Traffic Generators: Difference between revisions
→Curl
(Created page with " = Siege = Siege is an open source stress / regression test and benchmark utility. siege -c100 -t30S -d10 -b -v aman.info.tm Some Extensions --header="Cookie: SESSb43b2...") |
(→Curl) |
||
| (28 intermediate revisions by the same user not shown) | |||
Line 1:
[[Category:Lab]]
__TOC__
<br />
= Siege =
Line 7 ⟶ 8:
Some Extensions
The some-urls.txt file is just a simple list of URLs on newlines:
Line 16 ⟶ 17:
= Httperf =
Installation:
sudo apt install httperf
Usage:
httperf --server waf.avitest.com --port 80 --num-conns 100 --rate 10 --timeout 1
= Curl =
curl -s -o /dev/null -w "%{http_code}" http://waf.avitest.com
seq 100 | parallel -j0 curl -s -o /dev/null -w "%{http_code}" http://waf.avitest.com
for i in `seq 1 99999`; do echo "Status Code:"; curl -s -o /dev/null -w "%{http_code}" https://10.1.1.1; sleep 1; done
;Bash Script
<pre>
# get output, append HTTP status code in separate line, discard error message
OUT=$( curl -sfI --connect-timeout 1 http://10.52.200.32/ | grep "HTTP" )
# get exit code
RET=$?
if [[ $RET -ne 0 ]] ; then
echo "Time out: $RET"
else
echo "$OUT"
fi
</pre>
while true; do ./monitor.sh; sleep 1; done
= Apache Benchmark =
ab -t 1 -n 1000 -c 300 http://waf.avitest.com/
Increase Apache MaxClients (now called MaxRequestWorkers in new versions) on Backend Server:
sudo nano /etc/apache2/mods-available/mpm_event.conf
MaxRequestWorkers 1000
ServerLimit 565
= Locust =
Installation:
sudo apt-get update
sudo apt-get -y install python-pip python-dev libxml2-dev libxslt-dev
sudo pip install locustio
Install pyzmq for tests across multiple servers to increase the testing capacity:
sudo pip install pyzmq
Set file limit to unlimited to ensure there is no OS problems with files at concurrency:
ulimit -n 9999
== Locust File ==
Single “task” which gets a specific webpage:
<syntaxhighlight lang="python">
from locust import HttpLocust, TaskSet, task
class UserBehavior(TaskSet):
@task
def get_something(self):
self.client.get("/something")
class WebsiteUser(HttpLocust):
task_set = UserBehavior
</syntaxhighlight>
The test below logs a user in to groupster then requests what will be their home page 1 time for every 3 times it requests a group page:
<syntaxhighlight lang="python">
from locust import HttpLocust, TaskSet
def login(l):
l.client.post("/login/process", {"email":"me@someemail.com", "pass":"password"})
def index(l):
l.client.get("/")
def group(l):
l.client.get("/group/1/")
class UserBehavior(TaskSet):
tasks = {index:1, group:3}
def on_start(self):
login(self)
class WebsiteUser(HttpLocust):
task_set = UserBehavior
min_wait=5000
max_wait=9000
</syntaxhighlight>
== Executing ==
* Go to the directory with locust file & execute below command:
* This only starts the interface to control the tests it does not actually start a test:
locust -f ./locastfile.py --host=http://somedomain.io --master
Go to the web interface to control the test variables and to start and stop it.
http://localhost:8089
== Distributed Mode ==
*To generate significant load it is almost always necessary to run it in distributed mode.
*Master server do not simulate any user itself.
*For this we have to start one or more slaves using –slave flag
*Both the master and each slave machine, must have a copy of the locust test scripts when running Locust distributed.
*Run below command on each slave nodes as well
ulimit -n 9999
Start “master” node:
locust --host=http://somedomain.io --master
Then start any “slave” nodes, giving them a reference to the master node:
locust --host=http://somedomain.io --slave --master-host=192.168.10.100
== Without WebUI ==
Runs the cli mode as below:
locust -f locustfile.py --no-web -c 1000 -r 100 --run-time 1h30m
In Distributed mode it will wait until all slave nodes have connected before starting the test:
locust -f locustfile.py --no-web -c 1000 -r 100 --run-time 1h30m --expect-slaves
= Slowhttptest =
Source: [http://manpages.ubuntu.com/manpages/xenial/man1/slowhttptest.1.html ubuntu.com]
{{UC}}
= Bombardier =
Source: [https://softwaretester.info/http-benchmarking-with-bombardier/ softwaretester.info]
Preparation
sudo apt install -y curl git
curl -O https://storage.googleapis.com/golang/go1.8.linux-amd64.tar.gz
tar xvf go1.8.linux-amd64.tar.gz
sudo chown -R root:root go
sudo mv go /usr/local/
Configure go (for user)
mkdir ~/.go
echo "GOPATH=$HOME/.go" >> ~/.bashrc
echo "export GOPATH" >> ~/.bashrc
echo "PATH=\$PATH:/usr/local/go/bin:\$GOPATH/bin" >> ~/.bashrc
source ~/.bashrc
go version
Install bombardier
go get -u github.com/codesenberg/bombardier
bombardier --help
Usage:
*Run with 10 connections on 5 sec and show latency statistics.
bombardier -d 5s -c 10 -l -k https://www.heise.de
*Test for 60 seconds at 250 Insecure connections per second :
bombardier -c250 -k -d 60s "https://10.70.28.12/tools/healthcheck?api-key=0145463c-df61-4ebe-bbf1-ed3f43t4f57"
<br />
;References
<references/>
<br />
<br />
<br />
{{DISQUS}}
| |||