By Kevin Howard
This blog post is a guest post by Kevin Howard from MRTGExamples.com. This site is a fantastic resource for all things related to implementing and getting the most out of MRTG. I have it listed in my links on the right side of my site under the Blogroll.
What I have realized that might be of interest to those who deploy or support web servers is that MRTG can be of great help to internet sites, specifically viewing machine level monitoring and historical data, without logging into the host.
A web page either served from the host or a separate monitoring server can report vital data from other machines to the monitoring staff. This may include a 24×7 Network Operations Center or other support staff that may not have shell access to the machine. As web sites grow larger and larger, simply logging into one or two of the machines and typing a command to get the current status would rely on your memory to recall the previous administrative check.
Of course, MRTG tracks data, by the day, week, month and year. Monitoring or system administrative staff can rely on MRTG monitoring vital services to reveal something “out of the ordinary”. Current data, past hours, days, or weeks could then be examined to reveal a problem, symptom or trend. Monitoring staff can then page or contact a technical engineer with some information as to what is going on with the web, application or database server.
Application level information can also be fed and monitored via MRTG such as general stats, threads, open files, disk space or any command you can run CLI, the numeric output can be used.
Other data can be gathered via script or command, that is interpreted by a simple perl script revealing a (1 or 0 for example) that a process is “ok” or “not ok”, that can be simply graphed as well. SNMP is a straight way of gathering counter information and graphing it as well, this is much more focused on a particular item, and needs interpretation done by the monitoring staff. Basic training on what is a HIGH number or LOW number, or what indicates trouble and what does not.
Examples of this and more are at: http://www.mrtgexamples.com
