Limits & Quotas

Concurrent runs

Overview

Execution concurrency depends on the service scope, available infrastructure, and scheduling configuration. Confirm the supported concurrency and queue behavior for your environment before a release or large validation campaign.

Distinguish accepted work, waiting work, active execution, and completed results. A waiting run is not evidence that tests passed or failed, and a large queue can make a time-sensitive result stale.

Step-by-step procedure

Estimate demand

List the suites, expected duration, trigger frequency, and shared dependencies. Account for overlapping scheduled jobs, repeated CI events, and retries.

Observe the execution boundary

Use the run and execution information available to your team to compare waiting and active work. Record timestamps and check the health and ownership of any customer-managed execution infrastructure.

Reduce avoidable contention

Stagger nonurgent workloads, remove duplicate triggers, and use targeted suites when appropriate. Confirm the effect of cancellation before relying on it to free capacity.

Key concepts

Concurrency: The number of operations executing at the same time within a defined boundary.
Queue age: Time work has waited before execution; useful for deciding whether the eventual result will still be relevant.

Best practices

Plan for the capacity of the application under test as well as the test executor.
Keep release-critical work distinguishable from background experiments.
Review capacity changes with infrastructure and account owners before scaling.

Common issues

Unexpected backlog: Check duplicate event delivery, unavailable execution resources, unusually long tests, and dependencies that serialize work before assuming a quota increase is required.

Was this page helpful?