Table of contents
Project description
Tracers is an Open-Source APM (Application monitoring) project that offers you minimal overhead wrappers for profiling your code execution flow
🛈 Finished transaction: 3.81 seconds
# Timestamp Net Total Call Chain
1 0.00s 0.00s [ 0.0%] 3.81s [100.0%] ✓ async function_a
2 0.00s 0.10s [ 2.6%] 0.10s [ 2.6%] ¦ ✓ async asyncio.tasks.sleep
3 0.10s 0.50s [ 13.1%] 0.50s [ 13.1%] ¦ ✓ time.sleep
4 0.60s 0.00s [ 0.0%] 3.21s [ 84.2%] ¦ ✓ async function_b
5 0.60s 0.10s [ 2.6%] 0.10s [ 2.6%] ¦ ¦ ✓ async asyncio.tasks.sleep
6 0.70s 0.00s [ 0.0%] 0.70s [ 18.4%] ¦ ¦ ✓ async function_c
7 0.70s 0.10s [ 2.6%] 0.10s [ 2.6%] ¦ ¦ ¦ ✓ async asyncio.tasks.sleep
8 0.80s 0.50s [ 13.1%] 0.50s [ 13.1%] ¦ ¦ ¦ ✓ time.sleep
9 1.31s 0.00s [ 0.0%] 0.10s [ 2.7%] ¦ ¦ ¦ ✓ async function_d
10 1.31s 0.10s [ 2.6%] 0.10s [ 2.6%] ¦ ¦ ¦ ¦ ✓ async asyncio.tasks.sleep
11 1.41s 2.00s [ 52.5%] 2.00s [ 52.5%] ¦ ¦ ✓ time.sleep
12 3.41s 0.10s [ 2.7%] 0.10s [ 2.7%] ¦ ¦ ✓ async asyncio.tasks.sleep
13 3.51s 0.00s [ 0.0%] 0.10s [ 2.7%] ¦ ¦ ✓ async function_d
14 3.51s 0.10s [ 2.7%] 0.10s [ 2.7%] ¦ ¦ ¦ ✓ async asyncio.tasks.sleep
15 3.61s 0.10s [ 2.7%] 0.10s [ 2.7%] ¦ ¦ ✓ async asyncio.tasks.sleep
16 3.71s 0.00s [ 0.0%] 0.10s [ 2.7%] ¦ ¦ ✓ async function_e
17 3.71s 0.10s [ 2.7%] 0.10s [ 2.7%] ¦ ¦ ¦ ✓ async asyncio.tasks.sleep
Count Net Total Function
3 3.00s [ 78.7%] 3.00s [ 78.7%] ✓ time.sleep
8 0.81s [ 21.2%] 0.81s [ 21.2%] ✓ async asyncio.tasks.sleep
1 0.00s [ 0.0%] 3.21s [ 84.2%] ✓ async function_b
1 0.00s [ 0.0%] 0.70s [ 18.4%] ✓ async function_c
1 0.00s [ 0.0%] 3.81s [100.0%] ✓ async function_a
2 0.00s [ 0.0%] 0.20s [ 5.3%] ✓ async function_d
1 0.00s [ 0.0%] 0.10s [ 2.7%] ✓ async function_e
Some blocks (skews) occurred in the event loop ¹
# Timestamp Delay
0 1.40s 2.00s
1 0.09s 0.50s
2 0.80s 0.50s
¹ Consider reviewing them carefully to improve the overall system throughput
Key Features
- Allows you to instrument any function or statement on your own code or any third party library
- Handles asynchronous code out-of-the box, no config required
- Exposes a high-level API:
- A decorator:
@trace, to instrument functions- Example:
@trace def any_function(*args, **kwargs): pass @trace async def other_function(*args, **kwargs): pass
- Example:
- A function:
callto instrument code in-line- Example:
call(any_function, *args, **kwargs) await call(other_function, *args, **kwargs)
- Example:
- Sync/async cases are handled internally
calland@traceare equivalent, you can choose the one that fits you best
- A decorator:
- It's Thread-safe, Async-safe, Process-safe and Context-safe
- You'll get accurate results in any scenario
- Introduces minimal overhead and it's easy to deploy!
- It's a pure python implementation based on contextvars and decorators. Your code runs as fast as if you were not profiling it
- It's easy to pin-point performance problems:
- Gives you the total execution time in seconds and %
- Allows you to identify points in time where your async event loop got blocked
- Made with love by nerds, for humans :heart:
Quick Introduction
Let's start with a very basic example:
import time
from dateutil.parser import parse
def example():
time.sleep(2.0)
your_business_logic('Sat Oct 11')
def your_business_logic(date: str):
parse(date)
time.sleep(1.0)
example()
Tracing its flow and gathering profiling information is a matter of decorating your functions:
--- a/examples/without_tracers.py
+++ b/examples/with_tracers.py
@@ -1,15 +1,18 @@
import time
from dateutil.parser import parse
+from tracers.function import trace
+@trace()
def example():
time.sleep(2.0)
your_business_logic('Sat Oct 11')
+@trace()
def your_business_logic(date: str):
parse(date)
time.sleep(1.0)
example()
If you run it, all the functions you decorated will be traced and you'll have metrics of the execution flow:
🛈 Finished transaction: 3.00 seconds
# Timestamp Net Total Call Chain
1 0.00s 2.00s [ 66.7%] 3.00s [100.0%] ✓ example
2 2.00s 1.00s [ 33.3%] 1.00s [ 33.3%] ¦ ✓ your_business_logic
Count Net Total Function
1 2.00s [ 66.7%] 3.00s [100.0%] ✓ example
1 1.00s [ 33.3%] 1.00s [ 33.3%] ✓ your_business_logic
From the output you can conclude:
- executing function example took a total of 3.0 seconds to complete
- function example takes 66.7% of the execution time
- function *example encloses 100% of your code time
- function example called function: your_business_logic
- function your_business_logic took 1.0 seconds out of the 3.0 seconds the function example needed to complete
- function your_business_logic represents 33.3% of your execution time
Tracing code is not limited to your own code. You can trace any callable object including third party packages, Python's standard library, and almost anything
The level of detail is up to you!
--- a/examples/with_tracers.py
+++ b/examples/with_detailed_tracers.py
@@ -1,18 +1,18 @@
import time
from dateutil.parser import parse
-from tracers.function import trace
+from tracers.function import call, trace
@trace()
def example():
- time.sleep(2.0)
+ call(time.sleep, 2.0)
your_business_logic('Sat Oct 11')
@trace()
def your_business_logic(date: str):
- parse(date)
- time.sleep(1.0)
+ call(parse, date)
+ call(time.sleep, 1.0)
example()
🛈 Finished transaction: 3.00 seconds
# Timestamp Net Total Call Chain
1 0.00s 0.00s [ 0.0%] 3.00s [100.0%] ✓ example
2 0.00s 2.00s [ 66.6%] 2.00s [ 66.6%] ¦ ✓ time.sleep
3 2.00s 0.00s [ 0.0%] 1.00s [ 33.4%] ¦ ✓ your_business_logic
4 2.00s 0.00s [ 0.0%] 0.00s [ 0.0%] ¦ ¦ ✓ dateutil.parser._parser.parse
5 2.00s 1.00s [ 33.3%] 1.00s [ 33.3%] ¦ ¦ ✓ time.sleep
Count Net Total Function
2 3.00s [100.0%] 3.00s [100.0%] ✓ time.sleep
1 0.00s [ 0.0%] 0.00s [ 0.0%] ✓ dateutil.parser._parser.parse
1 0.00s [ 0.0%] 1.00s [ 33.4%] ✓ your_business_logic
1 0.00s [ 0.0%] 3.00s [100.0%] ✓ example
Installation
We are hosted on PyPI: https://pypi.org/project/tracers
Just run: pip install tracers
or use the package manager you like the most
Examples
Check them out in the examples folder
1 maintainer
Metadata
1 maintainer
Release files for tracers 21.8.6876
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| tracers-21.8.6876.tar.gz | 12.3 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| tracers-21.8.6876-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 24.3 kB
Release files / tracers-21.8.6876.tar.gz
| Download URL | tracers-21.8.6876.tar.gz |
|---|---|
| Size | 12.3 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
4e63593715b13a9d8c9aa23b1f0d04f4696c6005544b4e40f335f9466f897cab
|
|
BLAKE2b-256 checksum How to use checksums |
dc0b63b60b15ba76dd93d2c286cc34324fbaf354e397bf3110a789eb507f1379
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
poetry/1.1.5 CPython/3.8.9 Linux/5.10.44
|
Release files / tracers-21.8.6876-py3-none-any.whl
| Download URL | tracers-21.8.6876-py3-none-any.whl |
|---|---|
| Size | 12.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
162102069e463b12dc1991688a58a9def1790ba4029c3a72790ea54f73a14160
|
|
BLAKE2b-256 checksum How to use checksums |
4459632eefa787e2bcb59bb1ea5c1229e8d64e4672ef82e230ddf3afc1701fc3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
poetry/1.1.5 CPython/3.8.9 Linux/5.10.44
|
