Persistant dsf issue from Smart Array Controller to LSI SAS

[Steven E. Protter]

Shalom,

You can use either legacy or persistent names.

Either will work.

It is a good idea however to migrate to persistent names. That is the new thing and that is what HP wants you to do. So says my crystal ball.

SEP

[Earl_Crowder]

John,

Wipe the partition table off the disk:
#idisk -wR /dev/rdisk/disk9

Remove special files:
#rmsf /dev/rdisk/disk9_p?
#rmsf /dev/disk/disk9_p?
#rmsf /dev/dsk/c0t0d0s?
#rmsf /dev/rdsk/c0t0d0s?

Earl

[John Jimenez]

Thanks SEP. Great I can use either. Yes I will remember to always use persistent.

Thanks Earl.
I just remove the persistent one.
=> #idisk -wR /dev/rdisk/disk9
idisk: Write mode requires description file
idisk version: 1.44
********************** WARNING ***********************
If you continue you will destroy all partition data on this disk.
Do you wish to continue(yes/no)? yes


And will remove those 2 persistant special files and 2 legacy special files.

[John Jimenez]

finished removing from /dev/rdsk and /dev/dsk.

ioscan now looks good and I can not create vg01 with these 5 disks.....

=> #ioscan -m dsf
Persistent DSF Legacy DSF(s)
========================================
/dev/rdisk/disk6 /dev/rdsk/c0t5d0
/dev/rdisk/disk6_p1 /dev/rdsk/c0t5d0s1
/dev/rdisk/disk6_p2 /dev/rdsk/c0t5d0s2
/dev/rdisk/disk6_p3 /dev/rdsk/c0t5d0s3
/dev/rdisk/disk7 /dev/rdsk/c0t1d0
/dev/rdisk/disk8 /dev/rdsk/c0t4d0
/dev/rdisk/disk9 /dev/rdsk/c0t0d0
/dev/rdisk/disk10 /dev/rdsk/c0t3d0
/dev/rdisk/disk11 /dev/rdsk/c0t2d0
/dev/rdisk/disk13 /dev/rdsk/c1t0d0

One last questions. I created a hot spare for my mirrored O/S with EFI. is that hot spare only for my mirrored O/S? If I create vg01 with these remaining 5 disks, will that hot spare replace one of these disks? I assume that it will only hot spare my mirror, but was not 100% sure. If it does not I better not put anything importance on these 5 JBOD's.

[Earl_Crowder]

I see 8 drives in the sasmgr output.

From the sasmgr output, i see that bay 1 and 2 are mirrored, and 3 is the hotspare. bay 4 thru 8 (5 disks) are available to configure however you like. The hotspare isn't visible to the OS as a disk device.

So you should see 6 disk devices (assuming no other storage attached).

[John Jimenez]

Yes that sounds 100% correct, and I guess first guess is correct, and If I make a new vg01 with the remaining 5 that that hot spare will not spare, because the O.S. does not even know its there.
The O.S. does not even know about the spare or the mirror, becuause the E.F.I. created the mirror and spare.... I wonder if crating a mirror and spare was over kill....
I guess I better not put anything important on these other 5 internal drives, because if one fails we have no raid or hot spares.

[Steven E. Protter]

With this set up, going with hardware based raid and spares, saves you CPU cycles.

It is the way to go.

An OS based solution is normally not going to perform as well.

Much as I love mirror/ux

SEP

[John Jimenez]

Okay thats good to know. But the last 8 years I have ran PA RISC servers with "only" two internal drives and the reset on a SAN. just wondering if there was a way to make thse other 5 disks Now I have 8 internal drives, and just hoping that I could make these 5 disks more fault tollerant.

[Earl_Crowder]

You could bind the remaining 5 drives in a raid-5 set, which will give you redundancy, and the hot-spare will be available to replace either a failed mirror from the OS or a failed drive from the raid-5 set.

It will give you capacity of 280GB (roughly) with redundancy.

Performance will be meh on writes, so don't use it for anything performance-critical (oracle database, etc).

[John Jimenez]

Oh.....So that disk will fail over, if one of these drives goes down. Awesome, good to know.