Win2K "Disk Admistrator" related queries

NTDEV
1

Following are some questions related to disk subsystem
and are related to practical observations on the test
setup. In my setup, I connect a JBOD (containing
SEAGATE/IBM disks) to Emulex HBAs in Win2k server,
through a SAN switch. I have installed the Emulex port
driver on my server.

(A) When I connect/zone a disk to host and run Disk
Administrator, *sometimes*, it takes very long time to
show up the new disk. Mostly it waits after giving the
print: “Connecting to Logical Disk Manager Service”.
Why such a long delay in identifying a new disk ?

(B) If I remove/unzone a disk from server and run Disk
Administrator, it takes very long time to declare that
device is no more connected to the server. Why such a
long delay ? However, (mostly) subsequent invokation
of Disk Admistrator completes quickly.

(C) Though disk administrator is very slow (and at
times, unpredictable), while handling addition/removal
of disks, Emulex configuration/monitoring utility
elxcfg is very quick to identify any changes in device
configuration. My question is that if HBA can find the
change in device configuration so quickly, why disk
administrator takes so long ?

(D) Is there an API, which a priviledged program can
use, to tell OS to scan for any newly added or removed
disk devices ? I guess “disk administrator” must be
using something like that, but I don’t know what that
is ?

Your response will be deeply appreciated.

Thanks,
Anuj

Do You Yahoo!?
Sign up for SBC Yahoo! Dial - First Month Free
http://sbc.yahoo.com

2

(A) When I connect/zone a disk to host and run Disk
Administrator, *sometimes*, it takes very long time to
show up the new disk. Mostly it waits after giving the
print: “Connecting to Logical Disk Manager Service”.
Why such a long delay in identifying a new disk ?

Are u adding some disks and removing some at the same time?

(B) If I remove/unzone a disk from server and run Disk
Administrator, it takes very long time to declare that
device is no more connected to the server. Why such a
long delay ? However, (mostly) subsequent invokation
of Disk Admistrator completes quickly.

If the disk administrator thinks the disk is still there( i.e. the miniport has not reported the disk gone to the port driver ) , it scans the disk once u open it. And until the miniport declares the device as disappeared , disk administrator will wait. Once Disk administrator knows the disk is gone( after a timeout or maybe multiple timeouts the Emulex miniport might declare the disk as gone ), subsequent invocations of disk admin will not scan that disk anymore.

(C) Though disk administrator is very slow (and at
times, unpredictable), while handling addition/removal
of disks, Emulex configuration/monitoring utility
elxcfg is very quick to identify any changes in device
configuration. My question is that if HBA can find the
change in device configuration so quickly, why disk
administrator takes so long ?

Though the miniport knows about the disk going away( Link went down, target disappeared…whatever ), the OS and hence Disk admin might now know about it if the miniport does not have support for dynamic reconfig, or if it does not report the disk as gone.

(D) Is there an API, which a priviledged program can
use, to tell OS to scan for any newly added or removed
disk devices ? I guess “disk administrator” must be
using something like that, but I don’t know what that
is ?

Yes, the miniport can ask for a bus rescan. I think Disk administrator, when opened, scans the disks it knows about already. So to initiate a full rescan, either the miniport has to ask for it, or a manual rescan can be done using disk administrator.

I think those are the answers. Guys, correct me if I am wrong!

Mukul

3

> (D) Is there an API, which a priviledged program can

use, to tell OS to scan for any newly added or removed
disk devices ?

CM_Reenumerate_DevNode. Or Device Manager’s “Scan For Hardware
Changes”.

Max