Introduction
We wanted to verify whether the sharing created by running on a virtual machine made any statistical difference to the PingER results. Thus we compare the results from PinGER running on a bare metal machine (pinger.slac.stanford.edu) vs running on a VM.
Installation of PingER Measurement Agent (MA pinger2.pl)
Spin up the virtual machine.
- Its nickname is pingervm
- It is called dhcp-nebula-124.26.slac.stanford.edu (134.79.124.66)
- Install Apache
- yum install -y httpd
Install pinger2.pl
- You need a writeable /usr/local (SLAC uses AFS
- Get root access (sudo -s)
- Set usrlocal=local in /etc/taylor.opts
- sudo taylor everything to make the change take effect (example)
- Install lynx (example), XML::Simple (example)
- It also needs ping (usually pre-installed in /bin/ping), ping6 (usually pre-installed in /bin/ping6), dig (usually pre-installed in /usr/bin/dig), and mail (usually pre-installed in /bin/mail)
- Follow the instructions at: Installation Overview
- Then
Verify installation
- To see if the cronjob is running look at the dates on the files /usr/local/share/pinger/pingerCronStat.stdout and /usr/local/share/pinger/pingerCronStat.stderr, they should be within the last 30 mins (example)
- Examine the above files (in particular the stderr file) to verify pinger2.pl is properly configured (an example of a typical error in the stderr file). If all is well then the stderr file will be empty. Example of a normal stdout file.
- Look at the latest file in /usr/local/share/pinger/data/ and look at the most recent data (example).
Installation of traceroute.pl and ping_data.pl
These are CGI scripts to enable a reverse traceroute and ping server (traceroute.pl), and to enable the gathering of data (ping_data.pl) from the archive sites. The two scripts were installed in the standard CGI script location /var/www/cgi-bin as defined in /etc/httpd/conf/http.conf. They are accessible from within SLAC as: http://dhcp-nebula-124-66.slac.stanford.edu/cgi-bin/ping_data.pl and http://dhcp-nebula-124-66.slac.stanford.edu/cgi-bin/traceroute.pl. Since they are on port 80 there is no access to the web server from offiste.
To ensure the web server (Apache) restarts after a reboot I also had to issue:
[cottrell@dhcp-nebula-priv-52-7 ~]$ sudo -s [sudo] password for cottrell: [root@dhcp-nebula-priv-52-7 cottrell]# /sbin/chkconfig httpd on [root@dhcp-nebula-priv-52-7 cottrell]# /etc/init.d/httpd restart Stopping httpd: [FAILED] Starting httpd: [ OK ] [root@dhcp-nebula-priv-52-7 cottrell]# ps -efl | grep httpd 1 S root 3056 1 0 80 0 - 46710 poll_s 07:42 ? 00:00:00 /usr/sbin/httpd 5 S apache 3058 3056 0 80 0 - 46710 inet_c 07:42 ? 00:00:00 /usr/sbin/httpd 5 S apache 3059 3056 0 80 0 - 46710 inet_c 07:42 ? 00:00:00 /usr/sbin/httpd 5 S apache 3060 3056 0 80 0 - 46710 inet_c 07:42 ? 00:00:00 /usr/sbin/httpd 5 S apache 3061 3056 0 80 0 - 46710 inet_c 07:42 ? 00:00:00 /usr/sbin/httpd 5 S apache 3062 3056 0 80 0 - 46710 inet_c 07:42 ? 00:00:00 /usr/sbin/httpd 5 S apache 3063 3056 0 80 0 - 46710 inet_c 07:42 ? 00:00:00 /usr/sbin/httpd 5 S apache 3064 3056 0 80 0 - 46710 inet_c 07:42 ? 00:00:00 /usr/sbin/httpd 5 S apache 3065 3056 0 80 0 - 46710 inet_c 07:42 ? 00:00:00 /usr/sbin/httpd 0 S root 3083 3029 0 80 0 - 26328 pipe_w 07:46 pts/0 00:00:00 grep httpd