On Fri, Oct 18, 2013 at 12:34 PM, John R Pierce pierce@hogranch.com wrote:
On 10/18/2013 5:49 AM, Paolo De Michele wrote:
I have a dedicated server with several services running: ssh, ftp, httpd (with several sites andactive domains), the mail server (dovecot, postfix), dns.
I'd like to monitor all of these services in a graphical, easy, setting of thresholds and alerts via email. I would also like that if a customer wanted to see the graphs I could create codes read-only.
to echo whats already been said, and perhaps clarify...
Nagios is the classic alert package. while its usually run on a dedicated server and monitors a bunch of other servers, it certainly can be used on a single system. But, Graphing in Nagios is a bit of a pain.
Cacti is a excellent graphing package, same thing, its usually run on a central server that monitors lots of stuff on other servers, it can be run on the same machine. however, setting up alerts in Cacti is difficult. You may find older references to 'mrtg', well, mrtg was rewritten as rrdtool, and rrdtool is the basis of Cacti.
both of these systems have web based displays, and use 'agent' based data collection. Its not unusual to use both at once for their respective strong points.
with any of these systems, you typically have a line in the agent script for each thing you want to monitor on a given host. utility scripts such as check_postgresql.pl let you get extensive data out of postgres databases
There is also OpenNMS: http://www.opennms.org. Probably overkill for a single host, but good if you intend to scale up and since a yum repository is maintained, installing isn't bad. It normally uses snmp and remote probes of network protocols instead of a dedicated local agent, but does have the ability to use some NRPE stuff from nagios if you want. Don't think there is a usable 'read only' login for outsiders but it does have an embedded jasper reports server for publishing fancy reports that can be emailed as pdfs.