neorants

The day started off with some nasty AT&T backbone issues, one of which managed to convince me that my web and mail server was down for a while this morning because ICMP was getting through just fine yet I couldn’t make a TCP connection to save my life. After the relief of finding out most people could get to it just fine I’ve gone on to have a great day, though…

A friend and coworker, whom I’ve mentioned before, has been working on a script to create some SLA reports. SLA stands for service level agreement. You know, the kind of contract you give your customers that says “we guarantee you 99.9% uptime or you can beat us up and take our lunch money.” Anyway, he had two scripts written, a perl script and a shell script, totaling 14 lines. I told him there was room to make the scripts more efficient, but he said it would take a lot of extra lines of code.

I told him I didn’t think it would. In fact, I bet I could make the changes and end up with less lines of code than he currently had. He started counting lines and said there was no way.

“Bet me,” I said.

“No way, you can’t do it in less,” he replies.

“Bet me!” I insisted.

“You’re not gonna count my whitespace and comments, are you?” he stated, obviously intrigued.

“Of course not. No tricks, just bet me.”

“OK, what are the stakes?”

“How about the loser buys the winner lunch?”

“You’re on.”

Silly, silly man. I’ll take that lunch with extra groveling. ;) Here’s the final product, which ended up as 7 lines (we counted lines as lines ending with a semicolon, which constitutes the end of a perl statement). It’s a little dirtier than I would have liked, but pretty elegant considering what we had to work with:

#!/usr/bin/perl -w

use DB_File;

open (DATA, 'zcat /var/log/cacti/archive/cacti-output.log.2003-02* | ') ||

   die "can't open logs!\n";

dbmopen (%RESULTS, 'results.dbm', 0666);

dbmopen (%RESULTSNU, 'resultsnu.dbm', 0666);

while (<DATA>) {

   /\/usr\/local\/www\/cacti\/rra\/([^\s]+)/ && $RESULTS{"$1"}++;

   /\/usr\/local\/www\/cacti\/rra\/([^\s]+).+N:U/ && $RESULTSNU{"$1"}++;

}
while ($key = each(%RESULTS)) {

   print "$key = $RESULTS{$key} " . ($RESULTSNU{$key} || 0) . " " .

      (($RESULTS{$key} - ($RESULTSNU{$key} || 0)) / $RESULTS{$key} *

      100). "%\n";

}
Of course, he hates me now.