All,
I've been going back and forth with development on a way of best leveraging a log rotation capability in our newly deployed LMS 3.2 rollout on Solaris 10.
My issue:
As of now, we use the services facility built into Solaris to start/stop the syslog service and logrotate (via logadm) our LMS log file (syslog_info) on a weekly basis. I added an argument in logadm to HUP the syslogd service which I think resets the file pointer and/or forces syslogd service to re-read the syslog.conf file. The problem with this setup is that in order for me to logrotate, I additionally have to stop and start a whole chain of processes so that the LMS application can carry on its duties of collecting and analyzing this data. So what I would have to do without drilling down a lenghty process dependencies is to stop/restart LMS which I really dislike doing for a small part of the entire application suite.
Cisco TAC recommends that insead of using logadm, that we should use logrot.pl which I have used in the past on a limited basis. TAC informs me that this will essentially logrotate and stop/start the LMS as well but what I haven't been able to get out of them is how this script deals with the syslog service. I looked over the code but didn't see any reference to the actual syslog service.
In summary, I need to know:
1) why is logrot.pl preferred over logadm.
2) how does logrot.pl handle the syslog service.
Your feedback would be most appreciated...
Michael
