Vedro profiling

Vedro profiling - plugin for Vedro framework for measuring resource usage of tests

The plugin measures CPU and memory usage during test execution and exports metrics in k6-compatible NDJSON format, making it easy to integrate with performance monitoring tools like Grafana, InfluxDB, and other observability platforms.

Installation

$ vedro plugin install vedro-profiling

$ pip3 install vedro-profiling

# ./vedro.cfg.py
import vedro
import vedro_profiling


class Config(vedro.Config):
    class Plugins(vedro.Config.Plugins):
        class VedroProfiling(vedro_profiling.VedroProfiling):
            enabled = True

Usage

Basic Usage

Enable profiling for your test run:

$ vedro run --enable-profiling

This will create a .profiling/profiling.ndjson file with CPU and memory metrics in k6-compatible format.

With Custom Run ID

$ vedro run --enable-profiling --run-id load-test-2026-01-26

With Visualization

Generate matplotlib plots alongside the metrics:

$ vedro run --enable-profiling --draw-plots

Configuration

Advanced Configuration

# ./vedro.cfg.py
import vedro
import vedro_profiling


class Config(vedro.Config):
    class Plugins(vedro.Config.Plugins):
        class VedroProfiling(vedro_profiling.VedroProfiling):
            enabled = True
            enable_profiling = True
            
            # Profiling methods: "default" (psutil), "docker" (containers)
            profiling_methods = ["default", "docker"]
            
            # Polling interval in seconds
            poll_time = 1.0
            
            # Generate plots
            draw_plots = True
            
            # Docker Compose project name for container monitoring
            docker_compose_project_name = "my-project"
            
            # Custom run identifier
            profiling_run_id = "staging-load-test"
            
            # Additional tags for metrics
            additional_tags = {
                "env": "staging",
                "team": "performance",
                "region": "us-east-1"
            }

Output Format

The plugin generates metrics in NDJSON (newline-delimited JSON) format compatible with k6 and other performance monitoring tools.

File Location

• Metrics: .profiling/profiling.ndjson
• Plots (if enabled): .profiling/*.png

NDJSON Structure

The output file contains metric definitions followed by data points:

{"type":"Metric","metric":"cpu_percent","data":{"type":"gauge","unit":"percent"}}
{"type":"Metric","metric":"memory_usage","data":{"type":"gauge","unit":"megabytes"}}
{"type":"Point","metric":"cpu_percent","data":{"time":"2026-01-26T10:00:00.123Z","value":25.5,"tags":{"target":"app-1","method":"docker","run":"my-test-123"}}}
{"type":"Point","metric":"memory_usage","data":{"time":"2026-01-26T10:00:00.123Z","value":512.3,"tags":{"target":"app-1","method":"docker","run":"my-test-123"}}}

Metrics

• cpu_percent - CPU usage percentage (gauge)
• memory_usage - Memory usage in megabytes (gauge)

Tags

Each data point includes the following tags:

• target - Container name or process name
• method - Profiling method (docker or default)
• run - Unique run identifier
• Custom tags from configuration

Processing Data

Extract Data Points Only

cat .profiling/profiling.ndjson | jq -c 'select(.type=="Point")'

Filter by Tags

cat .profiling/profiling.ndjson | jq -c 'select(.type=="Point" and .data.tags.env=="staging")'

Convert to InfluxDB Line Protocol

cat .profiling/profiling.ndjson | jq -r '
  select(.type=="Point") | 
  "\(.metric),target=\(.data.tags.target),method=\(.data.tags.method) value=\(.data.value) \(.data.time | fromdate * 1000000000)"
'

Aggregate Metrics

# Average CPU by target
cat .profiling/profiling.ndjson | jq -s '
  [.[] | select(.type=="Point" and .metric=="cpu_percent")] |
  group_by(.data.tags.target) |
  map({target: .[0].data.tags.target, avg_cpu: ([.[].data.value] | add / length)})
'

Features

• Multiple Profiling Methods: Monitor both system-level metrics (via psutil) and Docker container metrics
• k6-Compatible Format: Export metrics in NDJSON format for easy integration with monitoring tools
• Custom Tags: Add custom tags for filtering and grouping metrics
• Visualization: Generate matplotlib plots for quick visual analysis
• Non-Blocking: Uses background threads to minimize impact on test execution
• Flexible Configuration: Configure via code or command-line arguments

Integration Examples

Grafana + InfluxDB

1. Convert NDJSON to InfluxDB Line Protocol
2. Import into InfluxDB:

influx write --bucket performance --file profiling.influx

3. Create Grafana dashboard using InfluxDB as data source

Custom Analysis

import json

# Read and process metrics
with open('.profiling/profiling.ndjson') as f:
    for line in f:
        point = json.loads(line)
        if point['type'] == 'Point':
            metric = point['metric']
            value = point['data']['value']
            tags = point['data']['tags']
            # Process metrics...