CI pipeline execution analysis
CI pipeline execution2/5 runs distributed across 7 agents
Agent utilization
Agent 2 / 2 tasks completed / 2m 11s active45%
Agent 4 / 6 tasks completed / 1m 43s active35%
Agent 0 / 2 tasks completed / 1m 13s active25%
Agent 1 / 4 tasks completed / 1m 05s active22%
Agent 5 / 1 tasks completed / 1m 01s active21%
Agent 7 / 0 tasks completed
Agent 3 / 0 tasks completed
Improving agent utilization
An ideal agent utilization report will show even, high utilization across all agents. You might not see that for all distributed task executions though. For the most accurate assessment, watch how your agents perform on a CI pipeline execution where your entire project graph is affected (for example, after an NPM dependency is changed).
If you see low or uneven utilization across agents, here's some things you can try:
- Parallelize wherever you can
- Break apart projects with long-running targets, so their tasks can be more easily distributed across agents
- Break apart long-running targets such as e2e tests, so that they can run in parallel
Run durations
Improving run timings
Ideally you'd like to see all the runs in a CI pipeline execution happen in parallel. That may not be possible though, depending on how your tasks depend on one another. If you do see room for improvement, here's some things you can do to improve your run timings:
- Parallelize wherever you can