The Linux box currently hosting stonehenge.com is in a rented space at a co-location facility. As a result of the Internet shakeout happening everywhere, the co-lo facility was bought by a larger networking company and we’ve been having network interruptions, including complete loss of service, from time to time. The administrator of the box came to me looking for evidence that these outages had been going on for extended periods of time so that he could take that to the new owner, get some of his money back, and pass the savings along to me.
The Linux box currently hosting stonehenge.com is in a rented space at a co-location facility. As a result of the Internet shakeout happening everywhere, the co-lo facility was bought by a larger networking company and we’ve been having network interruptions, including complete loss of service, from time to time. The administrator of the box came to me looking for evidence that these outages had been going on for extended periods of time so that he could take that to the new owner, get some of his money back, and pass the savings along to me.
It occurred to me that I could use the logs from my Web server, and it would have been nice to use the database-driven logs as I did in last month’s column, but I needed data from a year ago. I did have the database dumps, but not enough disk space to restore all the data. Fortunately, I also had the standard Apache access_log files broken up by day, compressed separately and dating all the way back to my first hosting machine. It was from these files that I was able to glean the data.
First, let’s consider the requirements. I wanted an hour-by-hour hit count for a given range of dates. I also wanted to filter out any local hits, since those hits don’t reflect network connectivity issues. I also wanted to be reasonably efficient about date parsing. That ruled out running Date::Manip’s general date recognition routines or anything else…
Please log in to view this content.
Not Yet a Member?
Register with LinuxMagazine.com and get free access to the entire archive, including:
