Web Access-Log Analysis

Web Access Logs

DCN maintains recent web access logs that may help web publishers analyze document requests for their web sites.

The log files are in NCSA-standard format and are located at:

/usr/local/etc/httpd/logs/

on Wheel. Log files currently are rotated weekly. The current (live) access log is named access_log.

Previous logs are named in the format YRMODY.access_log where YRMODY is the date following the last full day the log received records. Log files for at least the last month should always be available.

Log Analysis Tools

Since log files are in a popular format, there are many tools available for analysis and experienced users may prefer to search out their own tools. (Yahoo's Log Analysis Tools List is a good starting point.)

DCN provides two UNIX command-line tools which may satisfy many users:

  • wwwstats is a modified version of UC Irvine's popular log analysis program. It can consolidate statistics from multiple log files and provides breakdowns of:
    • Daily Transmission Statistics
    • Hourly Transmission Statistics
    • Total Transfers by Client Domain
    • Total Transfers by Reversed Subdomain
    • Total Transfers from each Archive Section

  • myhits is a little easier to use and can render usage statistics in fancy bar graphs. It's output format is also nearly identical with the old DCN web-based log analysis tool and is much more straightforward then wwwstats.

Using the Programs

You may run either program by typing its name at the UNIX command prompt. Each -- by default -- searches the current log file for document requests for files stored in your account. Each creates a formatted HTML file, access.html, of its results in your public_html directory where you may read it with a web browser.

Each program has many options that allow it to analyze different log files, search for different URLs, and much more. To learn the options, invoke either program with a "-h" command-line argument.

Read the examples page for a simple tutorial on using myhits for log analysis.

Experienced UNIX hands may also run either program as a batch job to cut connect time. Please run large jobs only during light server load periods or you may lose friends.