2011-05-20 10:23 AM
One of the counters returned by a "stats show lun" is "queue_depth_lun". The description would lead me to believe it's exactly what it claims it is: The queue depth of the LUN. However, the data it returns doesn't make any sense. In my case, it returns a very high number (in the hundres of millions). My belief is that this is a raw statistic, and needs to have it's previous value subtracted from it to get it's real value. Despite doing this, the value still doesn't make any sense. For instance, I get these values:
(These are the actual values returned by the "stats" command
minute 1: 1593764771
(These are "delta" values I get when I subtract the raw value from the previous minute's raw value)
minute 1: 3176.0000
minute 2: 3164.0000
minute 3: 3230.0000
minute 4: 47690.0000
minute 5: 57797.0000
minute 6: 60833.0000
The odd thing is the first 3 data points are from when I'm running heavy I/O testing to the LUN, and the second 3 are when it's idle. Clearly, it doesn't make sense for the queue depth to be higher when idle.
Can anyone tell me if there is another place I should be looking to get queue depth for a LUN? If not, is this the correct counter? If so, what am I doing wrong?
I'm running ONTAP 8.0.1P2 on a new HA pair of 6280s.
Solved! SEE THE SOLUTION
2011-07-05 11:53 AM
You can get LUN "Queue Length" using this command :
lun stats -o -i 1 -c 1
-o = Extra stats
-i = Interval
-c = Count
Read Write Other QFull Read Write Average Queue Partner Lun
Ops Ops Ops kB kB Latency Length Ops kB