https://railway.com/project/0192182b-560f-4469-9a29-0f8e53d2777f?environmentId=99a912e2-1d3a-4ce2-a92a-23cab17dd163

The cost estimations might not be accurate at times. I would suggest calculating estimates manually.

These are the rates:

$10/GB RAM/month

$20/vCPU/month

$0.15/GB Storage/month

$0.05/GB Egress

The system is showing peak ram usage for the database which is not being used in production at all. similar projects with same template with 400 monthly users costs $10, with no users $7. But this template project did cost $65. somehow system is calculating max ram usage even there is nothing using that much RAM

Can you provide a screenshot of your project's usage tab?

why thread is closed in discord? how railway pro customer support is getting this in last days ://

this is from the project with the bugs. only RAM usage is causing this

Cost looks correct. Do you know which services in your project are using a lot much memory?
You can see the cost on a per-service basis by pressing "view cost by service".

Duplicate threads divide support efforts. Having a single support thread is the best way forward.

Either way, as mentioned in Discord, all calculations on our end are correct, and an admin is not required here, so I'll step out and leave this to the community.

it is default railway supabase template and not used. was used for testing (no users). same template (with 400-500 users) with plausible, react apps and one nodejs backend costs less than this. it was calculation bug for RAM usage anyways which one project can't use same RAM for a month. The real RAM usage without bugs look like this and admins still think this is my problem. ://

Could you show a screenshot of the cost by service?

here it is

this is from production project:

"Kong" is https://github.com/Kong/kong, correct?
And are you aware of any configuration differences between "adorable-patience" and prod?
Could you screenshot the 30d memory usage of the Kong service?

no changes, both are railway supabase template. Kong service usage from production or from test project which has $37 usage?

Test project one

I am going to pay anyway, but it really hurts the relationship between the Railway and me. I brought at least 7-10 developers to this provider because it really worked well so far. Now I have to pay for the usage caused by a potential deployment bug or calculation bug. It would be cool if they could schedule a call and at least open my project, my invoice and check together. All they can say is: "all calculations on our end are correct".

You can say: Is it stable at 5 GB RAM usage?

It's likely a bug in the Kong service—not a Railway bug.

Have you tried redeploying, or recreating the service?

I removed the deployment, but I don't think the problem is with the Kong service. It is impossible to use 5 GB of RAM when the service is not used by customers. Even if there were connection errors, 5 GB of RAM is a huge usage.

What version of Kong is the test project using? And what about prod's?

both same version, i tried with this project then setup production env.

Which version though?