Cloudflare Developers•5mo ago

I'll take the other side and suspect it is _not_ billed, but we'll have to wait for someone on the t

I'll take the other side and suspect it is not billed, but we'll have to wait for someone on the team to weigh in

JJohn Spurlock I'll take the other side and suspect it is _not_ billed, but we'll have to wait ...

gruntlord6•8/14/25, 1:47 PM

I would love that to be the case so I hope I am wrong, I think paying egress from cloudflare to cloudflare is insane

Hhenri there is only one instance, which has two versions, but the instante id should b...

kian•8/14/25, 2:17 PM

The team couldn't find any issues in the logs & it seemed to be healthy to them - are you still having the issue?

CD_CNTNew•8/15/25, 8:50 AM

Anyone come across containers become harder and harder to be active after several times of successful startup&stop, mostly with "no container instance" or "not listening on TCP" errors? On the dashboard, the logs show containers seem like never get requests from workers to wake up.

CCD_CNTNew Anyone come across containers become harder and harder to be active after severa...

gruntlord6•8/15/25, 6:25 PM

I haven't. I have an image processing container so that I can use sharp to batch convert images and store them in R2. I start it up frequently, upload more images, start it up again to do those etc.

what is you use case where this occurs?

gruntlord6•8/15/25, 11:52 PM

is there a way to see billable container usage?

Ggruntlord6 I haven't. I have an image processing container so that I can use sharp to batch...

CD_CNTNew•8/16/25, 3:15 AM

Just normal http backend in the containers, stateless even. Thought it's a simple usecase for testing the containers on cloudflare, but never thought what happened like this. I guess containers never got requests from workers or durable objects bound with them, sort of bugs?

gruntlord6•8/16/25, 3:16 AM

sorry I'm not really following, you have a web service running from the container directly and it never responds?

CD_CNTNew•8/16/25, 3:17 AM

actually never start up I think. Once start up successfully, they can respond quite well

gruntlord6•8/16/25, 3:18 AM

it should start as soon as you run the attached durable object

CD_CNTNew•8/16/25, 3:19 AM

yes, at the begining just deployed, they worked well but after several times start and stop, they appeared harder to wake again

gruntlord6•8/16/25, 3:20 AM

no idea, I've even had 10 instances running for the same job without any issues starting them and stopping them

CD_CNTNew•8/16/25, 3:21 AM

how about the sleepafter of your containers?

gruntlord6•8/16/25, 3:21 AM

although I always communicate with the durable objects associated with the container

gruntlord6•8/16/25, 3:21 AM

its set to 30s so I can keep them warm, I was worried the image conversion might take longer then 10s and I don't want it shutting down mid conversion

gruntlord6•8/16/25, 3:22 AM

but I also don't want them running longer then i need them

CD_CNTNew•8/16/25, 3:22 AM

I set it to 3m

CCD_CNTNew I set it to 3m

gruntlord6•8/16/25, 3:22 AM

that would mean that unless you end the task properly it says alive for 3m doing nothing

gruntlord6•8/16/25, 3:23 AM

well assuming it finishes the task in that frame, for example when I was using one instance it would keep running as long as I kept it queued with tasks

CD_CNTNew•8/16/25, 3:23 AM

just want them to run as long as I need, coz after they turned inactive, hard to wake them up

gruntlord6•8/16/25, 3:24 AM

theres a 3 second cold start, is that accounted for when you are waking them up?

gruntlord6•8/16/25, 3:24 AM

or do they just fail to start at all

CD_CNTNew•8/16/25, 3:25 AM

fail to start

gruntlord6•8/16/25, 3:25 AM

is it something in your code then?

CD_CNTNew•8/16/25, 3:28 AM

I tested the image locally, worked well, and on cloudflare they could work after just deployed, so I think the codes are ok

CD_CNTNew•8/16/25, 3:32 AM

on the worker side, I've got the messages like no container instance or no tcp port listening, on the container side, no requests arrived so that I think no chance for my codes to run

Alexander•8/17/25, 4:43 PM

Hey all! I tried to find sandbox instance limits but there is no clear answer in the docs anywhere - am I missing anything? the reference to container limits is a bit confusing. Does it mean that max sanbox instance == max container instance for the user account? (or project)?

Appreciate the reply

AAlexander Hey all! I tried to find sandbox instance limits but there is no clear answer in...

gruntlord6•8/17/25, 6:00 PM

sanbox?

gruntlord6•8/17/25, 6:00 PM

do you mean the max instances you define?

Alexander•8/17/25, 8:36 PM

yep referring to @Cloudflare/sandboxes

aren't max instances referring to max container instances? or 1 sanbox == 1 container instance used?

Khagan•8/17/25, 9:35 PM

Hello, I'm using Next.js + Django. Is it possible to deploy this on cloudflare containers?

AAlexander yep referring to @Cloudflare/sandboxes aren't max instances referring to max c...

1984 Ford Laser•8/17/25, 10:14 PM

Yeah a sandbox is a container instance

1984 Ford Laser•8/18/25, 7:49 AM

Any news on containers allowing arbitrary TCP/UDP traffic? RTMP in/out is a major use case for me

TobiLG•8/18/25, 12:55 PM

I currently see very slow container deployments, around 10min from what I can see in th Containers Dashboard... And then it takes a few minutes afterwards until I no longer see

A call to blockConcurrencyWhile() in a Durable Object waited for too long. The call was canceled and the Durable Object was reset

A call to blockConcurrencyWhile() in a Durable Object waited for too long. The call was canceled and the Durable Object was reset

errors. Is this to be expected currently? Edit: It took nearly 15min after the intial deployment until requests were served

Marco Aleixo•8/18/25, 2:22 PM

hey team - is there a limit of subrequests in containers as well?

darkpool•8/18/25, 7:18 PM

do containers support grpc? Couldnt find any mention of it in docs. If not, is support on roadmap?

MrBithles•8/18/25, 7:30 PM

Is the concept containers only for local/isolated tasks that can be executed by a worker.
As I've been struggling to get a simple container to run which can handle requests to call external sites with fetch requests or Playwright.
Issues like:

Singletons objects in worker are not scoped to code that runs in container, its obv to say but not obv in code what is scoped.

Contains can't do network requests? (

✘ [ERROR] Error: [unenv] https.request is not implemented yet!

✘ [ERROR] Error: [unenv] https.request is not implemented yet!

)

Not sure why examples/templates as for hono app vs the worker request class, I was initially treating index.ts as the entry to Docker

The way of getting a container seems odd, following the example of using getRandom always throws an error first:

✘ [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: The operation was aborted
Container exited with unexpected exit code: 1 {"exitCode":1,"name":"Error","stack":"Error: Container exited with unexpected exit code: 1"}
[wrangler:info] POST / 500 Internal Server Error (2153ms)
✘ [ERROR] Error: [unenv] https.request is not implemented yet!

✘ [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: The operation was aborted
Container exited with unexpected exit code: 1 {"exitCode":1,"name":"Error","stack":"Error: Container exited with unexpected exit code: 1"}
[wrangler:info] POST / 500 Internal Server Error (2153ms)
✘ [ERROR] Error: [unenv] https.request is not implemented yet!

TobiLG•8/18/25, 10:12 PM

Deployments really seem to be oscilliating today.... Earlier the containers were available within a minute again, but now >10min and nothing... How can this be debugged? Is this expected?

MMrBithles Is the concept containers only for local/isolated tasks that can be executed by ...

kian•8/19/25, 12:06 AM

That unenv error sounds more related to your Worker than the Container itself

TTobiLG Deployments really seem to be oscilliating today.... Earlier the containers were...

kian•8/19/25, 12:21 AM

What exactly are you measuring, the time when making a request to an inactive instance?

Kkian What exactly are you measuring, the time when making a request to an inactive in...

TobiLG•8/19/25, 5:22 AM

Recurrent curl requests to the endpoints, until the service gets available. Had everything between 1-20mins yesterday. And that is after the Dashboard says the deployments are done… Honestly, this is not really tolerable for a Container Service when people want to iterate quickly during development. Also, sometimes it seems that I receive cached responses, even though all the old instances are shown to be shut down, which is strange. I‘m aware that this is in beta, but multi-minute downtimes upon redeployments might be a showstopper.

TobiLG•8/19/25, 5:26 AM

I‘ve been part of the private beta, and I didn’t see this behavior before. Unsure if there‘s something I can do to improve this on my end, or how to debug this. The Worker logs show the timeouts to DOs, but no error messages

Nick•8/19/25, 9:50 AM

Does anyone know what happens if I redeploy the DO while a container is running? Will the container be killed?

Meathill•8/19/25, 3:37 PM

Hello, everyone. I tried to setup a Strapi Docker on worker, but now I am stuck at Failed to start container: The container is not listening in the TCP address 10.0.0.1:1337Failed to start container: The container is not listening in the TCP address 10.0.0.1:1337
Anyone can help me? Thank you all

Jjschlesser make sure you have a line like this in your Dockerfile `EXPOSE 1337` and dont fo...

Meathill•8/19/25, 4:22 PM

Yes, I think so, I have test this Image locally and it works well

Jjschlesser locally use localhost instead of 10.0.0.1 , you will have to do some environment...

Meathill•8/19/25, 4:24 PM

I tried 0.0.0.0 before, and it didn't work. So

localhost

localhost

should be better?

✘ [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: The operation was aborted Container exited with unexpected exit code: 1 {"exitCode":1,"name":"Error","stack":"Error: Container exited with unexpected exit code: 1"} [wrangler:info] POST / 500 Internal Server Error (2153ms) ✘ [ERROR] Error: [unenv] https.request is not implemented yet!

I'll take the other side and suspect it is _not_ billed, but we'll have to wait for someone on the t

Similar Threads

Similar Threads

Similar Threads