What is a Flow Run?
Every time a Pipefy integration flow is triggered by an event, the system automatically creates a unique Run ID (acting as a digital receipt or Job ID for that specific execution).
A Run represents the journey of an entire flow from the initial trigger to the final action. Pipefy handles these runs via asynchronous processing. This means the system acknowledges the trigger instantly, allowing your users to move on with their day while our background workers handle the heavy lifting behind the scenes without interrupting your live operations.
Tracking the Queue and Concurrency Limits
To keep the integration engine stable, predictable, and fair for all users, our platform operates under clear governance and capacity rules:
The Concurrency Limit (50 Simultaneous Runs): Your organization can run up to 50 individual flow runs at the exact same millisecond. If you send up to 50 flows runs simultaneously, they will find open processing slots and execute immediately.
The "Queued" Status: If an integration workflow bursts and triggers more than 50 runs at once (for example, during a massive data import or an unexpected spike), the extra runs will temporarily enter a Queued status.
First-Come, First-Served: Queued runs are held safely in a centralized waiting line and will automatically execute in a strict order of arrival the moment your active concurrent slots drop below 50.
How to Read the Queue Status Widget
On the top right corner of the Runs dashboard, you will find the new interactive Queue Status widget. This component provides an immediate health check of your active execution window:
Total Runs: The total volume of unique flow runs currently handled within the monitoring view.
Succeeded (Green bar): The exact number of flow runs that completed their tasks from end to end successfully.
Failed (Red bar): The number of flow runs that stopped due to structural errors, data payload issues, or an execution that breached governance boundaries.
Managing Expectations & Performance Optimization
Seeing a run in a Queued status does not mean your flow is stuck or crashed. The queue is a vital architectural design built for high-reliability asynchronous processing, ensuring that no data is ever dropped when volume spikes occur.
💡 Architectural Best Practice: If your business needs to process large datasets without stacking runs into the waiting queue, we recommend utilizing pagination and small delays. By passing your data in paced, cadenced batches, you ensure your flows stay comfortably under the 50 simultaneous slots limit — allowing your data to process instantly without ever needing to wait in line.