cancel
Showing results for 
Search instead for 
Did you mean: 

RAID problem

SOLVED
Go to solution
Ola Carlstrom
Occasional Contributor

RAID problem

Hello all!
I am trying to configure a RAID5 disk using ORCA. I have 14 146.8 GB disks, a
5302 controller in a DS15 server, running OpenVMS 7.3-1.
I configure it as one big raid with all disks included. I get a 1777.6 GB RAID.
When I try to initialize the RAID in OpenVMS I get a message from system:

%INIT-F-IVADDR, invalid media address

sho dev/full works well
---------------
Disk XXXXXX$DKC0:, device type COMPAQ LOGICAL VOLUME, is online, file-oriented
device, shareable, served to cluster via MSCP Server, error logging is
enabled.

Error count 0 Operations completed 9
Owner process "" Owner UIC [SYSTEM]
Owner process ID 00000000 Dev Prot S:RWPL,O:RWPL,G:R,W
Reference count 0 Default buffer size 512
Total blocks 3727729008 Sectors per track 255
Total cylinders 57328 Tracks per cylinder 255
---------------
Any clue?

Regards, Ola
7 REPLIES
Jan van den Ende
Honored Contributor
Solution

Re: RAID problem

Ola,

AFAIK, there is a limit of 1G blocks/volume.
That gives 0.5 Tb = 512 G max.
Seems you will have to split up into "small" enough sizes....
Maybe someone in engeneering can conform or contradict this?

Jan
Don't rust yours pelled jacker to fine doll missed aches.
Uwe Zessin
Honored Contributor

Re: RAID problem

The controller supports 2 TByte big logical volumes, but last time I checked OpenVMS was limited to 1 TByte (2**31 blocks).
.
Martin P.J. Zinser
Honored Contributor

Re: RAID problem

Hello Ola,

did you ever consider backup or worse restore of this 1.7 TB volume you intend to create? You really might be much better off with a few slightly smaller volumes.

Greetings, Martin
Ola Carlstrom
Occasional Contributor

Re: RAID problem

Hello All,
thank you very much for answers.

Regards, Ola
Uwe Zessin
Honored Contributor

Re: RAID problem

Well, the disks are pretty big. There are two ways to make smaller logical disks:

- build multiple smaller arrays and carve one logical disk from each array. Of course, the RAID overhead will increase and that leaves less data for the user.

- keep one large array and carve multiple logical disks from it. I guess you will loose a little space for additional meta data, but this is nothing in comparison to the first choice. There is one downside with this idea that I have seen on the MSA1000 storrage array that is based on Smart Array technology, too. I have no direct experience with the SA5302A controller, but I beleive it has that limitation, too:
If you create multiple logical disks within an array, then you can only delete them in reverse order.

Anyway. Please let us know how you decided and share your experience with us!
.
Mike Naime
Honored Contributor

Re: RAID problem

UWE is correct, There is a 1TB volume limit for what a VMS system will recognize in a volume. If you require a larger volume than this, you can bind several smaller volumes together at the OS level.

The EVA will make up to a 2TB LUN, but a VMS system will not recognize anything larger than 1023GB. 1024 doesn't work... we tried it. :-)
VMS SAN mechanic
Uwe Zessin
Honored Contributor

Re: RAID problem

Some history for those who like it:

Who would have thought about such big volumes about 10 years ago? It took until VAX/VMS V6.0 that the volume size was limited to 2**24 blocks (8.5 GBytes).

If I recall correctly that was due to a limit in the Volume Control Block - 3*8=24 and the upper byte of the longword was used for something else.

Several people ran into problems when they tried 9.1 GByte disks on VAX/VMS V5.5-2 and filled them up near the limit. There was a wrap-around and data got corrupted or the system crashed.
.