Operating System - Tru64 Unix
1751709 Members
4982 Online
108781 Solutions
New Discussion юеВ

what different between them?

 
SOLVED
Go to solution
Rambo_1
Regular Advisor

what different between them?

#mkfdmn /dev/rz2c domain-1
#mkfdmn /dev/disk/dsk2c domain-1


what different between them ?
4 REPLIES 4
Hein van den Heuvel
Honored Contributor
Solution

Re: what different between them?

Tru64 V4 style versus V5+

In V4 it was a direct mapping. A formula based on Bus-Lun-Target. Very annoying, as that will change when an adapter is added, or a disk is stuck in a different slot.

V5 attempts to create a persistant mapping based on wwid. This is most critical for clusters. If a given disk is visable trough a shared bus #1 on system A, but that same shared bus is #3 on system C in the cluster, then that disk should still have the same name.

The rzNx mapping is hard and sparse. The dskNx mapping is soft and dense, starting at 0 for the first disk found. The tool dsfmgr can be used to change/rearrange dsk devices and so on.

hth,
Hein.
Michael Schulte zur Sur
Honored Contributor

Re: what different between them?

Hi,

another thing. While it is an advantage to trace a disk by its wwid in cluster and assign the same device to it, it makes it more difficult to replace a disk, because it gets a new device.

Also in 4.0X all disk/tape devices were in /dev, now they are in /device/dsk, /devices/rdsk, /devices/tape and /devices/ntape.
/dev is just a logical link now.

Michael
Hein van den Heuvel
Honored Contributor

Re: what different between them?

Michael wrote: "While it is an advantage to trace a disk by its wwid in cluster and assign the same device to it, it makes it more difficult to replace a disk, because it gets a new device."

Correct. That can be a drag at first.
That's why I mentioned dsfmgr. I should have indicated that. In an example... say you have dsk0, dsk1, dsk2 as devices. Now dsk1 dies and gets physically replaced. The new disk is likely to be called dsk4. This is no problem if all you have to do is create a new volume with the old name on that disk. And it is not really a big problem if you have it listed once in fstab. It can be annoying however if there are a bunch of scripts with say lsm commands dealing with the dsk name (you probably should not have those scripts but hey!).
So what you do is: dsfmgr -e dsk4 dsk1
This will give the old name back, and 'reserves' dsk4 as a name just in case that old disk comes back alive.

In my weird an wack benchmark world, with entire EVA's coming and going on a whim, I carried the renaming a step furher.
We actually 'reseved' (in our minds) dsk0, dsk1... for basic system drives. dsk10, dsk11,.. for Oracle Redo, dsk20, dsk21 for Oracle data. Dsk90, dsk91... for our backups.
So when watching 'monitor' or similar we woudl instantly know what kind of IO was happening! We had a superscript that read a private datafile mapping those dsk names onto user selectable EVA/HSG container names. It would then go out and 'discover' the drives on the SAN and map as desired.
It woudl then generated the LSM script and lauch a restore. Good clean fun! (both to write and to use).

CHeers,
Hein.
Ralf Puchner
Honored Contributor

Re: what different between them?

the device mechanism is well documentation in the admin guide. Give it a try....
Help() { FirstReadManual(urgently); Go_to_it;; }