I've noticed the scale changes to accommodate all of the values that need to be graphed. It will use the narrowest range of numbers it can until a larger one is needed. I've seen this myself on some tests, where one graph looks totally all over the map, and I then realize we're talking about a difference of a couple of ms.

----- Original Message ----- From: Michael Beatty [mailto:Michael.Beatty at sherwin.com] Sent: Tuesday, January 08, 2013 08:56 AM To: xymon at xymon.com <xymon at xymon.com> Subject: [Xymon] Ping tests showing different levels percision

I just noticed that different hosts are displaying ping results with very different levels precision. On 3 of the hosts, the test seems to be retuning results that are much more precise than on the other 3. If you look at the graphs and data, data points are frequently changing up and down with an active graph. On the other 3, the data is basically a flat line, almost always showing the same value.

The one thing that that I can see that the more precise graphs are showing is that all of those systems have ping results measured in milisecods and are on the same network, where as the 3 that are not as precise are not. One of them is my local system itself pinging itself and its data points are measured microseconds. The other two are over a satellite and their pings are between 1 and 3 seconds. For example, one of the hosts is almost always showing 40 ms where as another is changing with ever report between about 1000 ms and 3000 ms.

Is this normal, is there some kind of normalization that happens on hosts whos tests don't change much in value?

-- Michael Beatty

Xymon mailing list Xymon at xymon.com http://lists.xymon.com/mailman/listinfo/xymon