If you migrate an RRD-based tool like SmokePing, Cacti or ZeroD you may run into the problem, that the target system can no longer use the historic data, if the default number representation changes (big/little endian, 32/64 Bit, …).
In all these cases the following procedure should work. It’s important you dump on the original architecture and restore on the target system.
The default installation of snmpd in Debian 6 Squeeze will put warnings into the daemon.log.
For quick analysis of the options and health of a Java VM I used the jconsole tool (part of JDK) in the past. But that changed.
In the standard rsyslog setup of Ubuntu a kernel message goes to 3 log files (syslog, messages and kern.log), i.e. running LOG targets in your firewall setup can really hog your /var/log volume.
I decided to simply discard the kernel messages after logging them to kern.log.
Using RAID to improve availability is fine, but don’t forget to monitor it properly. Otherwise you will never know, that half of your mirror has already gone out of service for months.
If you are tired of scanning all kind of CERT advisories in your mailbox and you are running a nice Debian-based landscape using mostly package-based software, you should consider using check_apt from the Nagios plugins.
Without special reconfiguration on the SNMP daemon, you are able to check processes on a Linux resp. services on a Windows host remotely:
For the basic remote checks I evaluated several SNMP checks for Nagios and finally moved on with the following to begin with:
The Zero Frontend is currently facing a major upgrade or tuning, introducing views to aggregate components from different hosts.
Popups for the graphs are now integrated into the same window, there is an automatic refresh.