Scanning the syslog

[kozlow]

I created a straight forward script the would scan the syslog.log and report any entries from the current day. I read through it and check if the current Month and Day equals the Month and Day in the log. If any entries are found, I email them to my group so we can review when we get in. This helps us get notified of any potential Hardware issues. I then run this at 23:59 each night...

It works great, unless the system has not been recycled in over a year. This is the case with many of our servers as production is 7X24X365.... I find the Month and Day, but there is no year in syslog.log. So it is now reporting errors from yesterday and a year ago.

In AIX I use "errpt -a -dH -dU -s `date +%m%d0000%y`"

Is there any product in HP-UX to display Hardware or Software errors by day.

 

[yai]

Hi,
I would recommend some kind of logfile rotation, as logfiles may get very big over the years. And by default, the size of syslog files is limited to 2GB (which can, of course, be changed). Scripts for logfile rotation may be found on the internet.
I also recommend to log all machines to a central syslog server, this has the following advantages:
- all messages of all servers are in chronological context, e.g. a network problem occuring on all servers simultaneously
- a central place for log rotation scripts
- a central place to backup your syslogs (you should keep them as long as you keep your data!)
- logs will be written to the same log even after multiple reboots
...
Goof luck.