pReplay

pReplay replays the dependency graph. It starts from the first activity which is loading the root html. Then according to the graph, it it encounters a network activity, it makes a request for the corresponding url, or if it encounters a computation activity, it waits for a specific amount of time (mentioned in the graph). Once a particular activity is finshed it checks whether it should trigger dependent activities based on whether all activities that a dependent activity depends on are finished. pReplay keeps walking through the dependency graph this way until all activities in the dependency graph have been visited.

usage

./pReplay server testfile [http|https|http2|phttpget] [max-connections] [cookie-size]

• server : DNS name or IP address
• testfile : relative path to test file in json format
• protocol :
  • http : http 1.1
  • https : http 1.1 with SSL
  • http2 : http 2
  • phttpget : requests are handled by PHTTPGET. PHTTPGET has to be started seperately
• max-connections : maximum amount of concurrent connections
• cookie-size : size of cookie - works with http1 only

phttpget

• checkout the HTTPOverSCTP repo
• switch to multistream branch
• make
• put the phttpget binary in the pReplay directory
• run pReplay with phttpget option

https://github.com/NEAT-project/HTTPOverSCTP

BRANCH: multistream

tested with FreeBSD and Linux

thttpd

• checkout the thttpd repo
• switch to the multistream branch
• ./configure + make + make install
• /usr/local/sbin/thttpd -C /usr/local/www/thttpd_config

  
  https://github.com/nplab/thttpd/

BRANCH: multistream

Sample `thttpd_config`:

dir=/usr/local/www/data cgipat=/cgi-bin/* chroot logfile=/usr/local/www/logs/thttpd_log pidfile=/var/run/thttpd.pid max_age=0

tested with FreeBSD

## info
pReplay has been tested with curl-7.47.1