Getstats 3.0 Set CommandsSome of the settable options apply to subsequent read commands. The other options apply to subsequent write commands.
Command arguments that contain uppercase characters must be quoted. If they are all lowercase you can leave off the quotes.
Turn this on (before reading the log files) if you want byte count information included in the reports. Turn it off for compatibility with old report files which show only hit counts.
This is usually the first command in your getstats command file. The reporting date is part of the Extract filenames created by set xtroot and Domain Report filenames created by set domroot.
set date 19980214
Note: the dates that appear in the report headers are obtained from the log records, not the set date command.
Set this to the hostname of your web server. It must be set before any read commands are issued since it will be applied to any log records that do not contain a vhost field. We only put the vhost field on the virtual host requests to save space in the log file.
set defaulthost www.columbia.edu
Specifies the number of domain levels to keep in memory. This will be the upper bound for subsequent write domain commands. Note that domlevels must be at least 2 if you are creating Domain Reports for a set of subdirectories using the set domroot command.
We recommend setting domlevels to 2 because 1 level does not provide enough detail in the domain reports, and 3 levels requires too much memory.
Specifies the number of dir tree levels to keep in memory. This will be the upper bound for subsequent write dirtree, write error, and write redir commands. The maximum value for this option is 6. To increase this upper bound, modify the GRANULARITY symbol in gs-tree.c, and recompile getstats.
Specifies the URL of the image that will appear on each report. Issue this command AFTER selecting a server with the set servername command. You will need to set icon again after each set servername command.
If your reports are on a secure server (https:), your icon URL should also be on a secure server. If your reports are on a regular web server (http:), make sure your icon URL also points to a regular server. This will avoid web browser warnings about mixing secure data with non-secure data in the same document. Here is an example:
set servername "http://www.columbia.edu/"
set icon "http://www.columbia.edu/httpd/getstats/getstats.gif"
The 2 level Domain Report for the specified server will be written to the specified directory. Here are a couple of examples:
set domroot www.columbia.edu /www/data/httpd/reports/columbia/dom
set domroot www.ciaonet.org /www/data/httpd/reports/ciao/dom
In addition to the complete Domain Report the set domroot command will automatically create separate Domain Reports for each subdirectory of the specified directory. This means you can create dozens of reports with a single command. To create a domain report for the /cu/libraries directory you would simply create a subdirectory called /cu/libraries, under the domain root /www/data/httpd/reports/columbia/dom. The showstats CGI script will find it there.
You must set date before you set domroot. Directories that contain a .nodomain file will not get a Domain Report. Use the write domain command to create additional Domain Reports.
Turn this off to include all file types in subsequent Dir Tree Reports. Turn this on to count only the HTML files, e.g.:
set ignoreimages no
write dirtree 6 /www/data/httpd/report/columbia/19980214.dtrreq6.html
set ignoreimages yes
write dirtree 6 /www/data/httpd/report/columbia/19980214.htrreq6.html
Turn this on to include only the institutional directories in subsequent Dir Tree Reports (no public_html directories). Turn this off to include only the public_html directories in subsequent Dir Tree Reports. For example:
set ignorepublic yes
write dirtree 6 /www/data/httpd/report/columbia/19980214.dtrreq6.html
set ignorepublic no
write dirtree 6 /www/data/httpd/users/columbia/19980214.dtrreq6.html
Specifies how often getstats will print a status message while it is running.
Include only the requests that begin with the specified string. This command would have to be used before reading the log files.
set reqmask "/~beecher"
After reading the log files, you must set servername to select a server, then issue the set icon command and create all the desired reports for that server. Repeat the process for all virtual servers that are included in those log files.
set servername "http://www.columbia.edu/"
set icon "http://www.columbia.edu/httpd/getstats/getstats.gif"
Set this to alpha if you want subsequent Dir Tree Reports sorted alphabetically. Set it to nreq if you want subsequent Dir Tree Reports sorted by number of requests.
set sortkey alpha
write dirtree 6 /www/data/httpd/report/columbia/19980214.dtralp6.html
set sortkey nreq
write dirtree 6 /www/data/httpd/report/columbia/19980214.dtrreq6.html
Extract files will be written to each subdirectory of the specified directory. Here are a couple of examples:
set xtroot www.columbia.edu /www/data/httpd/log/xt/www
set xtroot www.ciaonet.org /www/data/httpd/log/xt/ciao
This means you can create dozens of extract files with a single command. To create an extract file for the /cu/libraries directory you would simply create a subdirectory called /cu/libraries, under the extract root /www/data/httpd/log/xt/www. getstats will keep about 50 files open simultaneously. If you have more extract files than that, getstats will run a bit more slowly.
The extract files will be anonymized: first part of the host name converted to "x", and the last part of the IP address converted to "0". The virtual host field will be removed, since there is a separate extract file for each virtual host. The parentheses will be removed for compatibility with other web log analyzers. The extract files will be compressed after they are written.
You must set date before you set xtroot. Directories that contain a .noextract file will not get an extract file.
Academic Information Systems
212 854.1919
consultant@columbia.edu