Running AWS
Signals
While running, AWS accepts three break signals:
- C
- Causes AWS to exist as soon as all current tasks have exited.
AWS will log a message to syslog about the current task count
if it's non-zero. You can send a break D to get an update on
the current number of tasks sent to syslog. A final
message will be logged when AWS exits.
- D
- Causes AWS to send some statistics to syslog:
- activetasks
- The number of tasks currently running
- hightasks
- The maximum number of tasks run simultaneously so far
- requestcounter
- The number of requests handled by the server so far
- E
- Causes AWS to send the current configuration to syslog.
- F
- Causes AWS to close and reopen the log files. If you've just
renamed them, this will cause it to open new versions. Otherwise,
it just keeps appending log messages to the old log files.
Maintenance
There are two rexx scripts that summarize the two log files
generated by AWS:
- access.rexx
- Reports on total requests and bytes shipped, the counts for
various request statuses, and the top 10 most popular URLs,
clients, referers and hosts. Finally a list of unparsable lines
is generated.
- error.rexx
- Reports on counts for various types of lines it recognizes. It
also generates a list of clients that asked for things they
couldn't handle (Mozilla/0.9 has a bad habit of claiming it
can't handle text/html, so it turns up a quite a bit here). It
then generates a list of the URLs of referers that referenced
non-existent URLs, and the URLs they referenced, followed by
a list of all URLs that were not requested but not found on the
server. Finally, it lists the lines that it didn't recognize.
You should rename the log files and run the analyzer over them - or
at least the error log analyzer - at regular intervals.
Mike Meyer