I’m running PostrgreSQL on a Windows VM in my office, and I’m averaging about 324% usage on the C: disk.
I thought the scale on that was 0 to 100.
Any clues?
Microsoft explains the “feature.”
Heh.
I just added the counters MS suggested. [1]
And to thwart my efforts, the system stopped bogging my disk down and now my disk queue length is zero… % disk time is averaging 15%…
I just opened that box with the cat in, apparently…
[1] Counters:
* Avg. Disk sec/Read
* Avg. Disk sec/Write
* Avg. Disk sec/Transfer
Hmmm.
Lemme petition the boss for funding on a half-dozen Raptors to put in a RAID array hanging off of the back of my system… I’ll let you guys know how that goes…
Whats your disk que length at? That’s a pretty reliable count on how busy the disk is.
It was bouncing around between 2 and 7 most of the time.
The user interface was not very usable when it got above 3.
This was an install of openNMS monitoring just over 1000 servers.
It was set to check each server on both ICMP and HTTP every 5 minutes, which worked out to 24000 queries per hour.
As an experiment, I backed the rate down to one checkup of both ICMP and HTTP per hour.
Disk swamping has ceased. I’m considering scaling it back to just HTTP, since IIS is the critical app for these servers anyway. I’ll experiment with ramping the monitoring rate back up 'til I find the saturation point as it were… or my boss gets me the half-dozen Raptors.