https://github.com/sloev/metrace

Metrace - cpu/memory tracing for process trees
https://github.com/sloev/metrace

cpu memory metrics monitor python trace tracing tracking

Last synced: about 1 month ago
JSON representation

Metrace - cpu/memory tracing for process trees

• Host: GitHub
• URL: https://github.com/sloev/metrace
• Owner: sloev
• License: mit
• Created: 2019-09-17T09:43:31.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2020-09-02T21:39:24.000Z (over 4 years ago)
• Last Synced: 2024-11-01T19:52:32.654Z (about 2 months ago)
• Topics: cpu, memory, metrics, monitor, python, trace, tracing, tracking
• Language: Python
• Homepage: https://pypi.org/project/metrace/
• Size: 1.14 MB
• Stars: 10
• Watchers: 2
• Forks: 1
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
[![Latest Version](https://img.shields.io/pypi/v/metrace.svg)](https://pypi.python.org/pypi/metrace) [![Python Support](https://img.shields.io/pypi/pyversions/metrace.svg)](https://pypi.python.org/pypi/metrace)

# Metrace - cpu/memory tracing for process trees



Metrace makes it easy to get a nice interactive html plot of the cpu/memory usage of a whole process tree.







## Install

Metrace depends on `Plotly` and `Psutil`, both will be installed with metrace using pip:

```

$ pip install metrace

```

## Client usage (optional)

Metrace lets you annotate specific parts of your code with scopes and these will become part of your final plot output and end up look like this:







Metrace gives you a context manager `trace` you can wrap code blocks with:

```

$ cat foobar.py

from metrace import trace

with trace('creating 1000 objects'):

    l = []

    for i in range(1000):

        l.append(object())

```

To collect these traces you run your script with metrace as usual:

```

$ metrace run python foobar.py

```

## Commandline Usage

After installation you will have the **metrace** command available.

```

$ metrace run python main.py && metrace plot

```

**It has two subcommands:**

### run

```

$ metrace run python foobar.py

```

Runs a process and collects cpu/memory metrics for both the process and its children seperately.

Metrace outputs the metrics in a json file in current working directory named like: `metrace_1568385178098218.json` with the number being milliseconds since epoch.

### plot

```

$ metrace plot

# or

# to not open the html report automatically:

$ metrace plot --autoopen=no

# or

# to specify input/output files

$ metrace plot -i metrace_1568385178098218.json -o metrace_report.html

```

Plots cpu/memory usage for a given, or latest found, `metrace json file`.

Output is a html file where you can download the images, zoom in and hover to inspect the different traces.

## Samples

Docs folder contains a sample of a [metrace json file](./docs/metrace_1568385178098218.json) and a [metrace html report](./docs/metrace_report.html)

# Attribution

[plotly.min.js](https://github.com/plotly/plotly.js) Licensed under MIT, has been included in this library here: [metrace/plotly_latest.min.js](./metrace/plotly_latest.min.js)