glance utility

[skravi17]

Hi

When i am using glance in hp-ux 11.0 always
i saw diskutil 100%? Our configuration is
8 cpu , raid configuration 0+1 and 8gb ram.
Our database is size 80gb. I want to know
glance utility show correct information or
how to check real diskutilization? or any
other way to find out diskutilization.

regards

ravi

[Scott Van Kalken]

Disk Util bar: disk utilization for the busiest currently mounted disk device. The segments indicate the percentages of File system ("F") and Virtual Memory ("V") disk IO activity over the last update interval.


It's more about utilisation and I/O rates than anything else.

If you drill down to the disk screen, you can see which volume or mounted filesystem is getting the most IO and thus spread your datafiles accordingly.

[Michael Tully]

The glance utility actually gives a snapshot of what is going on. Your best best is to make sure that you actually collecting information, so that it can be reviewed later. This is done through the 'extract' command from measureware.

If this is occurring all the time then these are what I would look at first.

What sort of disk is it? Is the buffer cache setting too high, the recommended setting is somewhere between 300-500Mb depending on the amount of RAM you have and the type of disk you have.

What is the timeout value for your disks? e.g.

# pvdisplay /dev/dsk/c7t2d4
--- Physical volumes ---
PV Name /dev/dsk/c7t2d4
VG Name /dev/dscvg
PV Status available
Allocatable yes
VGDA 2
Cur LV 14
PE Size (Mbytes) 4
Total PE 2558
Free PE 771
Allocated PE 1787
Stale PE 0
IO Timeout (Seconds) default
Autoswitch On

The default timeout is 30 seconds, many disk arrays have this value set to 180.

[Animesh Chakraborty]

Hi,
Are you using EMC disk.
You can also try sar -d 1 5

[Fragon]

One more solution:
#sar -u 5 5
and check whether the %wio is too large.

-Live as you wish-
Gerald

[Michael Tully]

From the information provided were you able to determine which type of disk it is.

Operating System disk
Database disk

# sar -d 5 4

should give a reasonable snapshot as well. If there is a filesystem that is on say your /dev/vg00 that should perhaps be elsewhere could indicate something that may need to be done.