|
|
 |
« on: December 04, 2006, 05:44:41 PM » |
|
the cpu_* agents are collecting values from vmstat every 5 minutes, but these are instantaneous values, they don't represent what was happening during the 5 minutes.
So maybe it can be replaced by:
cat /proc/stat | head -1 | awk '{print $N}'
where N represents the wanted field:
#2 user: normal processes executing in user mode
#3 nice: niced processes executing in user mode
#4 system: processes executing in kernel mode
#5 idle: twiddling thumbs
#6 iowait: waiting for I/O to complete
#7 irq: servicing interrupts
#8 softirq: servicing softirqs
this returns how many hundredth of a second the processor was occupied by user/total/idle... and they are counters incremented on and on.
So, there are 60*100*5 of 1/100s in 5 minutes, so the percentage is (currentValue-previousValue)/30000.
The only problem is that we CAN'T do any computation in the pandora console when defining a module... (I added this in the wish list).
But never mind, the data won't show percentage, but they'll be an image of it.
Then in Pandora console, take the value as generic_data_inc (wich does currentValue-previousValue).
|
