Limits & quotas
Concrete limits enforced by the LakeSail API and platform. Numbers come from the OpenAPI spec; soft limits (anything that requires raising via support) are noted.
API request limits
| Limit | Value | Notes |
|---|---|---|
| Max page size on list endpoints | 100 | The pageSize query parameter caps at 100. Use cursor-based pagination for larger results. |
| Rate limits |
Resource naming
| Field | Min | Max |
|---|---|---|
| Resource name (most entities) | 1 | 255 chars |
| Description | — | 1023 chars |
| Identity provider name | 1 | 255 chars |
| Identity provider type display name | 1 | 100 chars |
Jobs
| Limit | Value | Notes |
|---|---|---|
| Cron expression | 100 chars | Standard 5-field cron |
| Cron timezone | 64 chars | IANA zone names |
| Min concurrent runs | 1 | When concurrencyPolicy is allow |
Job timeouts and resource limits are configured per-job; there is no platform-imposed hard ceiling beyond what your cluster can run.
Queries
| Limit | Value | Notes |
|---|---|---|
| Query text | 1 | 8191 chars (the maximum a single query can store inline). |
| Catalog schema name | — | 255 chars |
For longer SQL, split into multiple queries or reference SQL files from a Python wheel job.
Notifications
| Limit | Value | Notes |
|---|---|---|
| Channels per organization | ||
| Rules per organization |
Sessions
| Limit | Value | Notes |
|---|---|---|
| Concurrent sessions per cluster | ||
| Idle timeout before close | ||
| Token lifetime | Short — measured in minutes, not days |
Clusters and infrastructure
| Limit | Value | Notes |
|---|---|---|
| Min management nodes | 1 | Hard floor; a cluster needs at least one system node |
| Max management nodes | Soft limit; raise if you keep hitting it | |
| Networks per cloud account | ||
| Clusters per network | ||
| Provisioning time, typical | 5–10 minutes | Network or cluster, end-to-end |
| Provisioning time, worst case | 30+ minutes | If AWS quota requests need to land first |
The cluster's allocatable CPU and pod-per-node count are constrained by the management node instance type. See Set up a cluster for the validation warnings that fire when sizing is insufficient.
AWS-side quotas
LakeSail provisions resources in your AWS account, so your account's service quotas apply:
- EC2 instance limits per region — most relevant for cluster compute scaling.
- VPCs per region — five by default.
- CloudFormation stacks per region — typically not a constraint.
- Elastic IPs, NAT gateways, load balancers — used per cluster.
If a network or cluster fails to provision with a quota error in CloudFormation, raise the relevant quota in your AWS account before retrying.
Auth and identity
| Limit | Value | Notes |
|---|---|---|
| MFA recovery codes | ||
| Session cookie / token lifetime | ||
| OIDC scopes per identity provider | — | Limited only by what the IdP supports |
Raising a limit
For soft limits, contact . Include:
- Your organization ID.
- Which limit you're hitting (with an exact resource ID or operation).
- Why — a brief description of the workload that needs the higher limit.
Hard limits (the ones noted "Hard floor" / "Hard ceiling") can't be raised; redesign around them.
API reference
- Authentication —
ListPermissionsreturns the authoritative permission catalog. - The full schema (including every field's min/max constraint) is at the Full Spec.