Raid Array 230/Plus & Alpha 1200

[dave kone]

I have this card and 3 drives in a raid 5 set on my 1200 DRA0:

I'd like put in 3 more hard drives into my Alpha and create an additional separate raid set DRA1:

How do I configure this in VMS? v. 7.1-1H2
Is there someone in the Washington DC area that could help me with this?

thanks

[Verne Britton]

I believe this card is also called a KZPAC and has one or three (depending on the model) UltraSCSI ports.

You need the config program on a floppy; you shut down the Alpha and switch to the ARC console, then run the program from the floppy. I always use a VGA monitor to do it on; I think it can be done as well using a serial console.

Once in the utility, you define disks and raid sets. The configuration is stored on the card (in NVRAM ?) and all the Alpha and VMS will see are the DRAxxx devices.

The VGA (GUI) utility is called RA200RCU.EXE; the serial version is RA200SRL.EXE.

An old note I have says the Firmware CD contains these utilities in some subdirectory, as well as (again an old note) being at ftp://ftp.digital.com/pub/Digital/Alpha/firmware/v6.6/utility/swxcrmgr/

I hope this will get you started ... good luck !

[dave kone]

Thanks! Now how do you get to the ARC console?

I think then I'm set. I've been reading the 203/plus manuals and came close to what your reply said.

David

[Verne Britton]

shut down VMS normally, specifying it to not automatically reboot :-)

Then at the console prompt (>>> often), enter the three letter command ARC and press return (if memory serves).

This will cause most Alphas to switch to the ARC console (mostly used by Windows NT) and eventually you will get a GUI menu. Choose RUN MAINT. PROGRAM (or something similar) from the main menu or the utility menu (sorry I don't have an Alpha handy that I can shut down at the moment).

You may or may not have to enter A: as a prefix to the RA200RCU.EXE program name.

Once you are all finished with the RAID utility and have exited back to the main ARC menu, just power cycle your box (or press the INIT button) and it should restart and come back into the normal SRM (VMS) console environment.

[dave kone]

Thank you, I got right into the arc console and my raid configuration utility. Soon to have another raid set will I.

[Paul Janssen]

The next thing you will want to do is to monitor the raid set without going to the floppy (and shut down vms) (well I presume you would want to do this ;-)).

There are two ways and both work.
Either your machine has Motif and then you install the motif application. Or you install the agent on vms and the PC application on your desktop/notebook.

I do not remember just how these files are named, but they are/were in the website of the RA200 controller. You would find stuff in the docs.

One tip for using the PC version. You will need to reach the vms machine by name. So make an entry in your host file or in your dns server. Also the PC must be known in tcpip or ucx.

Good luck,

Paul Janssen

[dave kone]

I see the drive when I do a show dev d
its dra1:

Its on-line but not mounted. How do I mount this drive so its ready to go?

thanks
David - who knows little about vms

[Uwe Zessin]

Dave,
you first need to put a file system on a newly created RAID set. E.g.:

$ initialize DRA1: DATA01

Then you need to make the volume available to the operating system system:

$ mount /system DRA1: DATA01

The mount is not persistent. All volumes except for the system disk (which is implicitly mounted during the system boot) need to be mounted as part of the system startup. The usual location is the file SYS$MANAGER:SYSTARTUP_VMS.COM

[dave kone]

thanks, I have a lot more to learn I guess before I can start using it. Its mounted but I am unable to copy file to the new raid set dra1:

[Ian Miller.]

what error do you get when wrting to the new mounted DRA device?

[dave kone]

when I try to copy files this is what I get:



%COPY-E-OPENOUT, error opening DRA1:[DAVE]*.*; as output
-RMS-E-DNF, directory not found
-SYSTEM-W-NOSUCHFILE, no such file

[Hein van den Heuvel]

Well Dave, VMS means every single word of that error message. Apparently there is no directory [DAVE] on DRA1:.. and there wouldn't be if you freshly initialized that disk.
You can create the directory with:
$CREATE/DIR DRA1:[DAVE]

Depending on the exact task you want to accomplish, you may also want to check out using BACKUP to copy instead of COPY. The Backup tool will create top directories if need be.

It is probably too late now, but IMHO a 3-member raid-5 set is just silly. You loose 33% of you space, and have a tremendous write performance overhead (2 read + 2 writes for each write). Instead I would suggest raid 0+1 where the 'wasted' space is 50%, the write overhead is just a second write IO (2 disk writes for each application write), and two disks to choose from for each read (verus just 1 for raid-5).

If you have that option I would encourage you growing the raid set instead of adding an other inefficient one.

Good luck,
Hein.

[dave kone]

I know 3 drives is nuts, but this controller has a size limitation of 3 nine gig drives per logical drive. Its crazy!!

I might go mirrored all the way down the road.

[Ian Miller.]

I think the limitation is 32Gb per logical drive and I agree that with current disk sizes it is stilly but it's an old raid controller.

[dave kone]

Even back then (5.5 years ago) when we got the sytem 18/9/4 gig disks were around. I still think the limitation is low especially for a server with 7 drive slots @ 4.3 gigs each minimum.

[Jan van den Ende]

Dave,

I immediately believe you have the system just 5.5 years, but the last time I played with such card was at a customer site where I was until 10.5 years ago.
And just 8.5 years ago, when building our current cluster, I remember the excitement that, just in time for us, there became available those GIANT 4 G drives... and that so shortly after those big 2 G ones came out.

At the time those cards were designed, they were truely OVERdimensioned.

(Oh boy, do such stories make one feel old!)

fwiw,

Proost.

Have one on me.

jpe

[grant dittmer]

i also have the same controller in my alpha (and yes it is quite old). i am wondering if there is a replacement raid controller that i could replace my old controller with, that would allow my drive designations to remain DRAx. if not, what would you recommend as a new controller that has 3 scsi ports and allows more capacity (currently i have 3, 9 gig drives in 3 separate raid 5 arrays). if i can not keep the drive desination DRAx, should i just use a system logical (or symbol)???

thank you in advance

grant...

[Jan van den Ende]

Grant,

on the hardware part someone else will have to give the answer.

On the DRAxn: naming part, I strongly hold a hard principle (and just this question shows why):

Any physical drive name (such as DRA0:) should occur at exactly ONE place in the entire system/cluster, and then, all of them together in just ONE file:
the file in which /EXEC mode, cluster- or systemwide logicals for each device are defined.
Converting to shadowing, upgrading hardware, whatever change in drive hardware naming now are administered in ONE file only.
Second, there should be only ONE file, in which those devices are referenced: the file in which /EXEC mode Concealed Devices are derived from the devices.

(@ John Gillings: I am aware that you prefer to directly couple the Concealed Devices to hardware names, as had to be done before the restrain was lifted that Concealed Devices needed to be /TERMINAL. Easily to be combined with the rest of this principle)

Now, you only have to ENFORCE that ANY access to any disk content is VIA those concealed devices.

It really is the basis of transportable software and data, and as such is THE conditio sine qua non of good programming practise.

I only regret that this principle is not more often, and more strongly, brought over the footlights.

But, if you find you are stuck with hardware names spread all over your system, then it IS possible to just redefine those old names as pointing to the new location. But now YOU are reponsible that NONE of those redefs EVER leads to conflicts, and that ANY data that once got allocated to the same device, STAYS together, irrespective of its further having no relation whatsoever.

fwiw.

Proost.

Have one on me.

jpe

[Ian Miller.]

Grant,
welcome to the itrc vms forum. You should start a thread of your own so you can award points. See
http://forums1.itrc.hp.com/service/forums/helptips.do?#33

What model of AlphaServer is it?
There has been some more recent raid cards.

[Ian Miller.]

There are
http://h18002.www1.hp.com/alphaserver/products/storage/sa5300a/index.html
http://h18002.www1.hp.com/alphaserver/products/storage/kzpcc.html
http://h18002.www1.hp.com/alphaserver/products/storage/kzpcc_ac.html

however non of them list the AS1200 as a supported system.

[Aaron Lewis_1]

Grant, I don't think there is a replacement for a 1200, however, depending on the model of card, you might be able to upgrade the amount of Cache on the controller. The 230/Plus cards were released with as little as 1 or 2MB, don't remeber for sure, but I do know they maxed out at 8MB.

[grant dittmer]

thanks for the info. currently i have a ds20 and an es40. yea, i know i should have something defined in a com file, but as this application is about 15 years old (migrated from and ancient VAX), and has had about 6 systems admins, my only real recourse is to define it now and search replace away all the previous sins. but it would have been nice to simply replace my scsi card with another one that kept the DRAx intact... oh well...

thank you for all the advice...

grant

[Ian Miller.]

Look at the links I mentioned - for a ES40 or a DS20 then they are supported for
SA530A

kzpcc (for ES40, DS20e)

[Aaron Lewis_1]

Grant, for the ds20 / es40, you could replace the cards, and use logicals to point any DR* accesses to the new drive names.

[Lawrence Czlapinski]

Dave: The DRA designation is for the raid controller card not a SCSI controller.
Yes, it would be nice to have a higher total GB limit than 32 GB for a logical drive on the Raid Array 230/Plus. We have had problems with Raid controllers not recognizing 9.1 GB drives with old firmware revisions.
Lawrence