UPTIME version 0.2

Source: http://www.illuin.org/Software/uptime-0.2.tar
Source: http://www.illuin.org/Software/uptime-0.2.tar.gz

This is a program I wrote to produce uptime and availability statistics for the fileservers on campus. It makes use of the BER support produced for MRTG. HOWEVER it only uses a single socket and doesn't need to wait for responses before polling other devices.

How it works.

The process sends the SNMP queries to each host in the file over $maxtime, polling each device approximately every $polltime seconds. The use of $maxtime means that you can prevent the script from flooding the network with queries. The use of $maxtime gives the shortest period a host could be reloaded in without the script noticing. (eg if a system restarts twice in $maxtime the script only sees one of these events).

After the script has been running for $compresstime seconds it compresses the logfiles and exits. I make use of the supervise program from Dan Bernstein to keep uptime.pl running, it prevents more than one instance running at once, and allows the program to be run from a crontab which makes it more resilient to failures. Though I find that the script can run for days/weeks without crashing.

crontab entry:
5 * * * * cd /info/uptime ; supervise /info/uptime ./uptime.pl >/dev/null 2>&1 &

calculation.pl

This script compresses the log-file, removing all redundant elements from it. It writes out a series of HTML tables for including in status pages

crontab entry:
6 * * * * cd /info/uptime ; ./calculation >/dev/null 2>&1

TODO:

decide if it really need running undersupervise/crontab
make 'peak-periods' configurable
allow 'deaddays for holidays
mysql back-end
php generation of html pages

References:

MRTG

daemontools package