Quick feedback: my containers crash when using `startAndWaitForPorts` Works fine on local but immed

Quick feedback: my containers crash when using

startAndWaitForPorts

startAndWaitForPorts

startAndWaitForPorts

startAndWaitForPorts

Works fine on local but immediately crashes on CF infra without any logs / infos

Works without (by just getting and sending requests to it) though

tao•11/7/25, 8:38 AM

We’re frequently encountering network connection lost errors in our application.
Could someone please help investigate or provide guidance on resolving this issue?

tao•11/7/25, 8:47 AM

EElia Quick feedback: my containers crash when using `startAndWaitForPorts` Works fin...

EliaOP•11/7/25, 12:36 PM

Another thing, think it'd be great if there's an easy way to extend the sleepTime from inside the container

We solved it by having the container invoke the API pinging the container (causing a reset) -- but an internal API would be far more elegant imo

Alex Dahl•11/8/25, 6:44 PM

Any news of persistent disk for containers? Would love to switch over from fly.io, but the lack of persistent disk is a blocker

AAlex Dahl Any news of persistent disk for containers? Would love to switch over from fly.i...

gruntlord6•11/9/25, 4:29 PM

yea a bit of a bummer, you can store data in R2 but then you would have to retrieve it each time before it could be used

Ggruntlord6 yea a bit of a bummer, you can store data in R2 but then you would have to retri...

Alex Dahl•11/10/25, 8:54 AM

Yeah! If containers were colocated with R2 this would have been fine, but that doesn't seem to be the case, i did some benchmarking here:
containers-beta

Justy•11/12/25, 10:54 PM

Hi everyone, just getting started on CF infrastructure and was hoping someone could help:
I have a FastAPI instance running in a container that I'd like to interact with MongoDB, but I'm struggling to find a way to route all egress through a set IP range that I can whitelist in Mongo. Is this possible yet? We are not on an enterprise plan, and likely won't ever be - I've combed the docs for a while now and haven't been able to find anything. Any help would be much appreciated!

JJusty Hi everyone, just getting started on CF infrastructure and was hoping someone co...

1984 Ford Laser•11/12/25, 10:55 PM

Cloudflare uses the same set of IPs across all applications for egress, you can whitelist the following IPs: https://www.cloudflare.com/en-au/ips/

Justy•11/12/25, 11:01 PM

Is the standard way everyone does it in production? Just feels like it wouldn't be the optimal way to secure the connection... essentially be allowing all apps hosted on Cloudflare potential access to our Atlas instance? I'm still relatively new to backend dev, so I could be completely wrong, just want to make sure we're as secure as possible.

11984 Ford Laser Cloudflare uses the same set of IPs across all applications for egress, you can ...

PhoenixSpace•11/12/25, 11:10 PM

Containers don't egress from these IPs iirc

PhoenixSpace•11/12/25, 11:10 PM

containers-beta

JJusty Is the standard way everyone does it in production? Just feels like it wouldn't ...

PhoenixSpace•11/12/25, 11:12 PM

Depending on how the application is set up, you can either use the MongoDB Atlas Admin API to whitelist the container's specific IP address on startup (if its longer lived applications where that won't be happening very frequently), or proxy it through a VPS with a fixed IP address

PhoenixSpace•11/12/25, 11:12 PM

The second is very straightforward to do with Cloudflare infrastructure

Justy•11/12/25, 11:17 PM

Okay awesome thanks, I will look into those options

PPhoenixSpace Containers don't egress from these IPs iirc

1984 Ford Laser•11/13/25, 1:03 AM

huh TIL

11984 Ford Laser Cloudflare uses the same set of IPs across all applications for egress, you can ...

Chaika•11/13/25, 1:42 AM

It's mostly just CDN/workers fetch at this point that goes through them.
TCP Connect/TCP out of Workers use WARP IPs which aren't in that list too (as well as WARP itself)

Paul•11/13/25, 2:53 AM

@cloudflare-container-team, can you confirm this issue?

Fix Cloudflare Container Instance Type Error

Problem: Used invalid instance type "standard-1" which doesn't exist in Cloudflare API.

Error Message:
Expected "containers.instance_type" field to be one of ["dev","basic","standard"] but got "standard-1".

---
Root Cause

Mismatch between documentation and reality:
- Documentation shows: "lite", "standard-1", "standard-2", "standard-3", "standard-4"
- API actually accepts: "dev", "basic", "standard" (only 3 types)

The larger instance types ("standard-1" through "standard-4") are documented but not yet implemented in the Cloudflare Containers public beta.

---
Currently Valid Instance Types

Based on the API validation error, only these 3 types exist:

| Instance Type | Resources (estimated) |
|---------------|-----------------------------------------|
| "dev" | Minimal (likely ~2 GB disk) |
| "basic" | Small (4 GB disk) |
| "standard" | Medium (8 GB disk, 4 GiB RAM, 1/2 vCPU) |

---

PPaul @cloudflare-container-team, can you confirm this issue? Fix Cloudflare Contai...

Paul•11/13/25, 2:54 AM

The website says use 'standard-1': https://developers.cloudflare.com/containers/platform-details/limits/

But wrangler doesn't accept this.

Cloudflare Docs

Limits and Instance Types

The memory, vCPU, and disk space for Containers are set through predefined instance types.
Six instance types are currently available:

Paul•11/13/25, 2:55 AM

wrangler 4.29.1 (update available 4.47.0)
─────────────────────────────────────────────
✘ [ERROR] Processing wrangler.toml configuration:
- Expected "containers.instance_type" field to be one of ["dev","basic","standard"] but got "standard-1".

Logs were written to "/home/runner/.config/.wrangler/logs/wrangler-2025-11-13_02-38-01_715.log"
Error: The process '/opt/hostedtoolcache/node/20.19.5/x64/bin/npx' failed with exit code 1
Error:

Action failed

PPaul ⛅️ wrangler 4.29.1 (update available 4.47.0) ─────────────────────────────────...

Chaika•11/13/25, 3:00 AM

Upgrade to latest wrangler (or past 4.41.0 when they were added)

Paul•11/13/25, 3:01 AM

Ok I'm using a Github Action: cloudflare/wrangler-action@v3

Let me see if I can somehow force the updgrade after the action installs it's version of wrangler...

PPaul Ok I'm using a Github Action: cloudflare/wrangler-action@v3 Let me see if I can...

Isaac McFadyen•11/13/25, 3:05 AM

You should specify that in the Action config, don't try to manually upgrade it after:
https://github.com/cloudflare/wrangler-action?tab=readme-ov-file#configuration

Jadugar_Jaggu•11/13/25, 4:59 AM

Hey facing container issues from last 6-7 hours. I just checked the dashboard and it showing this container has no instances. It was working fine from past 30 days.

Jadugar_Jaggu•11/13/25, 5:00 AM

Is there any global issue in containers?

Abdxlahs•11/13/25, 11:56 AM

Have anyone deployed n8n on containers?

AAbdxlahs Have anyone deployed n8n on containers?

Paul•11/13/25, 4:09 PM

No, but actually have had this though - let us know if you do it!

AAbdxlahs Have anyone deployed n8n on containers?

xvvvyz•11/13/25, 4:13 PM

wouldn't you need persistent storage?

Paul•11/13/25, 4:15 PM

I haven't actually tried, this so not sure about storage, but I wonder if you could execute a workflow on demand? Like just spin up the container, execute the workflow then stop it. That could just use the OS ephemeral storage, this approach assumes you load in all the context for each workflow execution. It would have slower startup times.

Shoaib019•11/13/25, 7:42 PM

Hi everyone,

I’m currently using Cloudflare Containers along with a Worker. The Worker sends a request to the container to handle video processing, since Workers themselves have limited resources.

Here’s what’s happening:

Everything works fine for smaller videos (up to 1 GB).

I’m trying to process larger videos (around 5 GB) stored in Cloudflare Stream.

Once the video reaches the “ready” state, my Worker does two things:

Generates thumbnails — this works correctly even for 5 GB videos.

Sends a request to the container to generate trailers — this is where the issue occurs.

The trailer generation starts, and I can see logs initially, but then the logs stop. After some time, the Worker throws a “CPU exceeded” error.

What’s strange is:

This issue doesn’t happen consistently — sometimes it even occurs for smaller videos.

The same video might work once, but fail if I upload it again as a new video.

So my questions are:

If Containers aren’t meant to handle heavy computation, what’s their real use case?

Why does the CPU exceed error appear in the Worker logs instead of the container logs? It seems like the Worker is waiting for the container’s response and times out, causing the CPU limit error.

If that’s the case, doesn’t it mean containers are still limited by the Worker’s execution time, making them unsuitable for long-running or compute-heavy tasks?

Any insights or guidance from the Cloudflare team or anyone who has faced a similar issue would be really helpful.

Thanks,
Shoaib

SShoaib019 Hi everyone, I’m currently using Cloudflare Containers along with a Worker. The...

Jadugar_Jaggu•11/14/25, 4:24 AM

Have you tried using container from workflows ?

Workflows can wait for container response as long as you want

JJadugar_Jaggu Have you tried using container from workflows ? Workflows can wait for contain...

Shoaib019•11/14/25, 5:02 AM

I don’t want them to wait because trailer generation takes time i just want this process to happen in background.

SShoaib019 I don’t want them to wait because trailer generation takes time i just want this...

Shoaib019•11/14/25, 11:37 AM

issue is resolved contianer have different instance types by default it uses the lite i changed to 4.
thanks guys

https://developers.cloudflare.com/containers/platform-details/limits/

Cloudflare Docs

Limits and Instance Types

The memory, vCPU, and disk space for Containers are set through predefined instance types.
Six instance types are currently available:

Yigit Erol•11/14/25, 2:15 PM

hi all. I'm trying to push my first image to the registry. It says it pushed the image but it doesn't seem to be there when I list. It also pushes instantly, which is very unlikely for a 2 gig image.. I am trying to push the image first and I'll configure the worker afterwards. Is this not supported yet? Documentation says I can push to the registry this way but it doesn't seem to be working. any help is much appreciated, thanks.

YYigit Erol hi all. I'm trying to push my first image to the registry. It says it pushed th...

Nikita Sharma•11/14/25, 4:27 PM

to list images you would have to use wrangler containers images listwrangler containers images list

NNikita Sharma to list images you would have to use `wrangler containers images list`

Yigit Erol•11/14/25, 6:02 PM

That's perfect, thanks a lot. I can see the images now. The only thing that would help significantly is to show the image upload progress. There is nothing happening in the console while uploading the image, making you think it is not working.

Eric•11/15/25, 12:42 AM

Thanks for containers y’all! This was legitimately a blast to make:

https://x.com/ericclemmons/status/1989490376198242503?s=46

Eric Clemmons, CEO of JavaScript Fatigue (@ericclemmons)

Ok, this is the coolest thing I've built YET with @Cloudflare

No-code Playwright E2E Testing + CI

30s from Prompt to Automated Test

90s E2E Test CI

It's basically an uptime checker for products

•

11/15/25, 12:27 AM

AAierie Hello! Does anyone have a recommended way to work with D1 from a container? Have...

1984 Ford Laser•11/15/25, 8:06 PM

We are still figuring things out! Best way right now is to use the REST API, bindings will be coming some time in the future

divineflu•11/15/25, 11:26 PM

couldnt find this on the pricing, is pulling the image to initilaize a container get counted in the data transfer pricing like aws ?
cause i have huge images but under the 2gb limit

amorfati•11/16/25, 12:20 AM

I have 4 sandbox container instances that were created 5-10 hours ago that I did not cleanup at the time and now I'm trying to stop them. How can I do that, or is the only way to delete the container at the moment?

Ddivineflu couldnt find this on the pricing, is pulling the image to initilaize a container...

Hello, I’m Allie!•11/16/25, 5:07 AM

01101010•11/16/25, 5:55 PM

Hello.

Cloudflare containers seem like being 95% of the way to being the best infra for game servers.

I’m currently building a game in Bevy (rust) which I package into a WASM client a binary server docker image. Each server instance just runs one instance of the game

I then deploy this server docker image to cloudflare containers with one container per game room. This is soo simple to manage! Great work.

My containers keep shutting down even if they hold open websocket connections to active clients. These clients then reconnect and cause another container to start. I’ve set the sleepAfter to “3m” as described in the websocket example on GitHub. And precisely 3m after connecting a single client and playing the game the container shuts down.

IMO cloudflare should ensure that the containers shut down if there are no active websocket connections for 3 minutes.

01101010•11/16/25, 5:56 PM

Is there a workaround I can do for now to ensure my games stay alive for as long as there are active players?

01101010•11/16/25, 7:18 PM

Also I would love to be able to manually choose the region where the container is created.

01101010•11/16/25, 7:19 PM

I’m currently running into some pretty bad lag / performance problems with cloudflare compared to railway and don’t know where they are coming from / how to fix. I think they happen because even if I’m in SF the containers sometimes spawn in Texas which adds at least 60ms of latency

01101010•11/16/25, 7:25 PM

Cloudflare above (104ms) an railway below (46ms)

From the container logs it seems that cloudflare is spawning the container in Dallas even though I’m in California. My railway instance is in us-west. .

Not sure why. Would love to be able to customise this.

Idyll•11/17/25, 10:53 PM

I'm hoping to get an update around mounting R2 into a container... I know we can call the API just like with D1 but at some point that approach will run out of road for me due to the global API limits.

Ideally binding D1/R2 works in the next 18 months?

My primary use case for containers is converting PDFs into images, one image per page, but I have seen analogous examples above manipulating other files.

I imagine things like:
- malware scanning
- making specific thumbnails for content
- pdf conversion
- etc.

Idyll•11/17/25, 10:55 PM

If there's plans for this or established patterns I'd love to find out more.

Prydar•11/17/25, 11:46 PM

so weird bug -
Even just importing a Typescript MSSQL ORM to a cloudflare container crashes my container. This is a working container with node-mssql as a proof of concept, and then it just breaks if you just import an ORM with one line. I've tried multiple workarounds. I've tried PrismaORM, Sequelize, TypeORM. This is on npx wrangler dev with local development. could be a bug just in local devl.

Quick feedback: my containers crash when using `startAndWaitForPorts` Works fine on local but immed

Similar Threads

Similar Threads

Similar Threads