

1984 Ford Laser•7/22/25, 9:43 AM

Maybe do a fetch to a Worker from within the container to retrieve the files from R2? Also apparently rclone can mount R2 via S3 API but haven't tried it myself

M

Manuel Ogomigo•7/22/25, 9:52 AM

Okay somehow env doesn't work there the wrangler doesn't send the env in the container

M

Manuel Ogomigo•7/22/25, 10:04 AM

2 warnings found (use docker --debug to expand):

SecretsUsedInArgOrEnv: Do not use ARG or ENV instructions for sensitive data (ENV "R2_ACCESS_KEY_ID") (line 37)
SecretsUsedInArgOrEnv: Do not use ARG or ENV instructions for sensitive data (ENV "R2_ACCESS_SECRET_KEY") (line 38)

Getting this warning when using env in Dockerfile, if there's anyway to do this correctly, I'll really appreciate

M

Maurice•7/22/25, 4:14 PM

Any idea on how to get hot reload working on a containerized deno app?

"You may prefer to set up your own code watchers and reloading mechanisms, or mount a local directory into the local container images to sync code changes. This can be done, but there is no built-in mechanism for doing so in Wrangler, and best-practices will depend on the languages and frameworks you are using in your container code."

https://developers.cloudflare.com/containers/local-dev/

Cloudflare Docs

Local Development

You can run both your container and your Worker locally, without additional configuration, by running npx wrangler dev in your project's directory.

M

Maurice•7/22/25, 4:14 PM

In a multi-container setup

U

Usman•7/22/25, 4:32 PM

Folks, does hyperdrive work with containers? I passed

this.env.HYPERDRIVE.connectionString

this.env.HYPERDRIVE.connectionString

this.env.HYPERDRIVE.connectionString

this.env.HYPERDRIVE.connectionString from my worker to container. But my Python code inside the container is not able to reach the host. Works fine if I use my origin database connection string inside Python. Also worker can connect to db with hyperdrive connection string. Only container can't. Any ideas?

could not translate host name "3d9c675c6d0235ddd9a54021ff8d1c14.hyperdrive.local" to address: Name or service not known

could not translate host name "3d9c675c6d0235ddd9a54021ff8d1c14.hyperdrive.local" to address: Name or service not known

could not translate host name "3d9c675c6d0235ddd9a54021ff8d1c14.hyperdrive.local" to address: Name or service not known

could not translate host name "3d9c675c6d0235ddd9a54021ff8d1c14.hyperdrive.local" to address: Name or service not known

UUsman Folks, does hyperdrive work with containers? I passed `this.env.HYPERDRIVE.conne...

U

Usman•7/22/25, 4:36 PM

Okay, found this. So basically there's no reliable way to connect to a native or low latency database yet, not even D1.

can we connect to the hyperdrive using

MManuel Ogomigo Hi is it possible to connect wrangler binding like R2 into the cotainer? I have...

B

Brett (Fiberplane)•7/22/25, 7:37 PM

i just built R2 support into our container last night. i have the access key id and secret key in the env of the worker, and i set the environment variables when i invoke the container itself.

this.container.start({
  // Need egress
  enableInternet: true,
  // Set env vars for R2 access in the container
  env: {
    R2_ENDPOINT_URL: `https://${this.env.CLOUDFLARE_ACCOUNT_ID}.r2.cloudflarestorage.com`,
    R2_ACCESS_KEY_ID: this.env.R2_ACCESS_KEY_ID,
    R2_SECRET_ACCESS_KEY: this.env.R2_SECRET_ACCESS_KEY,
    R2_BUCKET_NAME: this.env.BUNDLE_R2_BUCKET_NAME,
  },
});

this.container.start({
  // Need egress
  enableInternet: true,
  // Set env vars for R2 access in the container
  env: {
    R2_ENDPOINT_URL: `https://${this.env.CLOUDFLARE_ACCOUNT_ID}.r2.cloudflarestorage.com`,
    R2_ACCESS_KEY_ID: this.env.R2_ACCESS_KEY_ID,
    R2_SECRET_ACCESS_KEY: this.env.R2_SECRET_ACCESS_KEY,
    R2_BUCKET_NAME: this.env.BUNDLE_R2_BUCKET_NAME,
  },
});

this.container.start({
  // Need egress
  enableInternet: true,
  // Set env vars for R2 access in the container
  env: {
    R2_ENDPOINT_URL: `https://${this.env.CLOUDFLARE_ACCOUNT_ID}.r2.cloudflarestorage.com`,
    R2_ACCESS_KEY_ID: this.env.R2_ACCESS_KEY_ID,
    R2_SECRET_ACCESS_KEY: this.env.R2_SECRET_ACCESS_KEY,
    R2_BUCKET_NAME: this.env.BUNDLE_R2_BUCKET_NAME,
  },
});

this.container.start({
  // Need egress
  enableInternet: true,
  // Set env vars for R2 access in the container
  env: {
    R2_ENDPOINT_URL: `https://${this.env.CLOUDFLARE_ACCOUNT_ID}.r2.cloudflarestorage.com`,
    R2_ACCESS_KEY_ID: this.env.R2_ACCESS_KEY_ID,
    R2_SECRET_ACCESS_KEY: this.env.R2_SECRET_ACCESS_KEY,
    R2_BUCKET_NAME: this.env.BUNDLE_R2_BUCKET_NAME,
  },
});

no need to use ENV commands in the Dockerfile!

B

Brett (Fiberplane)•7/22/25, 7:40 PM

******

btw for the containers team: i have a small feature request for the dashboard!

if there is any way to provide metrics on when containers are killed for exceeding memory, that'd be awesome to surface in the dashboard. right now you have to cross-reference the P99 of memory usage with the instance sizee

BBrett (Fiberplane)i just built R2 support into our container last night. i have the access key id ...

M

Manuel Ogomigo•7/22/25, 7:53 PM

Thank you for this, this worked, how do you send the reqeusts, you use it's api right?

B

Brett (Fiberplane)•7/22/25, 7:56 PM

yeah so i'm using a bun/hono api and just load in the aws s3 sdk to upload the file

also btw, where did you find the r2 access key id? i resorted to looking up the id of my api key in my browser inspector network tab from inside the dashboard

M

Manuel Ogomigo•7/22/25, 8:00 PM

I got it from the API.

I'm using rust to run ffmpeg, I think that's the problem, it's not working proeprly. I'll check again.
Thank you for the tip!

R

Ricardo Gonçalves•7/22/25, 8:02 PM

Just for confirmation, is it possible to run docker-in-docker with Cloudflare Worker Containers? (I guess the answer is "no")

Bbrainstorm Do you have your code publicly available? I'd love to take a peek at how you str...

B

Brett (Fiberplane)•7/23/25, 10:34 PM

So the code is on a private repo but we just finally got around to refactoring to use the new Container class so we’re setting the env vars on the class itself per the docs — so it looks slightly different than what I posted (though that was working)

B

Brett (Fiberplane)•7/23/25, 10:34 PM

Pretty sure cf bindings in the container are an inevitability though

R

ralphjsmit•7/24/25, 11:01 AM

Hey guys, I am getting the following error all the time when making requests from local:

There is no Container instance available at this time.  This is likely because you have reached your max concurrent instance count (set in wrangler config) or are you currently provisioning the Container. If you are deploying your Container for the first time, check your dashboard to see provisioning status, this may take a few minutes.

There is no Container instance available at this time.  This is likely because you have reached your max concurrent instance count (set in wrangler config) or are you currently provisioning the Container. If you are deploying your Container for the first time, check your dashboard to see provisioning status, this may take a few minutes.

There is no Container instance available at this time.  This is likely because you have reached your max concurrent instance count (set in wrangler config) or are you currently provisioning the Container. If you are deploying your Container for the first time, check your dashboard to see provisioning status, this may take a few minutes.

There is no Container instance available at this time.  This is likely because you have reached your max concurrent instance count (set in wrangler config) or are you currently provisioning the Container. If you are deploying your Container for the first time, check your dashboard to see provisioning status, this may take a few minutes.

However, live usage appears pretty normal with just 4 live instances and all way below the concurrency limits.

L

laurynas•7/24/25, 3:35 PM

Heya, ran into this error when deploying a new worker (with containers) from CI:

 {"error":"SURPASSED_TOTAL_LIMITS","details":{"reason":"This application requires more than the allocated quota of disk in this account to run its requested instances of count 5.","used_disk_mb":90000,"requested_disk_mb":20000,"remaining_disk_mb":10000}}

 {"error":"SURPASSED_TOTAL_LIMITS","details":{"reason":"This application requires more than the allocated quota of disk in this account to run its requested instances of count 5.","used_disk_mb":90000,"requested_disk_mb":20000,"remaining_disk_mb":10000}}

 {"error":"SURPASSED_TOTAL_LIMITS","details":{"reason":"This application requires more than the allocated quota of disk in this account to run its requested instances of count 5.","used_disk_mb":90000,"requested_disk_mb":20000,"remaining_disk_mb":10000}}

 {"error":"SURPASSED_TOTAL_LIMITS","details":{"reason":"This application requires more than the allocated quota of disk in this account to run its requested instances of count 5.","used_disk_mb":90000,"requested_disk_mb":20000,"remaining_disk_mb":10000}}

I am not sure where these limits come from as the numbers are different than what's on https://developers.cloudflare.com/containers/platform-details/ ? What's the best course of action here - our deploys are blocked.

Llaurynas Heya, ran into this error when deploying a new worker (with containers) from CI:...

C

Chaika•7/24/25, 3:41 PM

Are they different? It says the max GB Disk for all instances combined is 100 GB. That error message is saying you're using 90 GB, asking for 20 GB more for this app but you only have 10 GB available. If you could use less/less max instances, should be able to get around it, otherwise there's the feedback link for higher limits

L

laurynas•7/24/25, 3:45 PM

the thing that's confusing is that we don't currently have live running instances. There are containers deployed but they're not live and running so until I hit this error I don't have a good intuition of how much my things are really taking up where is my limit etc.

C

Chaika•7/24/25, 3:49 PM

hmm yea, seems like it's just based on peak possible, (disk size for instance type * max instances)

S

spike•7/24/25, 8:45 PM

is it possible to set wranglers to use arm64 for local dev

Sspike is it possible to set wranglers to use arm64 for local dev

H

Hard@Work•7/24/25, 8:45 PM

It should by default?

HHard@Work It should by default?

S

spike•7/24/25, 8:46 PM

interesting, okay gotta dig deeper

F

FrozenFireOP•7/24/25, 10:48 PM

How do I ensure that a request is routed to the lowest latency zone, even if it means spinning up a new container? My application is extremely latency sensitive.

I set the max instances to 10, and then hit the workers.dev endpoint for my container deployment.

I had someone from the UK hit that same endpoint, and it didn't spin up a nearer instance.

FFrozenFire How do I ensure that a request is routed to the lowest latency zone, even if it ...

C

Chaika•7/24/25, 10:53 PM

Right now the routing just isn't very latency/geographic aware (Edit: well, your Durable Objects it spins up with random IDs for the getRandom/load balancing don't move with sort of pre-determine your location pools and mess with overall routing). Something they're working on: https://developers.cloudflare.com/containers/scaling-and-routing/#autoscaling-and-routing-unreleased
As for right now, "load balancing" is done via getRandom/just grabbing a random container id, which will create an object near the requester and container hopefully close as well. They never move though and sometimes the placement of the container instance themselves isn't great

F

FrozenFireOP•7/24/25, 10:56 PM

Huh. That's rough. Are there any strategies for dealing with this? I'm primarily interested in CloudFlare because I need low-latency placement.

F

FrozenFireOP•7/24/25, 11:01 PM

https://developers.cloudflare.com/containers/scaling-and-routing/#scaling-container-instances-with-get

I wonder if using env.container.get(<region>) would be effective-ish.

Cloudflare Docs

Scaling and Routing

Currently, Containers are only scaled manually by calling BINDING.get() with a unique ID, then
starting the container. Unless manualStart is set to true on the Container class, each
instance will start when get() is called.

F

FrozenFireOP•7/24/25, 11:40 PM

This sort of seems to work

export default {
  async fetch(request, env) {
    let container_id = request.cf?.colo || request.cf?.region || "default";
    return getContainer(env.MY_CONTAINER, container_id).fetch(request);
  },
};

export default {
  async fetch(request, env) {
    let container_id = request.cf?.colo || request.cf?.region || "default";
    return getContainer(env.MY_CONTAINER, container_id).fetch(request);
  },
};

export default {
  async fetch(request, env) {
    let container_id = request.cf?.colo || request.cf?.region || "default";
    return getContainer(env.MY_CONTAINER, container_id).fetch(request);
  },
};

export default {
  async fetch(request, env) {
    let container_id = request.cf?.colo || request.cf?.region || "default";
    return getContainer(env.MY_CONTAINER, container_id).fetch(request);
  },
};

S

spike•7/25/25, 6:34 AM

so

onStart()

onStart()

and the other lifecycle hooks only fire when you use

startAndWaitForPorts()

startAndWaitForPorts()

and onStart fires when the container is started or the container and the port is ready?

Do containers run as close to the edge as workers? I'm trying to figure out if I'm actually getting

Similar Threads

Similar Threads

Similar Threads