max -sample sample_time -allnames -samescale -hidelabels -scale value -ruler
seconds -splitwidthxheight -window number
It can show multiple graphs in one window, and the graphs can be detailing information about several machines.
At the bottom of the window, what graph types and their scale is displayed. These are displayed in the color of the graph.
While XSysStats is running, graphs will be downscaled and upscaled as require. And upscale doubles the graph scale, and only occurs when a new point is beyond the graph bounds.
Downscaling occurs when the running average falls below the scale, and all points on the graph are less than half the present scale. Downscaling halves the scale of the graph.
Upscaling and Downscaling by default do not occur on graphs of cpu time. Scaling can further be controlled by the -min and -max arguments.
A black and white display system will not likely be able to make much sense of multiple graphs in the same window. However, using the -split option can be used to create several windows, each detailing a different statistic.
collisions: number of incoming ethernet collisions since last update
context: number of context switches per second
cpu: percentage of cpu
time being used
disk: number of disk transfers per second
errors: number
of incoming ethernet errors since last update
interrupts: average number of device interrupts, per second
load1, load5, load15: The average number of jobs in the load queue for the last 1, 5, and 15 minutes.
packets: number
of incoming ethernet packets per second
page: page ins since last update
swap: swap ins since last update
Instead of just the type name, a host name can be appended to graph information from that host (see -host below.) Syntax would be -type page@host to graph page information from 'host'.
Note: Unlike perfmeter, collisions, errors, page and swap are not averaged over the sample time. This is a personal preference on my part, to be able to see the activity in these fields more clearly. For this reason, the scale of these graphs may appear different than perfmeter if the sample time is not one second.
Multiple graphs are selected by specifying multiple -type arguments. The graphs are graphed in the order of the -type arguments. This means that the first -type graph given will be drawn first, and graphs specified later will overwrite this graph data. The graph which you want to see most clearly should be the last -type argument.
If both min and max are the same, the graph will never change scale. By default, most graphs have a min value of 2 and a very large max value. The cpu graph is the one exception, both its min and max value is 100.
It seems to me that very seldom will people want to have XSysStats run with all the same resources more than once. If X Defaults were supported, then there would also need to be a way not to use them.
Timeouts of rstat on remote hosts will pretty much stop all the graphs.