Skip to content
Calcrivo

Concurrent Build Calculator

Determine how many builds can run concurrently given executor count, and how many builds will queue when demand exceeds capacity.

Inputs

executors

Total executor slots across all agents.

builds

Number of builds simultaneously triggered/requested right now.

Max Concurrent Builds

8builds

Builds That Will Queue

5builds

Executor Utilization

100.0%

Step by step

  1. Max concurrent = executors

    8

    = 8 builds

  2. Running now = min(executors, requested)

    min(8, 13)

    = 8 builds

  3. Queued = max(0, requested − executors)

    max(0, 13 − 8)

    = 5 builds

How it works

The maximum number of builds Jenkins can run at any single instant is bounded strictly by the total executor count across all connected, online agents — one executor runs exactly one build at a time. When the number of simultaneously requested builds exceeds available executors, the excess forms a queue rather than running immediately, waiting for an executor to free up; this is the same constraint that drives wait time in the build queue calculator.

Formula

queuedBuilds = max(0, requestedBuilds - executors)

B
Requested concurrent builds
E
Total available executors
Q
Builds that will queue

Frequently Asked Questions

Can a single build ever use more than one executor?

Not by default for a single job's main flow, but a declarative pipeline with a `parallel {}` block launching multiple `agent` directives effectively consumes one executor per concurrent branch, so a single pipeline run can occupy several executors at once.

Do executor labels affect this calculation?

Yes in practice — this calculator assumes all executors are fungible and available to every requested build; if builds require specific labels (e.g. `docker`, `gpu`), the effective competing pool is only the executors matching that label, not the cluster-wide total.

What happens to queued builds if none ever get an executor?

They remain queued indefinitely (subject to any configured queue timeout) until an executor becomes free or the job is manually cancelled; persistent queuing under peak load is a strong signal to add executors or improve throttling.

You might also need