----- Original Message ----- From: "Nic le Roux" <[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>> To: <[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>>
Sent: Friday, June 20, 2003 4:04 AM Subject: [mrtg] Per minute option. > A question on the way MRTG calculates averages, I'm not sure what part of the averaging you don't understand, so instead of providing a long dissertation on it, I'll wait till you give a more detailed question > and what happens when per minute option is added. By default, the output will be "per second". Meaning that (if the interval is 5 minutes) the 1st reading will be subtracted by the 2nd reading that is gathered 5 minutes later, and dividing the total by 300. So, if reading 1 is 100, and reading 2 is 1000, then 900/300 = 3 and therefore MRTG will graph 3 whatevers per second. If in the default mode, and monitoring a router or switch port, then it would be 3 octets (bytes) per second average over 5 minutes. Per Minute simply changes the division. In our example, (1000-100)/5 = 180 octets (bytes) per minute average over 5 minutes. > I added the perminute option to graph the errors that come through. Before > adding the per minute option it wouldn't graph anything, now it does and > "appears" correct, but I'm concerned as to why and what exactly the > perminute option does (I'm worried that I'm interpreting the graph > incorrectly.) MRTG will not graph anything less than 1. If in one 5 minute span, you only have 10 errors that's merely 0.03 errors per second. MRTG will log that as a 0. If you use "perminute" then MRTG will graph it as a 2. 10/300 = 0.03 10/5 = 2 >The X scale did not change, and appears exactly the same. I > know its supposed to multiply the values by 60 but what else does it do ? > It seems that by multiplying the values by 60 something else has to change > ass well. The x scale, on the daily graph anyway, is at 5-minute intervals and this will not change regardless of per minute, second or hour, or anyone's "ass". ;-) > Also, after doing so, I added the per minute option to the discards graph, > (which was graphing values, but very small), and after doing so some very > big values started to come through, I did a manual snmpget to see the > change over approximately 6 hours, and I received a value of 22000, which > appears very big, but accepted as correct. So it seems in order (the graph > anyway), but i need some way of verifying that all is well. I think I've answered this question already, but we'll use your example: 22000 discards over 6 hours is an average of 61.1 per minute and 1.0185186 per second. So since you were using the default of "persecond" then your chart would have had nothing but very low numbers. By the way, discards are a bad thing, you should get that fixed! As you can see, the data is always the same, but how its reported is what is changed by the perminute and perhour options. The only break of this rule is when you are dealing with very small numbers that MRTG ignores. HTH, Eric Brander Eric_Mailing_List at rednarb dot com -- Unsubscribe mailto:[EMAIL PROTECTED] Archive http://www.ee.ethz.ch/~slist/mrtg FAQ http://faq.mrtg.org Homepage http://www.mrtg.org WebAdmin http://www.ee.ethz.ch/~slist/lsg2.cgi
