Quotas and limits
How many of something you may hold, and how to ask for more.
A quota is not credit. Credit is whether you can afford to keep something running; a quota is how many of something you may hold at all. They refuse differently and are fixed differently.
The limits
| Metric | Scope | Hobby | Pro |
|---|---|---|---|
microvm.plany.cloud/instances | project + region | 3 | 50 |
microvm.plany.cloud/vcpus | project + region | 8 | 128 |
microvm.plany.cloud/memory_bytes | project + region | 16 GiB | 256 GiB |
microvm.plany.cloud/schematic_storage_bytes | project + region | 100 GiB | 2 TiB |
playpute.plany.cloud/game_server_groups | project + region | 2 | 25 |
game.plany.cloud/build_storage_bytes | account | 20 GiB | 500 GiB |
sandbox.plany.cloud/sandboxes | project + region | 5 | 100 |
sandbox.plany.cloud/max_instances | project + region | 10 | 200 |
agones.plany.cloud/autopilot_fleets | project + region | 3 | 50 |
vpc.plany.cloud/vpcs | project + region | 3 | 20 |
vpc.plany.cloud/firewall_rules | project + region | 25 | 250 |
secretmanager.plany.cloud/secrets | project + region | 50 | 500 |
registry.plany.cloud/repositories | project + region | 5 | 50 |
project.plany.cloud/projects | account | 3 | 50 |
iam.plany.cloud/api_keys | account | 5 | 100 |
instances, vcpus and memory_bytes count the machines running in a project and
region — game servers and the like — by number, by the vCPUs they reserve and by
their memory. sandbox.plany.cloud/max_instances is the sum of every sandbox's
maximum instance count.
Some things the platform creates for you are not counted: a project's default network, firewall rules opened for your own workloads, and secrets Plany stores on your behalf. So a project at its limit can still start a game server.
Ask about the right thing. A project-and-region metric asked about at the account level answers zero used rather than an error. Read a project metric against the project, with the region, and an account metric against the account. The console does this for you.
Reading them
Account → Quotas| Call | Answers |
|---|---|
ListQuotaMetrics | The catalogue: what exists, its unit, its scope. |
ListQuotaLimits | Your limits and what you are using. |
GetQuotaLimit | One of them. |
ListQuotaOverrides | Increases granted to you. |
Asking for more
You cannot raise your own limit — that would not be a limit. You ask, and every quota can be asked about.
Account → Quotas → Request more (beside the limit) → the limit you want, and whyrequested_limit is the limit you want, not how many more. The CLI asks about your
account's limits; for a project's, use the console or the API.
Instances, vCPUs, memory, storage, projects, API keys, sandboxes, fleets, networks, firewall rules, secrets, repositories — settled in moments, up to a ceiling for your plan.
Autoscaling groups, and any request above the automatic ceiling.
The automatic ceiling applies to the resulting limit, not to each step — so ten small requests cannot arrive somewhere one large one could not.
Nobody is refused by a machine
A request is granted automatically or goes to a person. Nothing declines one
automatically. If the answer is no, it comes from somebody at Plany, with their
reason, shown to you as they wrote it. A grant can also be partial: granted_limit
says what you actually got.
The justification is optional — most requests are settled automatically and nobody reads it. Write one when a person will decide: it is the only thing in front of them. Leave it out or write a sentence; a few characters are refused.
What happens next
A request a person will decide is acknowledged by email, and the decision — either
way — is another. Follow it with GetQuotaIncreaseRequest or
ListQuotaIncreaseRequests, or withdraw it with WithdrawQuotaIncreaseRequest.
Who may ask: account owners and admins, billing administrators, and project owners and editors. Viewers can read requests, so anybody wondering why a limit is still 3 can see that somebody has already asked.
When you hit one
A refusal names the metric, the limit and what you asked for, so you know exactly what to request. See Errors.