I already had my instance type set to the largest yesterday. Maybe the change took a really long tim

I already had my instance type set to the largest yesterday. Maybe the change took a really long time to take effect? I probably played with it for at least 30 minutes after changing to a larger instance. Thanks for sharing will keep it in mind if it comes back.

FFrancisco Hermida I already had my instance type set to the largest yesterday. Maybe the change to...

1984 Ford Laser•6/27/25, 11:28 PM

Did you spawn a new instance after making the change? I imagine an instance of a container can't have its instance type changed once its been created the first time

11984 Ford Laser Did you spawn a new instance after making the change? I imagine an instance of a...

Francisco HermidaOP•6/27/25, 11:40 PM

I only changed the instance type on wrangler config which didn't work for a while then suddenly worked hours later when I tried again

FFrancisco Hermida I only changed the instance type on wrangler config which didn't work for a whil...

1984 Ford Laser•6/27/25, 11:41 PM

When you spawned a new instance..? Or did you connect to a previously created one

11984 Ford Laser When you spawned a new instance..? Or did you connect to a previously created on...

Francisco HermidaOP•6/27/25, 11:55 PM

I don't fully know when a new instance is spawned or how to create or delete one. I just followed the container-template (which doesn't set the instance_type) and was hoping that on new deploys it would somehow update everything for me, which probably is not the case.

Francisco HermidaOP•6/28/25, 1:19 AM

has anyone come across a container demo that is based on node and not go?

terrxo•6/28/25, 9:03 AM

Are containers supported in multi config development (wrangler dev -c package/a -c package/b) ? Currently geting container image path resolving issues when running the dev command from root repository.

The image "./Dockerfile" does not appear to be a valid path to a Dockerfile

The image "./Dockerfile" does not appear to be a valid path to a Dockerfile

parisholley•6/28/25, 3:00 PM

would be great if we could define memory/cpu at runtime. i have a few async job flows i deploy via cloudrun, examples are generating videos or zip files, and the amount of memory/disk space is relative to number of input files. atm i use up to 8 vcpu/16gb, so probably awhile before containers is usable as a product

parisholley•6/28/25, 3:02 PM

(i suppose i could setup different instance types) and make it dynamic that way

slice•6/28/25, 6:51 PM

Is it possible to deploy discord bot on cloudflare containers ?

goofysystem•6/28/25, 7:34 PM

Is there a way to access/upload to local R2 from my container or do I have to run wrangler dev --remote to work with workers + containers + r2? I'm using S3 API to interact with R2 from my container.

Sslice Is it possible to deploy discord bot on cloudflare containers ?

slice•6/28/25, 7:49 PM

so how do i create it ?

the discord bot doesn't need any web traffic to start. it must run 24*7 without any web traffic.

Sslice so how do i create it ? the discord bot doesn't need any web traffic to start. ...

Hello, I’m Allie!•6/28/25, 8:02 PM

You can set an alarm to keep it alive, but you probably want multiple containers, as you aren’t necessarily guaranteed the container will keep running forever

✅

✅BattlefieldDuck•6/28/25, 8:04 PM

Hello, is it possible to consume message from cloudflare queue to cloudflare containers? is there any examples? thanks

HHello, I’m Allie!You can set an alarm to keep it alive, but you probably want multiple containers...

slice•6/28/25, 8:04 PM

I was planning to use long polling

slice•6/28/25, 8:05 PM

does that mean that there must be a http based entry point for containers to work ?

Iiamnotapipe cron?

slice•6/28/25, 10:39 PM

It would be lot of cronjobs

slice•6/28/25, 10:40 PM

Or delay in processsing in the bot

Francisco HermidaOP•6/29/25, 4:58 AM

made a demo showing how to use containers to create a image transformation service:
https://github.com/franciscohermida/boa-stack-img-transformer

GitHub

GitHub - franciscohermida/boa-stack-img-transformer

Contribute to franciscohermida/boa-stack-img-transformer development by creating an account on GitHub.

slice•6/29/25, 12:44 PM

hello, how do I have a mongodb database connectivity here ?

so everytime a container is spun it will create new instance of database ? is there any way to store the db instance in durable objects or something ?

Sslice hello, how do I have a mongodb database connectivity here ? so everytime a con...

Torianna•6/29/25, 2:42 PM

one idea could be to look into using R2 as a local storage sync, but you might lose data if the container stops before a sync finishes

Torianna•6/29/25, 2:43 PM

but as of right now, theres no way to access the container once its stopped. you could hook into the SIGKILL signal and perhaps sync that way, but its early days right now and i cant give a solid answer on how to do this

TTorianna one idea could be to look into using R2 as a local storage sync, but you might l...

slice•6/29/25, 2:45 PM

Thanks for the suggestion.

I think it’s better to wait few days instead of relying on such solutions for production.

Torianna•6/29/25, 2:46 PM

honestly, ive been thinking on this for a bit, if i come up with something, i'll give you a shout

TTorianna honestly, ive been thinking on this for a bit, if i come up with something, i'll...

slice•6/29/25, 4:36 PM

thanks

sandeep•6/29/25, 4:36 PM

Is anyone seeing their container instances getting killed quickly? In my case I have set

sleepAfter

sleepAfter

to "10m" but it seems to be inactive in 1-2 mins always

João Castro•6/29/25, 5:28 PM

Containers have some sort of simultaneous open connection limits like Workers do?

John Spurlock•6/29/25, 7:20 PM

Hmm, when deleting a worker in the dashboard with an associated DO, which is then associated with a container application: I was surprised the container application was not deleted when the worker (and DO namespace) were deleted. This leaves an orphan app associated with a DO namespace id that no longer exists.

Shafeeq Elanattil•6/29/25, 10:28 PM

I have a doubt about cloudflare containers. I have a video processing work load which downlaod the video from r2 and do some processing and send the results to database. Is there any time restrictions for container run time? because it probably take more time depend upon the video size. I think container should call from worker and worker have a time execution limit.

面

面条•6/30/25, 2:06 AM

Use Cloudflare Worker + Container to process images, relying on Photon, supporting functions such as scaling, cropping, watermarking, and filters. https://github.com/ccbikai/cloudflare-worker-image

GitHub

GitHub - ccbikai/cloudflare-worker-image: 使用 Cloudflare Worker ...

使用 Cloudflare Worker + Container 处理图片, 依赖 Photon，支持缩放、剪裁、水印、滤镜等功能。 - ccbikai/cloudflare-worker-image

thelovekesh•6/30/25, 8:11 AM

can we connect to the hyperdrive using containers?

つ

つんでれd•6/30/25, 9:07 AM

When running Containers locally on macOS,

fetch()

fetch()

ing to the Container throws a

Container port not found

Container port not found

error even after a while the Container has

this.ctx.container!.start()

this.ctx.container!.start()

ed, which is causing

startAndWaitForPorts()

startAndWaitForPorts()

to time out.
Is it just me or are there any workaround, or is it a known bug?
I'm using the latest template from Cloudflare.

つつんでれd When running Containers locally on macOS, `fetch()`ing to the Container throws a...

Darrin Deal•6/30/25, 12:16 PM

Not just you. I have not found a fix as of yet

robert@moyai.ai•6/30/25, 1:17 PM

Started playing around with Containers to deploy a Motherduck MCP server. Locally running the container works

local

docker run -p 8000:8000 motherduck-sse-mcp-amd64
WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested
[motherduck] INFO - 🦆 MotherDuck MCP Server v0.6.2
[motherduck] INFO - Ready to execute SQL queries via DuckDB/MotherDuck
[motherduck] INFO - Starting MotherDuck MCP Server
[motherduck] INFO - Using MotherDuck token to connect to database `md:`
[motherduck] INFO - Connecting to MotherDuck in SaaS mode
[motherduck] INFO - Database client initialized in `motherduck` mode
[motherduck] INFO - 🔌 Connecting to motherduck database
[motherduck] INFO - ✅ Successfully connected to motherduck database
[motherduck] INFO - Registering handlers
[motherduck] INFO - MCP server initialized in sse mode
[motherduck] INFO - 🦆 Connect to MotherDuck MCP Server at http://0.0.0.0:8000/sse
[uvicorn]    INFO - Started server process [18]
[uvicorn]    INFO - Waiting for application startup.
[uvicorn]    INFO - Application startup complete.
[uvicorn]    INFO - Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)

docker run -p 8000:8000 motherduck-sse-mcp-amd64
WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested
[motherduck] INFO - 🦆 MotherDuck MCP Server v0.6.2
[motherduck] INFO - Ready to execute SQL queries via DuckDB/MotherDuck
[motherduck] INFO - Starting MotherDuck MCP Server
[motherduck] INFO - Using MotherDuck token to connect to database `md:`
[motherduck] INFO - Connecting to MotherDuck in SaaS mode
[motherduck] INFO - Database client initialized in `motherduck` mode
[motherduck] INFO - 🔌 Connecting to motherduck database
[motherduck] INFO - ✅ Successfully connected to motherduck database
[motherduck] INFO - Registering handlers
[motherduck] INFO - MCP server initialized in sse mode
[motherduck] INFO - 🦆 Connect to MotherDuck MCP Server at http://0.0.0.0:8000/sse
[uvicorn]    INFO - Started server process [18]
[uvicorn]    INFO - Waiting for application startup.
[uvicorn]    INFO - Application startup complete.
[uvicorn]    INFO - Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)

But when trying to deploy to a container worker I never get a successful connection with motherduck database and instead get new error in the worker that seems to be complaining about a healthcheck endpoint:

Error checking 8000: The container is not listening in the TCP address 10.0.0.1:8000

Error checking 8000: The container is not listening in the TCP address 10.0.0.1:8000

. Anybody else receive these error messages when trying to connect to a container?

KZ•6/30/25, 2:40 PM

@Mike Nomitch | Workers PM can you guys create a containers-help and a bugs channel?

Mike Nomitch•6/30/25, 7:02 PM

Hey all

!

Sorry, I've been less active over last 48 hours on here due to some travel + personal commitments.

General updates on what we're working on now:

Fixing sleepAfter bug that currently exists in 0.0.13. If you revert to @Cloudflare/containers 0.0.11, the sleepAfter bug is actually gone, and the fixes between 11 and 13 have actually been solved at the runtime-layer, so this should be fine. - We'll be cutting an 0.0.14 release today with the fix.
Continue making regional placement better
- We fixed the core bug that was launching containers in the wrong region. Sorry, this was an embarrassing one . If you redeploy your container, it should be fixed. But we've got more to do here. If you spin up a bunch of containers, you'll fall back to other regions faster than we want. So working on making this better. And then also, if you launch a container via Workers CI ("Deploy to Cloudflare" button), it's not pre-warming properly.
Look into slow cold starts
- We're seeing some people report abnormally long cold starts. A few seconds (as of now) is not unexpected, but at 5+ seconds for our hello-world or even worse 10+ seconds, something is going wrong. We think what is happening is that we're clamping down on CPU too early. The goal is to allow for higher CPU for VM + Container boot and then move it down to whatever your instance size is afer cold start is done.
Fix memory accounting for small instance sizes
- The memory we're granting to instances is currently both container-memory and VM-memory. This meant that some people were OOMing at lower than expected values. We're switching that so the memory you get is just for the container. Should allow for rails/django on
```
dev
```
```
dev
```
  instances just fine.

Mike Nomitch•6/30/25, 7:02 PM

(and I'm going to be going thru open questions/issues today - sorry about the delays here!)

Kraus•6/30/25, 8:20 PM

Trying to build with containers but keep getting the following error:

Uncaught Error: Container is not enabled for this durable object class. Have you correctly setup your wrangler.toml?

Uncaught Error: Container is not enabled for this durable object class. Have you correctly setup your wrangler.toml?

Here is my part of my toml file:

# Containers
[[containers]]
name="transcription-containers"
class_name="TranscriptionContainer"
image="./Dockerfile"

[[durable_objects.bindings]]
name="TRANSCRIPTION_CONTAINER"
class_name="TranscriptionContainer"

# Containers
[[env.development.containers]]
class_name="TranscriptionContainer"
image="./Dockerfile"
name="transcription-containers"

[[env.development.durable_objects.bindings]]
class_name="TranscriptionContainer"
name="TRANSCRIPTION_CONTAINER"

# Containers
[[containers]]
name="transcription-containers"
class_name="TranscriptionContainer"
image="./Dockerfile"

[[durable_objects.bindings]]
name="TRANSCRIPTION_CONTAINER"
class_name="TranscriptionContainer"

# Containers
[[env.development.containers]]
class_name="TranscriptionContainer"
image="./Dockerfile"
name="transcription-containers"

[[env.development.durable_objects.bindings]]
class_name="TranscriptionContainer"
name="TRANSCRIPTION_CONTAINER"

And the code implementation:

import { Container } from "@cloudflare/containers";
export class TranscriptionContainer extends Container {
  // Port the container listens on (default: 8080)
  defaultPort = 8080;
  // Time before container sleeps due to inactivity (default: 30s)
  sleepAfter = "2m";
  // Environment variables passed to the container
  envVars = {
    MESSAGE: "I was passed in via the container class!",
  };
}

import { Container } from "@cloudflare/containers";
export class TranscriptionContainer extends Container {
  // Port the container listens on (default: 8080)
  defaultPort = 8080;
  // Time before container sleeps due to inactivity (default: 30s)
  sleepAfter = "2m";
  // Environment variables passed to the container
  envVars = {
    MESSAGE: "I was passed in via the container class!",
  };
}

Let me know if you have any questions, would like to get this resolved.

Kraus•6/30/25, 8:21 PM

@Mike Nomitch | Workers PM

goofysystem•7/1/25, 3:33 AM

I've encountered an inconsistency with how wrangler dev and wrangler deploy handle Docker build contexts.
The Issue:
When I place my Dockerfile in a subdirectory (e.g., cf-worker/containers/project-a/Dockerfile) and set the image field in wrangler.jsonc to './containers/project-a/Dockerfile':

wrangler dev builds successfully
wrangler deploy fails

Root Cause:
The two commands use different build contexts:

wrangler dev uses the Dockerfile's directory as the build context
wrangler deploy uses the worker's root directory as the build context

Current Workaround:
To make both commands work, I have to:

Move the Dockerfile to the worker root
Update the image field in wrangler.jsonc to './Dockerfile'
Modify COPY statements in the Dockerfile to account for the new context

Also deployed container has 30 second cold start and causing transaction timeout errors

Yash•7/1/25, 1:51 PM

I'm trying to deploy a Cloudflare container. It works occasionally, but most of the time I encounter the following error:

{"error":"Internal server error","message":"A call to blockConcurrencyWhile() in a Durable Object waited for too long. The call was canceled and the Durable Object was reset."}
Has anyone come across this issue before or have any reference on how to fix it?

Gabriel Francisco•7/1/25, 2:55 PM

I keep getting this error, even if its working

YYash I'm trying to deploy a Cloudflare container. It works occasionally, but most of ...

Tory•7/1/25, 2:59 PM

Yes, we get this error all the time. We're trying to migrate a service over to containers but right now it won't reliably accept requests and this is the most common error we see.

docker run -p 8000:8000 motherduck-sse-mcp-amd64 WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested [motherduck] INFO - 🦆 MotherDuck MCP Server v0.6.2 [motherduck] INFO - Ready to execute SQL queries via DuckDB/MotherDuck [motherduck] INFO - Starting MotherDuck MCP Server [motherduck] INFO - Using MotherDuck token to connect to database `md:` [motherduck] INFO - Connecting to MotherDuck in SaaS mode [motherduck] INFO - Database client initialized in `motherduck` mode [motherduck] INFO - 🔌 Connecting to motherduck database [motherduck] INFO - ✅ Successfully connected to motherduck database [motherduck] INFO - Registering handlers [motherduck] INFO - MCP server initialized in sse mode [motherduck] INFO - 🦆 Connect to MotherDuck MCP Server at http://0.0.0.0:8000/sse [uvicorn] INFO - Started server process [18] [uvicorn] INFO - Waiting for application startup. [uvicorn] INFO - Application startup complete. [uvicorn] INFO - Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)

# Containers [[containers]] name="transcription-containers" class_name="TranscriptionContainer" image="./Dockerfile" [[durable_objects.bindings]] name="TRANSCRIPTION_CONTAINER" class_name="TranscriptionContainer" # Containers [[env.development.containers]] class_name="TranscriptionContainer" image="./Dockerfile" name="transcription-containers" [[env.development.durable_objects.bindings]] class_name="TranscriptionContainer" name="TRANSCRIPTION_CONTAINER"

import { Container } from "@cloudflare/containers"; export class TranscriptionContainer extends Container { // Port the container listens on (default: 8080) defaultPort = 8080; // Time before container sleeps due to inactivity (default: 30s) sleepAfter = "2m"; // Environment variables passed to the container envVars = { MESSAGE: "I was passed in via the container class!", }; }

I already had my instance type set to the largest yesterday. Maybe the change took a really long tim

Similar Threads

Similar Threads

Similar Threads