REST Server for Performance Testing

This project provides a simple REST based server for performance testing container infrastructure.

Getting the container

The container is hosted on ghcr.io.

Pull container using the following commands.

Docker

$ docker pull ghcr.io/appform-io/perf-test-server-httplib

Podman

$ podman pull ghcr.io/appform-io/perf-test-server-httplib

Running the container

docker run -p 8000:8000 ghcr.io/appform-io/perf-test-server-httplib

To run on podman replace docker with podman in the above command.

To stop container, press <Ctrl-C>

Usage

To access the container hit it using any load testing tool on port 8000 (or whatever host port you map the container port to).

Sample wrk command

$ wrk -c 100 -t20 -d 30 --latency  http://localhost:8000
Running 30s test @ http://localhost:8000
  20 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   227.56ms  324.11ms   1.90s    79.61%
    Req/Sec     2.62k     2.20k   11.79k    63.51%
  Latency Distribution
     50%  440.00us
     75%  443.61ms
     90%  795.54ms
     99%    1.01s 
  1064444 requests in 30.02s, 106.39MB read
  Socket errors: connect 0, read 0, write 0, timeout 93
Requests/sec:  35453.26
Transfer/sec:      3.54MB

Sample ab command

$ ab -k -t 30 -n 1000000 -c 20 http://localhost:8000/
This is ApacheBench, Version 2.3 <$Revision: 1879490 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/

Benchmarking localhost (be patient)
Completed 100000 requests
Completed 200000 requests
Completed 300000 requests
Completed 400000 requests
Completed 500000 requests
Completed 600000 requests
Completed 700000 requests
Completed 800000 requests
Completed 900000 requests
Completed 1000000 requests
Finished 1000000 requests


Server Software:        
Server Hostname:        localhost
Server Port:            8000

Document Path:          /
Document Length:        12 bytes

Concurrency Level:      20
Time taken for tests:   28.274 seconds
Complete requests:      1000000
Failed requests:        2
   (Connect: 0, Receive: 0, Length: 2, Exceptions: 0)
Keep-Alive requests:    800005
Total transferred:      104799863 bytes
HTML transferred:       11999976 bytes
Requests per second:    35368.31 [#/sec] (mean)
Time per request:       0.565 [ms] (mean)
Time per request:       0.028 [ms] (mean, across all concurrent requests)
Transfer rate:          3619.72 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:        0    0   9.8      0    1031
Processing:     0    0   7.1      0    5001
Waiting:        0    0   0.2      0      15
Total:          0    0  12.1      0    5001

Percentage of the requests served within a certain time (ms)
  50%      0
  66%      0
  75%      1
  80%      1
  90%      1
  95%      1
  98%      1
  99%      1
 100%   5001 (longest request)

APIS

Main API : /

This returns an output (Hello World!) in text format.

$ curl -vvv http://localhost:8000/
*   Trying 127.0.0.1:8000...
* Connected to localhost (127.0.0.1) port 8000 (#0)
> GET / HTTP/1.1
> Host: localhost:8000
> User-Agent: curl/7.81.0
> Accept: */*
> 
* Mark bundle as not supporting multiuse
< HTTP/1.1 200 OK
< Content-Length: 12
< Content-Type: text/plain
< Keep-Alive: timeout=5, max=5
< 
* Connection #0 to host localhost left intact
Hello World!

File download API: /files/test.txt

This returns a simple small file.

$ curl -vvv http://localhost:8000/files/test.txt
*   Trying 127.0.0.1:8000...
* Connected to localhost (127.0.0.1) port 8000 (#0)
> GET /files/test.txt HTTP/1.1
> Host: localhost:8000
> User-Agent: curl/7.81.0
> Accept: */*
> 
* Mark bundle as not supporting multiuse
< HTTP/1.1 200 OK
< Content-Length: 10
< Content-Type: text/plain
< Keep-Alive: timeout=5, max=5
< 
Test File
* Connection #0 to host localhost left intact

Healthchecks

If configuring probes for checking controller health, / can be set as healthcheck and it will always return HTTP 200/OK. In case container needs to be made unhealthy, please point health check to /unhealthy API.

Controlling behaviour of the /unhealthy API

By default, the API will return 200 OK 3 times and then it will start returning HTTP 500 Server Error. To increase the count from 3 to any number, please pass an environment variable HEALTHCOUNT with the appropriate number.

$ curl  -I http://localhost:8000/unhealthy 2>&1 |grep HTTP
HTTP/1.1 200 OK
$ curl  -I http://localhost:8000/unhealthy 2>&1 |grep HTTP
HTTP/1.1 200 OK
$ curl  -I http://localhost:8000/unhealthy 2>&1 |grep HTTP
HTTP/1.1 200 OK
$ curl  -I http://localhost:8000/unhealthy 2>&1 |grep HTTP
HTTP/1.1 500 Internal Server Error

Controlling behaviour using environment variables

The following environment variables canb used to control different aspects of the container.

• CORES - By default the container detects the number of cores and creates a server thread for that. This can be controlled by setting the CORES variable to the required number. Usage: -e "CORES=20"
• NOLOG - Do not do access logging. Set this to stop console logging. This will significantly improve performance for the container. Usage: -e "NOLOG=1"
• HEALTHCOUNT - Number of counts after which the /unhealthy api starts returning HTTP 500/Internal Server Error instead of HTTP 200/OK. Usage: -e "HEALTHCOUNT=20"
• DIE_FAST - Exit the container after specified number of seconds. By default this is not set, i.e. container doesn;t exit by itself. Usage: -e "DIE_FAST=5".

Building

The server is written in C++. Code is hosted here.

Clone the code

$ git clone [email protected]:appform-io/perf-test-server-httplib.git

Fetch dependencies

$ git submodule update --init --recursive

Build

make

This will generate a binary called test-server in the current directory and a docker container. Tag and push the container as needed.

License

Apache 2.0

Base image

This container uses Redhat Universal Base Image