Module tenant_config 

Per-tenant configuration for multi-tenant A2A servers.

PerTenantConfig allows operators to differentiate service levels across tenants by setting per-tenant timeouts, capacity limits, rate limits, and other resource constraints.

Fairness under shared process-wide caps

Per-tenant limits bound what each tenant may use; they do not reserve capacity. Process-wide resources — the event-queue manager’s max_concurrent_queues, handler sweep thresholds, and the tenant-partition cap of the tenant store wrappers — are shared pools, so a tenant running at its own limit can still exhaust a shared pool and cause other tenants’ requests to be rejected as overloaded. Set per-tenant max_concurrent_tasks so that the sum across active tenants stays within the process-wide caps if noisy-neighbor isolation matters for your deployment. Data isolation is unaffected — it is enforced per-partition regardless of these limits.

Example

use std::time::Duration;
use a2a_protocol_server::tenant_config::{PerTenantConfig, TenantLimits};

let config = PerTenantConfig::builder()
    .default_limits(TenantLimits::builder()
        .max_concurrent_tasks(100)
        .rate_limit_rps(50)
        .build())
    .with_override("premium-corp", TenantLimits::builder()
        .max_concurrent_tasks(1000)
        .executor_timeout(Duration::from_secs(120))
        .rate_limit_rps(500)
        .build())
    .build();

// "premium-corp" gets premium limits:
assert_eq!(config.get("premium-corp").max_concurrent_tasks, Some(1000));

// Unknown tenants get defaults:
assert_eq!(config.get("unknown").max_concurrent_tasks, Some(100));

Structs

PerTenantConfig

Per-tenant configuration for timeouts, capacity limits, and executor selection.

PerTenantConfigBuilder

Builder for PerTenantConfig.

TenantLimits

Resource limits and configuration for a single tenant.

TenantLimitsBuilder

Builder for TenantLimits.