A straightforward python tool to generate some HTTP load with connections kept alive and not shared between clients.
httpload uses aiohttp which in turn builds on asyncio which is introduced in python 3.4. Furthermore isodate is used to parse ISO 8601 formatted durations given as command line options.
Install using pip (e.g. in a virtual environment created with pyvenv):
pip install git+https://github.com/TargetHolding/httpload.git
Output of python -m httpload -h:
optional arguments:
-h, --help show this help message and exit
-t TARGET, --target TARGET
-d DELAY, --delay DELAY
-r RAMPUP, --rampup RAMPUP
-c CONNECTIONS, --connections CONNECTIONS
-l LENGTH, --length LENGTH
-rc RECONNECT, --reconnect RECONNECT
Example which loads the server(s) at http://www.example.org/ for one hour with 1000 connections. Connections will be started at a rate of approximately 10 every second. The client will reconnect if the server disconnects. On each of these a message is sent every .5 seconds. I.e. the request rate will be approximately 2000 messages / second.
python -m httpload \
--target http://www.example.org/ \
--length 1H \
--connections 1000 \
--rampup 10 \
--reconnect yes \
--delay .5S
httpload writes a line to standard out around every 10 seconds with the total number of complete requests, the time passed, the request rate in the last 10 seconds and the total number of failed requests.
Furthermore a line is written to standard out when the requested number of connections was established.
When httpload completes or when httpload is interrupted (e.g. with CTRL+C or a SIGTERM signal) asummary is written to standard out with:
- the test duration,
- the number of completed and failed request,
- the average request rate over the entire test period,
- the mean latency and the standard deviation
- per HTTP response code and error the total number of occurrences
$ python -m httpload --target https://acceptatie.diatoetsen.nl/home/ --connections 1000 --rampup 10 --length 1H --reconnect yes --delay .5S
starting httpload, creating 1000 connections at 10.0 connections / second
1050 reqs completed in 0:00:10.049253 at 104.5/s 0 failed
4098 reqs completed in 0:00:20.049609 at 304.8/s 0 failed
8062 reqs completed in 0:00:30.050791 at 396.4/s 0 failed
11691 reqs completed in 0:00:40.051718 at 362.9/s 0 failed
15408 reqs completed in 0:00:50.054016 at 371.6/s 0 failed
19143 reqs completed in 0:01:00.055118 at 373.5/s 41 failed
22812 reqs completed in 0:01:10.058116 at 366.8/s 114 failed
26482 reqs completed in 0:01:20.061200 at 366.9/s 157 failed
30122 reqs completed in 0:01:30.091521 at 362.9/s 181 failed
-- created 1000 connections in 0:01:39.952067 --
33429 reqs completed in 0:01:40.100291 at 330.4/s 253 failed
37435 reqs completed in 0:01:50.110527 at 400.2/s 366 failed
^C
39036 reqs completed in 0:01:54.481615 at 366.3/s 388 failed
--------------------------------------------------------------------------------
test duration = 0:01:54.481703
completed = 39036
failed = 388 (0.0%)
avg. request rate = 341.0 / second
latency (mean) = 0:00:00.544008
latency (std) = 0:00:04.228234
response codes / errors:
200 = 39036
server disconnected = 388
--------------------------------------------------------------------------------
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent