RareCloud limits and quota requests: check them and ask for more

In one paragraph: every account has limits on the cloud resources it can run: vCPU, memory, servers, volumes, disk, public IPs and more. You can see your limits and live usage at any time, check whether a cluster fits before you order it, and ask for a higher limit. Every request opens a support ticket and a person from our team decides, and the new limit applies once it is approved.

See your limits

In the console: open Cloud, then Limits.

In the CLI:

rarecloud limits

Through the API: GET /v1/limits (scope services:read). Each entry in resources has resource, used, limit, state and ratio:

A limit of -1 means unlimited. nearingLimit is true when any resource is at warn or danger.

The same answer carries registry (the Container Registry's storage and repository figures) and sandboxes (running AI Sandboxes against your limit). Each is null when it does not apply or could not be read just now.

The MCP server reads this with get_quota_limits, and Terraform with the rarecloud_limits data source.

Will a Kubernetes cluster fit?

Ask before you order. Nothing is created or reserved.

rarecloud limits estimate --pool type=g-2vcpu-8gb,min=2,max=4

Through the API, it is POST /v1/limits/estimate (scope services:read) with up to 5 pools:

curl -s https://api.rarecloud.io/v1/limits/estimate \
  -H "Authorization: Bearer $RARECLOUD_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"pools":[{"machineType":"g-2vcpu-8gb","minimum":2,"maximum":4}]}'

minimum and maximum are 1 to 16. The answer tells you:

• fits: whether the pools' minimums fit in what is left. If they do not, the cluster cannot start.
• shortfall: what is missing, keyed the same way as a limit request, so you can send it as is.
• For each pool, reachable out of configured nodes, and the resource that is limiting it. A pool whose minimum fits but whose maximum does not still deploys and autoscales, up to reachable.
• available: false means we could not work it out. That is unknown, not "does not fit".

The MCP server has this as estimate_limits.

Ask for a higher limit

In the console: open Cloud, then Limits, and use Request limit increase.

In the CLI:

rarecloud limits request --set cores=64 --set ram=262144 \
  --reason "Scaling the production cluster for a launch" --yes
rarecloud limits requests

Through the API: POST /v1/limits/requests (scope services:write):

curl -s https://api.rarecloud.io/v1/limits/requests \
  -H "Authorization: Bearer $RARECLOUD_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"requested":{"cores":64},"reason":"Scaling the production cluster for a launch"}'

The MCP server has this as request_quota_increase, which needs your confirmation.

reason is 10 to 2000 characters. Tell us what you are building: it is what the person reviewing the request reads.

What happens next

1. The request is recorded as pending and opens a support ticket. The request carries the ticket number, and you can follow it under Support.
2. A person from our team reviews it. The decision arrives as a reply on the ticket and by email.
3. If it is approved, the new limit applies and the request shows approved, with what was granted in granted. Otherwise it shows denied.

You can have one pending request at a time. Sending a new one replaces it and posts the new numbers on the same ticket.

GET /v1/limits/requests (scope services:read), rarecloud limits requests and the MCP tool list_quota_requests list your requests. Requests from before every request went to review can also show auto_approved.

Troubleshooting

Links

• API getting started
• Cloud VM docs
• CLI docs, Terraform docs and MCP docs