If it's for users try offload to their personal computers with something like a companion app. If yo

If it's for users try offload to their personal computers with something like a companion app. If you have to do it remotely checkout the price of vm's. You can get them for betwen 100x to 1000x cheaper than containers depending on your workloads.

Ggnarley_farley.If it's for users try offload to their personal computers with something like a ...

Samandar Turdiev•8/9/25, 7:51 AM

Companion app?

gnarley_farley.OP•8/9/25, 7:52 AM

Things like ffmpeg or yt-dlp can be run off a users computer. You can make users install a binary or application on their personal computers that can interface with your service in some way. So they can make use of your software but using their personal resources

Ggnarley_farley.If it's for users try offload to their personal computers with something like a ...

Samandar Turdiev•8/9/25, 7:52 AM

I'll try to check vm options

gnarley_farley.OP•8/9/25, 7:53 AM

Check racknerd and try find their specials or some other service. Yearly costs are pretty low. You can upload to s3 or r2 and then make downloadable links from there.

Ggnarley_farley.Check racknerd and try find their specials or some other service. Yearly costs a...

1984 Ford Laser•8/9/25, 8:12 AM

Whats latency like if they don't have a nearby DC?

gnarley_farley.OP•8/9/25, 8:12 AM

Better than 13 second cold starts on containers. Forsure.

1984 Ford Laser•8/9/25, 8:13 AM

What does it cost to get more bandwidth?

gnarley_farley.OP•8/9/25, 8:13 AM

A lot

gnarley_farley.OP•8/9/25, 8:14 AM

Depends though on your use case. If you processing video for users

gnarley_farley.OP•8/9/25, 8:14 AM

This is way too expensive

1984 Ford Laser•8/9/25, 8:14 AM

no with racknerd I know how to read CF docs

gnarley_farley.OP•8/9/25, 8:14 AM

They have a lot of unlimited plans

gnarley_farley.OP•8/9/25, 8:15 AM

For like 200$+ a year for a pretty sick vm and unlimited download and upload

gnarley_farley.OP•8/9/25, 8:17 AM

Example 7tb per month for 28$ per year.

gnarley_farley.OP•8/9/25, 8:18 AM

1 tb for cf containers cost more than this yearly cost.

1984 Ford Laser•8/9/25, 8:18 AM

CF Containers isnt intended to be a 24/7/365 product

gnarley_farley.OP•8/9/25, 8:19 AM

where you get 84 terrabytes of egress.. yeah we not looking for 24/7 products.

1984 Ford Laser•8/9/25, 8:19 AM

the egress is also cheaper than what aws charges for ec2 at least in my region

gnarley_farley.OP•8/9/25, 8:31 AM

Let's compare to their own services instead. I would happily pay them 1 bar a month at the right price. The current fee's just make it something we won't ever use. End of story

Code Cowboy•8/9/25, 2:30 PM

Error checking 8787: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.

Error checking 8787: connect(): Connection refused: container port not found. Make sure you exposed the port in your container definition.

Why is it checking 8787? Also anybody found a way to fix this?

EEelco Downgrading to "@cloudflare/containers": "0.0.19", from 0.0.22 fixed it.

Code Cowboy•8/9/25, 2:40 PM

Wow @Eelco you are a legend, downgrading to 0.0.19 fixed local dev for me, super frustrating

spike•8/9/25, 5:50 PM

✘ [ERROR] Error checking if container is ready: Network connection lost.
✘ [ERROR] Error checking if container is ready: Network connection lost.
✘ [ERROR] Uncaught Error: Application called abort() to reset Durable Object. 
✘ [ERROR] Uncaught Error: Application called abort() to reset Durable Object.

✘ [ERROR] Error checking if container is ready: Network connection lost.
✘ [ERROR] Error checking if container is ready: Network connection lost.
✘ [ERROR] Uncaught Error: Application called abort() to reset Durable Object. 
✘ [ERROR] Uncaught Error: Application called abort() to reset Durable Object.

anyone else seeing this intermittent abort error when starting with startandwaitforports.

pandalambda2•8/10/25, 3:43 AM

do containers work with cloudflare tail log workers? trying to understand how I can push logs to axiom logging service without changing app code, sticking logpush product behind an enterprise account seems an odd decision given infrastructure manages the logs is a core part of the 12factor methodology.

Sscotto How can I check that if it's automated as far as I know

scotto•8/10/25, 6:34 PM

Any help @kian ?

berkant•8/10/25, 9:32 PM

Hi, do Cache Rules or stuff like {cf: {cacheEverything: true}}{cf: {cacheEverything: true}} apply to containers if called fetch on them?

nickGPT•8/11/25, 1:44 AM

Does anyone know, a way to minimize the container cold start time? Trying to get it to be 2-3 seconds, but is right now 13 seconds. Thought I could make it smaller by reducing the image size and using buildx to create a optimized build for the linux/amd64 container runtime, but no luck

Code Cowboy•8/11/25, 5:24 AM

Is there a way to increase 2GB image size limit?

pandalambda2•8/11/25, 7:11 AM

how does outbound http requests work from a container? does the infra still smash on CF-Worker request header?

pandalambda2•8/11/25, 7:13 AM

and cf.worker.upstream_zone

Jjmickelly Hey, I just have a question around Cloudflare containers, does it support nested...

jmickelly•8/11/25, 9:34 AM

Is there an update to this?

pandalambda2•8/11/25, 12:30 PM

you would definitely expect this channel to be driving their FAQ and early feedback especially this early in the product launch, it's shame there seems to be little CF staff activity here

pandalambda2•8/11/25, 12:30 PM

each time I read the docs I end up with more questions than I started

Sscotto Any help @kian ?

kian•8/11/25, 4:13 PM

It should be created as a normal API token under your account: https://dash.cloudflare.com/profile/api-tokens

CCode Cowboy Is there a way to increase 2GB image size limit? 😬

kian•8/11/25, 4:15 PM

basic

basic

and

standard

standard

instances should have 4GB limits - nothing higher available at the moment though

NnickGPT Does anyone know, a way to minimize the container cold start time? Trying to get...

kian•8/11/25, 4:18 PM

How are you measuring this - is it from a new deployment -> opening Worker or just an existing image being slow to start instances?

Kkian `basic` and `standard` instances should have 4GB limits - nothing higher availab...

Chaika•8/11/25, 4:19 PM

That should be just disk size that is increased and not image size, at least according to the docs https://developers.cloudflare.com/containers/platform-details/#limits

kian•8/11/25, 4:22 PM

I'll ask the team if the docs are outdated but I did ask about images :meowthinking:

kian•8/11/25, 4:25 PM

Docs need updating - image size scales with disk size.

Ppandalambda2 how does outbound http requests work from a container? does the infra still smas...

kian•8/11/25, 5:32 PM

It'll come from the IPs on https://www.cloudflare.com/en-gb/ips/ but otherwise no

CF-Worker

CF-Worker

behaviour.

pandalambda2•8/12/25, 12:53 AM

if there any patterns/practices how one could authenticate outbound requests from a container other than say mutual tls, the IPs are shared across everyone so that doesn't appear practical, is there any plan to support cf.worker.upstream_zone? perhaps unchaining outbound worker support from workers for platform to achieve this?

Hoa Luu•8/12/25, 8:43 AM

Hello everyone,
I want to build a container with Python for Cloudflare.
I have a crontab trigger set up to call the container on a schedule. However, when I try to build the container, I encounter an error.

Alexander Lewis•8/12/25, 11:50 AM

Has anyone come across this error when creating a custom instance? If I set the values of the custom instance to match a named instance, it works fine, but say I change vcpu to 1, this will happen?

Error rolling out application project-container due to an internal error (request id: undefined): VALIDATE_INPUT

AAlexander Lewis Has anyone come across this error when creating a custom instance? If I set the ...

kian•8/12/25, 1:18 PM

Do you have the right entitlements for 1 vCPU? I don’t believe that’s available on the general paid plan.

Kkian Do you have the right entitlements for 1 vCPU? I don’t believe that’s available ...

Alexander Lewis•8/12/25, 1:30 PM

Believe the account is on enterprise which assuming that should allow more vCPU allocation?

John Spurlock•8/12/25, 8:34 PM

hmm, wondering if egress from containers -> its associated DO is billed?

pandalambda2•8/13/25, 2:35 AM

I think it must be otherwise you just proxy everything and get free unlimited/egress?

pandalambda2•8/13/25, 2:35 AM

but then that does raise the question what about communication with KV, D1, queues, R2 etc.. ?

pandalambda2•8/13/25, 4:26 AM

can one tail cloudflare container stdout/logs via wrangler do we need to do the click next dance on the web admin ui?

CCode Cowboy Wow @Eelco you are a legend, downgrading to 0.0.19 fixed local dev for me, supe...

kian•8/13/25, 9:43 PM

It looks like this was a change where the port in the URL (i.e 8787 for local Wrangler dev) started being used

kian•8/13/25, 9:44 PM

You should be able to use

.containerFetch(request, <port>)

.containerFetch(request, <port>)

.fetch(switchPort(request, <port>))

.fetch(switchPort(request, <port>))

✘ [ERROR] Error checking if container is ready: Network connection lost. ✘ [ERROR] Error checking if container is ready: Network connection lost. ✘ [ERROR] Uncaught Error: Application called abort() to reset Durable Object. ✘ [ERROR] Uncaught Error: Application called abort() to reset Durable Object.

If it's for users try offload to their personal computers with something like a companion app. If yo

Similar Threads