weighttp¶

About¶

weighttp (pronounced weighty) is a lightweight and small benchmarking tool for webservers.

weighttp was designed to be very fast and easy to use and only supports a tiny fraction of the HTTP 1.x protocol in order to be lean and simple.
weighttp supports multithreading to make good use of modern CPUs with multiple cores as well as event-driven I/O for concurrent requests within a single thread.

Similar tools¶

• ab
• httperf
• httpress
• wrk2

Deployment¶

Obtaining¶

Use git to fetch the latest source:

$ git clone https://git.lighttpd.net/lighttpd/weighttp.git

or download a .tar.gz from https://git.lighttpd.net/lighttpd/weighttp/archive/master.tar.gz

Binary packages are available from https://build.opensuse.org for openSUSE and Debian based distributions:

• server:http
• home:stbuehler:lighttpd-utils

Building¶

meson¶

meson setup --buildtype debugoptimized --prefix /usr/local build
cd build
meson compile
#sudo meson install

autotools¶

$ ./autogen.sh
$ ./configure
$ make

The executable is created as src/weighttp

Alternatively, you can use gcc directly.

$ gcc -g2 -O2 src/*.c -o weighttp -lpthread -lm

Usage¶

weighttp uses similar command line arguments as apache bench (ab) for the subset of features supported by weighttp:

weighttp <options> <URI>
  -n num     number of requests      (mandatory)
  -t num     thread count            (default: 1)  (note: differs from @ab -t timeout@)
  -c num     concurrent clients      (default: 1)
  -k         keep alive              (default: no)
  -K num     num pipelined requests  (default: 1)  (note: not a feature of @ab@)
  -6         use ipv6                (default: no)
  -i         use HTTP HEAD method    (default: GET)
  -m method  use custom HTTP method  (default: GET)
  -H str     add header to request ("label: value"); repeatable
  -b size    socket buffer sizes (SO_SNDBUF, SO_RCVBUF)
  -B addr    local address to bind to when making outgoing connections
  -C cookie  add cookie to request ("cookie-name=value"); repeatable
  -F         use TCP Fast Open (RFC 7413)
  -T type    Content-Type header to use for POST/PUT data,
             e.g. application/x-www-form-urlencoded
                                     (default: text/plain)
  -A string  add Basic WWW Authorization   (str is username:password)
  -P string  add Basic Proxy-Authorization (str is username:password)
  -X proxy   proxy:port or unix domain socket path beginning w/ '/'
  -p file    make HTTP POST request using file contents for body
  -u file    make HTTP PUT request using file contents for body
  -d         do not report extended percentiles (beyond min,mean,max)
  -S         do not show standard deviation confidence warnings
  -l         (ignored; compatibility with Apache Bench (ab))
  -e file    output CSV file with percentages for total time to serve
  -r         (ignored; compatibility with Apache Bench (ab))
  -a time    max seconds to run benchmark (or until -n requests)
  -s cutoff  seconds to cutoff fails (default: 4)
  -q         quiet: do not show version header or progress
  -h         show help and exit
  -V         show version and exit

Troubleshooting¶

Benchmarking a webserver can result in a lot of sockets being created, especially if you don't use the -k parameter to reuse existing connections.
This can cause your system to run out of TCP port numbers and you will see the following error:

error: connect() failed: Cannot assign requested address (99)

Linux has a hardcoded 60 seconds timeout before it reuses a previously in use port. You can mitigate the problem by tuning 2 kernel parameters:

TCP_TW_REUSE¶

Linux can be configured to allow reuse of sockets which are in the TIME_WAIT state when it is safe from a protocol perspective.

echo 1 > /proc/sys/net/ipv4/tcp_tw_reuse

or add "net.ipv4.tcp_tw_reuse = 1" to your /etc/sysctl.conf to make the change permanent. You need to reload it via sysctl /etc/sysctl.conf

TCP_TW_RECYCLE¶

Note: TCP_TW_RECYCLE has been removed since Linux 4.12

Linux before Linux 4.12 can be configured to allow fast recycling of sockets which are in the TIME_WAIT state even if it is not safe from a protocol perspective.
You should not use this in a production environment. It can cause issues with load balancers and other mayhem.

echo 1 > /proc/sys/net/ipv4/tcp_tw_recycle

or add "net.ipv4.tcp_tw_recycle = 1" to your /etc/sysctl.conf to make the change permanent (bad idea). You need to reload it via sysctl /etc/sysctl.conf