I'm trying to graph traffic to our call switches - need a way of charting for patterns outside our 'usual behavior' for that time.
The docs for 1.8.2 note that the timeshifting in trigger allow me to go back in time. :
"""For example, avg(3600,86400) will return the average value for an hour one day ago. This way it is possible to compare average load today with average load of the same time yesterday:
{host:system.cpu.load.avg(3600)}/{host:system.cpu.load.avg(3600,86400)}>2
This expression would fire if average load for the last hour today exceeds average load of the same hour yesterday more than two times. """
This is great, and a huge improvement. I would like to, however, actually compare it to the average value for the previous week. (86400*[1,2,3,4,5,6]) ... any way to do this easily?
The docs for 1.8.2 note that the timeshifting in trigger allow me to go back in time. :
"""For example, avg(3600,86400) will return the average value for an hour one day ago. This way it is possible to compare average load today with average load of the same time yesterday:
{host:system.cpu.load.avg(3600)}/{host:system.cpu.load.avg(3600,86400)}>2
This expression would fire if average load for the last hour today exceeds average load of the same hour yesterday more than two times. """
This is great, and a huge improvement. I would like to, however, actually compare it to the average value for the previous week. (86400*[1,2,3,4,5,6]) ... any way to do this easily?