[net.unix-wizards] Final summary of RA81 responses

phil@sequel.UUCP (05/25/83)

Relay-Version:version B 2.10 5/3/83; site harpo.UUCP
Posting-Version:version B 2.10 5/3/83; site sequel.UUCP
Message-ID:<141@sequel.UUCP>
Date:Wed, 25-May-83 14:26:34 EDT
Organization:Sequel Computer Systems, Portland

Here are my experiences with RA81's to date.  Several people have
requested I post the responses to my inquiry.

4.1 through 4.1c support the  UDA50  with  RA80/RA81  drives.  We
have one each on one UDA50.  Our RA81 has been going offline once
every 3 days for the last month.  It can't be brought back online
without  a  full  power  off  of  the drive.  DEC know about this
problem and had the drives  on  hold  till  a  fix  is  found.  I
understand  the fix is all new boards in both RA81 and RA80 and a
new UDA50 (complete microcode rev).  I recommend  anyone  haveing
the  offline problems beat up DEC to get them upgraded as soon as
it hits the field.  It is an expensive upgrade for  DEC  (if  you
are  on  maintenance)  so it could take a while for the fix to be
completely available.

The performance of the RA81 is quite reasonable even in a 2 drive
environment.  If  this  fix  takes  care of the offline problem I
will be satisfied.  Here is the correspondence  I  have  received
from various people:

--------------------------------------------------------------------------
>From ogcvax!teklabs!arizona!kpno!stevens Fri Apr  8 10:00:49 1983
To: arizona!teklabs!ogcvax!sequel!phil
Date: 7 Apr 1983 5:19-MST
Subject: ra81

Phil,
   I would indeed be interested the continuing saga of your ra81's.
I know from past experience that the quality of DEC field service at
any location makes a lot of difference.  While at Kitt Peak we had the
first TS11's in the area - what a 1 year headache that was.  I have been
in touch with Mt Xinu and was favorably impressed over the phone.
Their prices are a factor of 4 cheaper than the other sources I had
found for binary bsd.
	Richard Stevens

--------------------------------------------------------------------------
>From phil Fri Apr  8 10:46:55 1983
To: ogcvax!teklabs!arizona!kpno!stevens
Subject: ra81
Cc: phil

It indeed is a continuing saga.  We have been getting 5 to 10 random
hard errors on the drive per day that claimed a "SDI" error, type 0353.
Our DEC guy wanted to take us down and run hours of diagnostics but I
asked around (Shannon and Kridle) and found our from Kridle that DEC has
the RA81 on hold because everyone is seeing the same type errors.  He claimed
that DEC has a microcode fix and hardware fix to the RA81 drive to correct
the problem that is in Beta testing now and the Field Service people should
find out about it in about 10 days.  Our Field Service guys don't know anything
about it but I suspect they will soon.  Enjoy..

	Phil Hochstetler
	Sequel Computer Systems, Inc.
	Portland Oregon
	(503) 626-5700

--------------------------------------------------------------------------
>From ogcvax!teklabs!ucbvax!mccallum@nbires.uucp Sun May  1 23:47:13 1983
To: ucbvax!teklabs!ogcvax!sequel!phil
Date: 1 May 83 11:55:46 PDT (Sun)
Subject: RA81's on a VAX
Message-Id: <8305011855.AA21122@UCBVAX.ARPA>
Received: by UCBVAX.ARPA (3.332/3.21),	id AA21122; 1 May 83 11:55:46 PDT (Sun)

We (NBI) have been running UDA50's with RA81's on a VAX for about 2 months.
The only problems we have had are drives going offline for no reason.  This
problem is known to DEC and an ECO is supposed to be coming soon to fix the
problem.  Other than that, they appear to work very well.  We have 2 750's.
Our configurations are:

1)	VAX 11/750
	1 UDA50 with 2 RA81's
	1 RM80	(swap is here)
	2 RK07's
	1 DZ11
	2 ABLE DH's
	1 TE16
	2 MB memory

2)	VAX 11/750
	1 UDA50 with 2 RA81's
	1 SI CMI controller with 2 Fujitsu drives (swap is here)
	1 SI CMI controller with SI tri-density tape drive
	2 ABLE DH's
	2 MB memory

3)	VAX 11/750
	1 UDA50 with 1 RA81	(swap is here)
	1 ABLE DH
	1 TU78
	3 MB memory
	1 DMF 32 (not being used)

VAXen 1 and 2 are in very heavy use with user files on the RA81's
VAX 3 is relativley new and is used for system development.  If your
problem isn't the same as ours, I would be interested in hearing what
it is.
	Doug McCallum
	NBI, Inc.

	UUCP:	{ucbvax,allegra,hao}!nbires!mccallum
	ARPA:	mccallum@utexas-11

PS:	We run 4.1aBSD on VAX 1 and 3, 4.1BSD on VAX 2.  I am in the process
	of bringing 4.1c up on VAX 3.

--------------------------------------------------------------------------
>From unisoft!kridle Mon May  2 02:06:10 1983
To: sequel!phil
Subject: Re:  ra81
Cc: ucbvax!kridle


Last I heard, (about a week ago) the ECO would be released in "two weeks."
I will be talking to someone tomorrow who may have a better idea. I will
ask.

Bob

--------------------------------------------------------------------------
>From unisoft!ucbvax!kridle@UCBCORY Tue May  3 04:06:02 1983
Date: 2 May 83 17:41:27 PDT (Mon)
From: ucbvax!kridle@UCBCORY (Bob Kridle)
Subject: RA81 Update
Message-Id: <8305030041.AA05351@UCBCORY.ARPA>
Received: by UCBCORY.ARPA (3.334/3.17)
	id AA05351; 2 May 83 17:41:27 PDT (Mon)
Received: from UCBCORY.ARPA by UCBVAX.ARPA (3.337/3.26)
	id AA04331; 2 May 83 17:38:04 PDT (Mon)
To: unisoft!sequel!phil


I have been advised that the ECO should be released to production in
one week. This means new drives will be released with the fix. As far
as when drives in the field are fixed, the inside word is "the wheel
that squeaks the loudest gets the grease." In other words, bug the hell
out of your DEC sales person and field service office. Apparently this
problem does not show on all systems and so may NOT be a mandatory
retrofit. It apparently makes the most difference on an HFC-50.

Sorry the news is not better.

Bob

--------------------------------------------------------------------------
>From ogcvax!teklabs!ucbvax!kpno!downey@arizona.uucp Tue May  3 13:12:12 1983
To: arizona!ucbvax!teklabs!ogcvax!sequel!phil
Date: 2 May 1983 9:25-MST
Subject: ra81
Message-Id: <8305030009.AA03814@UCBVAX.ARPA>
Received: by UCBVAX.ARPA (3.337/3.26),	id AA03814; 2 May 83 17:09:42 PDT (Mon)

phil:

we've had two ra81/uda50's on our 750's for 6 months. they are good drives
when used with dec's driver, but not perfect. one exhibits occasional
SDI errors (non-fatal, perhaps one or two a day) that do not interfere with
normal operation. the other is worse in that it will get a few of these in
a row then pop itself off-line, requiring a full reboot to reinit.

dec knows about these problems and is not trying to blame unix. the whole
uda/ra81 line is in engineering hold now to try to fix it. microcode
changes were tried in Phoenix that helped a little but not entirely. I would
go ahead and order them. they work very well generally and dec seems
responsive to the troubles. get the driver from decvax!aps if you don't
have it. (Armando Stettner). there is also a 750 FCO needed that fixes a
problem with BDP transfers that are not aligned on a long-word boundry.

good luck.
elwood downey, kitt peak national observatory, 950 n. cherry ave, tucson, az
               (602) 325-9221
               {{{ucbvax|utah-cs|purdue|mcnc}!arizona}|unc|hao}!kpno!downey

--------------------------------------------------------------------------
>From ogcvax!tektronix!iddic!charliep Wed May  4 08:11:20 1983
To: tektronix!ogcvax!sequel!phil
Subject: DEC RA81 on VAX 11/750's
Date: Wed May  4 07:55:56 1983

We have ordered a similar configuration.  I would be very interested
in seeing whatever results you obtain from the net.  PS.  Would you
care to describe your disk partition and /etc/fstab?

Thanks in advance,
Charles Perkins
Tektronix

PS. Good luck in whatever business you guys are in.  Nobody wants
to tell me what it is y'all are doing.
--------------------------------------------------------------------------
>From zeke Thu May  5 20:33:14 1983
To: phil
Subject: RA81


Electronic News, May 2 1983, page 14, says in part:

Problems with DEC's RA81 and RA60 disk storage systems have led to production
delays and/or stoppages of the large storage systems, the firm said last week.

The 456 megabyte RA81, which entered production late last year, was pulled
from volume manufacture early in April because of a problem related to software
used in the systems disk controller

...

The article is poorly written, but as best as I can understand it, either:
     1. production will resume in June or July
or   2. they didn't say when production will resume

The question is: How soon can we get an Eagle, and scrap the RA81?

--------------------------------------------------------------------------
>From pur-ee!decvax!ucbvax!RWS@MIT-XX Sat May  7 03:41:56 1983
Date:  3 May 1983 2042-EDT
From: Robert W. Scheifler <ucbvax!RWS@MIT-XX>
Subject: RA81's
Message-Id: <8305040038.AA18167@UCBVAX.ARPA>
Received: from MIT-XX (mit-xx.ARPA) by UCBVAX.ARPA (3.337/3.26)
	id AA18167; 3 May 83 17:38:17 PDT (Tue)
To: decvax!pur-ee!sequel!phil@BERKELEY

In case you can't send mail direct to me or would rather phone,
	Bob Scheifler
	617-253-1945

--------------------------------------------------------------------------
>From pur-ee!decvax!ucbvax!wedel@nbires.UUCP Sat May  7 03:41:59 1983
Date: Monday,  2 May 1983 08:08-MDT
From: ucbvax!nbires!wedel
Subject: RA81 experience
Message-Id: <8305021409.AA27883@UCBVAX.ARPA>
Received: by UCBVAX.ARPA (3.337/3.26)
	id AA27883; 2 May 83 07:09:46 PDT (Mon)
To: ucbvax!decvax!pur-ee!sequel!phil
Cc: mccallum

We have had RA81s installed on two VAX 11/750s since early December.
The first system is an all DEC system with an RM80 and RK07s as well
as the RA81s.  The driver obtained from Berkeley had some address problems
but after sorting those out the RA81s came up and ran fine.  The first day
in production each drive dropped off line once and had to be powered down
to be put back on line.  Since that day they have run very reliably.
We saw a few hard errors one day about a month ago but none since.

The second system has an SI UNIBUS disk as well as the RA81s.  This system
has had no end of problems.  Apparently the UDA50 controller and the SI
interface conflict with each other causing reports of hard errors on the
RA81s on the console.  The interesting thing is that the errors aren't
reported back to the program and in fact the data being read and written
is OK!  I've plugged those drives into the other system and they run
without problem.  Our solution is to install a CMI interface for the SI
drive which we should have done anyway.

The bottom line is that we are sufficiently pleased with the UDA50/RA81
combination to have ordered a third VAX configured around these disks.

					Wally Wedel
					NBI Engineering
					(303) 938-2923

--------------------------------------------------------------------------
>From pur-ee!decvax!aps Sat May  7 03:42:03 1983
To: pur-ee!sequel!phil
Subject: RA81 on VAX 11/750

I do not know about ECO's to the RA81's but I will send you the driver
in the form of a tar file in the next message.
	aps.

--------------------------------------------------------------------------
>From ogcvax!allegra!parsec!kolstad Thu May 12 16:40:42 1983
To: allegra!ogcvax!sequel!phil
Mailed: Thu May 12 11:40:32 1983
Subject: hi phil -- re: ra81
Cc:


Please post a summary of your results.....
			Rob

--------------------------------------------------------------------------
>From ogcvax!tektronix!zehntel!sytek!menlo70!hao!pag Wed May 18 09:40:25 1983
To: menlo70!sytek!zehntel!tektronix!ogcvax!sequel!phil
Subject: Re: Summary of RA81 responses
References: <168@sequel.UUCP>
Date: Tue May 17 05:10:33 1983

Phil,
    If you ever find out the exact numbers of the RA81 ECO's and/or
FCO's, please let me know.  Thanks a lot.

--peter gross
menlo70!hao!pag
decvax!brl-bmd!hao!pag
seismo!hao!pag
ucbvax!hplabs!hao!pag

---------------------------------------------------------------------------
>From pur-ee!decvax!cithep!mlb Wed May 18 16:55:45 1983
Date: Fri May 13 04:33 PDT 1983
Regarding your query for info about RA81 flakiness:  We don't have an RA81
(yet, though one is on order), but we do have an UDA50/RA80 on our 750.  We
did experience some weird problems with the RA80 a month or so ago:  Many
disk transfer errors, at first mostly correctable on retry, but things got
progressively worse over time.  DEC arrived, poked around for a bit, decided
that there seemed to be some problems related to the servo motor.  Replaced
motor, the board which drives the servo motor, the entire drive assembly,
and a cable which had allegedly been routed next to something hot and had
gotten partially melted!  (This all took place over a period of several days.)
Put things back together, ran diagnostics, all checked out OK.  Restored UNIX
from tape.  On fsck of third filesystem, errors all over the wazoo.  DEC tries
diagnostics again.  No problems, say diagnostics.  (Surely situations of this
sort aren't new to you ...)  Finally, we prevail on them to build us a VMS
system so that they can run UETP overnight.  (No small problem, since our
machine has NEVER run VMS, since all we have is the RA80, an RL02, and a
Brand X tape drive that VMS doesn't know how to talk to.  They managed to
build it from RL02s.)  Next morning, UETP log on console shows no disk errors.
However, while waiting for DEC to arrive, I decide to look at the VMS error
log.  Happily, it shows 187 errors of the same sort that UNIX has been finding.
(Though all soft; maybe VMS driver is more persistent on retries?)  Now, at
least, DEC can't deny existence of problem.  While they scratch collective
heads trying to decide what to do next, I call Armando Stettner.  He recalls
that they had a similar problem, which was allegedly fixed by an ECO to the
drive's microprocessor board.  He suggests that the ECO may not have been
turned into an FCO, so local field service offices may not know about it.
Meanwhile, after claiming for days that all the boards in the drive were the
most recent rev level, they discover that, lo and behold, there IS a newer
version of the microprocessor board (presumably incorporating the above-
mentioned ECO).  They installed it, and the drive has (so far) lived happily
ever after.  Rev level is (I think) `B1', whereas our DEC field service people
insisted for days that there was nothing newer than `B'.

In any case, since we have an RA81 due to arrive in June, I'd appreciate it if
you could forward to me any relevant information you get about RA81 flakiness.

Thanks!

				Mark Bartelt
				Hospital for Sick Children
				Research Development Corporation
				555 University Avenue
				Toronto, Ontario  M5G 1X8
				416/597-1500 ext 4588
				416/598-5982
				ucbvax!cithep!mlb
				decvax!cithep!mlb
-----------------------------------------------------------------------------

	Phil Hochstetler
	Sequel Computer Systems, Inc.
	Portland, Oregon

	(503) 626-5700
uucp:	ogcvax!sequel!phil
	pur-ee!sequel!phil