🚀 go-pugleaf

rbp@investor.pgh.pa.us (Bob Peirce #305) writes:

>I have an HSD scanner at the end of my SCSI chain.  Since I don't use it
>all the time and it gets pretty hot, I like to turn it off when I am not
>using it.  When I had Rel2.0 I did not seem to have a problem with this.
>Recently I installed 2.1 and now the system will go along for a couple
>of days and then suddenly fill up swap space.  It may be that this is a
>problem with 2.1 which is totally unrelated to the scanner but if I
>leave the scanner on swap doesn't seem to fill up.  Maybe I haven't
>waited long enough.  Anyhow I would appreciate any info on this.
I cannot see any correlation between turning off a SCSI device and
having your /private/vm/swapfile grow.

>Something I haven't tried is to actually disconnect the scanner from the
>CD drive and plug the SCSI terminator in.  That doesn't seem like a good
>thing to do on a running system but maybe it is okay.

Any of what you are doing isn't ok.  Never turn off any SCSI device
or remove it from the chain while the host machine is operating.
One of these days, you will find that your hard drive will go blank
and you won't be able to explain why.  Yes, this kind of thing can
happen by mucking with the SCSI chain while the host machine is
operating (esp. if it's writing to one of the disks!)
Theoretically, it is safe to remove stuff after the machine is
sync'd, halted  and in the Monitor (NeXT> prompt), but it's always better to
actually turn the machine off before you disconnect anything.

>-- 
>Bob Peirce, Pittsburgh, PA         rbp@investor.pgh.pa.us         412-471-5320
>venetia@investor.pgh.pa.us [NeXT Mail]  !uunet.uu.net!pitt!investor!rbp [UUCP]
-- 
David Lemson                                                 (217) 244-1205
University of Illinois NeXT Campus Consultant  / CCSO NeXT Lab System Admin
Internet : lemson@uiuc.edu                 UUCP :...!uiucuxc!uiucux1!lemson 
NeXTMail accepted                                   BITNET : LEMSON@UIUCVMD

Bob Peirce #305 writes
> I have an HSD scanner at the end of my SCSI chain.  Since I don't
> use it all the time ... I like to turn it off...
> Something I haven't tried is to actually disconnect the scanner
> from the CD drive and plug the SCSI terminator in.  That doesn't
> seem like a good thing to do on a running system...

The SCSI bus is actually an active bus.  To function correctly,
all devices on the bus should be powered on, the cables should
have the proper impedence, and the bus should be terminated at
both ends.  A *very* short bus can be run with termination only
at one end (e.g., just the on-board disk, which has a terminator
on it).

If a device doesn't receive power, it might not pass the bus
signals through correctly.  If it's at the end of the chain, the
terminator probably requires power to operate correctly.

For the same reason you shouldn't remove a device from the SCSI
bus while the system is running, your shouldn't power off a
SCSI device while the system is running.  Yes, these things
MIGHT work.  But, do you want to risk your data?  We've seen
instances where an improperly terminated SCSI bus can lead
to data corruption on the internal hard disk....

What to do?  Talk with the scanner vendors and ask them to
provide a "standby" mode, where the SCSI bus receives the
proper power, but the scanning light is shut off. (Caveat:
there might be scanners which do this already, and there
might be SCSI devices which handle bus signals and termination
when powered off.  Check with the device manufacturer to
be certain.)
--
Alan M. Marcum
NeXT Tech Support
amm@NeXT.COM

In article <1992Jul22.114709.9162@investor.pgh.pa.us>  
rbp@investor.pgh.pa.us (Bob Peirce #305) writes:
> I have an HSD scanner at the end of my SCSI chain.  Since I don't use it
> all the time and it gets pretty hot, I like to turn it off when I am not
> using it.  When I had Rel2.0 I did not seem to have a problem with this.
> Recently I installed 2.1 and now the system will go along for a couple
> of days and then suddenly fill up swap space.  It may be that this is a
> problem with 2.1 which is totally unrelated to the scanner but if I
> leave the scanner on swap doesn't seem to fill up.  Maybe I haven't
> waited long enough.  Anyhow I would appreciate any info on this.

I recently started having a similar problem.  I attached a Bernoulli  
Transportable 90 drive to my NeXT and used it for a couple days without  
any trouble.  The trouble started when I tried to eject the disk.  The  
manual says to drag the disk icon to the Recycler and only to press the  
drive's eject button when the disk disappears from the File Viewer.  Mine  
wouldn't.  I couldn't figure if the manual was wrong or what but, being  
impatient, I finally pressed the button.  The disk icon stayed... it  
stayed when I logged out and back in even.  It only went away when I  
turned the NeXT off and then on.  I put the Bernoulli disk back in and  
everything seemed fine.

Then next day I found the hard disk was full.  I had 100MB free the day  
before.  I poked around a little looking for what had grown so huge but  
then thought of the Bernoulli disk.  I ejected it (same problem as before)  
and then turned the machine off and on.  And there was now 100MB free.  I  
put the disk back in and there was a sudden drop to 65MB.  It seemed to  
stay there so I didn't worry too much.  I came in the next day and again  
no space on the hard drive.  Again, I ejected the disk (it worked right  
this time!) and restarted the computer and the 100MB was back.  I put the  
disk back in for a short time but the HD space dropped to about 80MB so I  
ejected it.  All day the space on the HD has steadily dropped.  Now its  
45MB and dropping.

The above posting sounded strangely familiar and so I checked my  
/private/vm/swapfile.  It is 58MB and seems to have grown since the last  
time I checked.  Maybe something else is going on but there seems to be a  
connection between this device and the HD space.  I'm stilling waiting to  
hear from Bernoulli but thought someone else might have a suggestion...

Thanks,
Marianne Guntow, The University of Chicago

marianne@stevie.bsd.uchicago.edu (Marianne Guntow) writes:

>In article <1992Jul22.114709.9162@investor.pgh.pa.us>  
>rbp@investor.pgh.pa.us (Bob Peirce #305) writes:
>> I have an HSD scanner at the end of my SCSI chain.  Since I don't use it
>> all the time and it gets pretty hot, I like to turn it off when I am not
>> using it.  When I had Rel2.0 I did not seem to have a problem with this.
>> Recently I installed 2.1 and now the system will go along for a couple
>> of days and then suddenly fill up swap space.  It may be that this is a
>> problem with 2.1 which is totally unrelated to the scanner but if I
>> leave the scanner on swap doesn't seem to fill up.  Maybe I haven't
>> waited long enough.  Anyhow I would appreciate any info on this.

>I recently started having a similar problem.  I attached a Bernoulli  
>Transportable 90 drive to my NeXT and used it for a couple days without  
>any trouble.  The trouble started when I tried to eject the disk.  The  
>manual says to drag the disk icon to the Recycler and only to press the  
>drive's eject button when the disk disappears from the File Viewer.  Mine  
>wouldn't.  I couldn't figure if the manual was wrong or what but, being  

I have a 90PRO transportable and I had to open the small vented front
panel and throw a two-position switch so the NeXT could see the Bernoulli
as a removable device.  Seems the default position (which suits SGI and 
others) of the switch is as non-removeable.  This panel is about 3/4"
by 1.5" and is about 3" to the left of the blue rubber button. 
BTW, this info was not in the manual but included on a separate slip of 
paper that came with my drive.

Enjoy, 
David Basiji

-- 
David Basiji 
"C neophyte, Bioengineering Grad Student, Internet Personality"

email: basiji@u.washington.edu          
NeXTmail: basiji@larry.pathology.washington.edu

		"Whining is just anger coming out of a very small hole."
            --Stuart Smalley

In <basiji.711941162@milton> basiji@milton.u.washington.edu (David Basiji) writes:
| I have a 90PRO transportable and I had to open the small vented front
| panel and throw a two-position switch so the NeXT could see the Bernoulli
| as a removable device.  Seems the default position (which suits SGI and 
| others) of the switch is as non-removeable.  This panel is about 3/4"

Drives with removable media should be so marked in the inquiry command
(presumably this is what the switch controls).  Other than our older
systems that had a bogus bias against removable media for the boot
device (only), SGI's OS has no problem with devices that say that they
have removable media.
--
Let no one tell me that silence gives consent,  |   Dave Olson
because whoever is silent dissents.             |   Silicon Graphics, Inc.
    Maria Isabel Barreno                        |   olson@sgi.com

In article <1992Jul23.213549.10270@midway.uchicago.edu> marianne@stevie.bsd.uchicago.edu (Marianne Guntow) writes:
>In article <1992Jul22.114709.9162@investor.pgh.pa.us>  
>rbp@investor.pgh.pa.us (Bob Peirce #305) writes:
>
> ...  The trouble started when I tried to eject the disk. ...
>
>Then next day I found the hard disk was full. ...
 
I bet that it is the "/tmp/console.log" which caused your HD flooding.
When some repeated error messages are issued (as after not quite clean eject
of the removable disk) the "console.log" grows to obscene sizes.
The reboot causes to clean the "/tmp" and you get the space back.

There is no "hiwater" mark for console.log.
If this file is removed, it is not created again, so a brutal remedy is to
remove it after login to the system - then you can do your *illegal* operations
without filling the HD up.

	-Z.

olson@anchor.esd.sgi.com (Dave Olson) writes:

>In <basiji.711941162@milton> basiji@milton.u.washington.edu (David Basiji) writes:
>| I have a 90PRO transportable and I had to open the small vented front
>| panel and throw a two-position switch so the NeXT could see the Bernoulli
>| as a removable device.  Seems the default position (which suits SGI and 
>| others) of the switch is as non-removeable.  This panel is about 3/4"

>Drives with removable media should be so marked in the inquiry command
>(presumably this is what the switch controls).  Other than our older
>systems that had a bogus bias against removable media for the boot
>device (only), SGI's OS has no problem with devices that say that they
>have removable media.
>--
>Let no one tell me that silence gives consent,  |   Dave Olson
>because whoever is silent dissents.             |   Silicon Graphics, Inc.
>    Maria Isabel Barreno                        |   olson@sgi.com

Though the Bernoulli literature explains that only early Sun OS's 
had this problem, it implicated all SGI OS versions.  No slight on
SGI intended (by me, anyway).  Perhaps you folks should contact Iomega
corp to set the record straight.  My literature is dated April 1992
and the phone number given is 1-800-456-5522.
-- 
David Basiji 
"C neophyte, Bioengineering Grad Student, Internet Personality"

email: basiji@u.washington.edu          
NeXTmail: basiji@larry.pathology.washington.edu

		"Whining is just anger coming out of a very small hole."
            --Stuart Smalley

In article <4455@rosie.NeXT.COM> amm@NeXT.COM (Alan M. Marcum) writes:
>Bob Peirce #305 writes
>> I have an HSD scanner at the end of my SCSI chain.  Since I don't
>> use it all the time ... I like to turn it off...
>> Something I haven't tried is to actually disconnect the scanner
>> from the CD drive and plug the SCSI terminator in.  That doesn't
>> seem like a good thing to do on a running system...
>
>The SCSI bus is actually an active bus.  To function correctly,
>all devices on the bus should be powered on, the cables should
>have the proper impedence, and the bus should be terminated at
>both ends.  A *very* short bus can be run with termination only
>at one end (e.g., just the on-board disk, which has a terminator
>on it).
>

This is bunk.  The SCSI bus is driven by open-collector devices.  A 
properly behaved SCSI peripheral may be left in the powered down state without
any ill effects.


>If a device doesn't receive power, it might not pass the bus
>signals through correctly.  If it's at the end of the chain, the
>terminator probably requires power to operate correctly.
>

SCSI signals are not passed through a SCSI device.  Rather the bus is
tapped by each device on the bus.  This is why it is called a "bus".
Quite often the terminator power comes from the host, not the drive.  This
is true of *both* the near and far terminators.


>For the same reason you shouldn't remove a device from the SCSI
>bus while the system is running, your shouldn't power off a
>SCSI device while the system is running.  Yes, these things
>MIGHT work.  But, do you want to risk your data?  We've seen
>instances where an improperly terminated SCSI bus can lead
>to data corruption on the internal hard disk....
>

This is good advice if for no other reason than you may choose to power 
up a device at the same time then NeXT decides to swap or something and
who knows how well behaved a device going through power up is.


>What to do?  Talk with the scanner vendors and ask them to
>provide a "standby" mode, where the SCSI bus receives the
>proper power, but the scanning light is shut off. (Caveat:
>there might be scanners which do this already, and there
>might be SCSI devices which handle bus signals and termination
>when powered off.  Check with the device manufacturer to
>be certain.)
>--
>Alan M. Marcum
>NeXT Tech Support
>amm@NeXT.COM

Nothing personal, Alan, but is this the kind of answer I can expect from
NeXT tech support at $225/Question ?


-Brian


-- 
Brian Cuthie
Systemix Software, Inc.
brian@systemix.com

brian@umbc4.umbc.edu (Brian Cuthie) writes:

>This is good advice if for no other reason than you may choose to power 
>up a device at the same time then NeXT decides to swap or something and
>who knows how well behaved a device going through power up is.

>Nothing personal, Alan, but is this the kind of answer I can expect from
>NeXT tech support at $225/Question ?

First of all, this was what I would consider a good answer.  You
said so above yourself, "This is good advice."  If anything, they
should 'err' on the side of safety.   Suggesting that all SCSI
devices be powered up and properly terminated is a safe way of doing
things, IMO.

Second, how much did this cost you?  I'm not saying that it should
have, I'm simply pointing out that the new pricing scheme is already
in place, and yet we still have a few NeXT people posting to the
net, helping out for free there.  Don't flame NeXT mercliessly when
they're the least of the evils of workstation support.

Ever call Sun tech support?  Every looked at how much AnswerLine
support costs?  Ever deal with their DAMN tech people on the other
end?   I had one lady who was actually what I could call mean to me.
I won't go into it here (this isn't alt.bash.sun, although I feel
that such a group should be created).
-- 
David Lemson                                                 (217) 244-1205
University of Illinois NeXT Campus Consultant  / CCSO NeXT Lab System Admin
Internet : lemson@uiuc.edu                 UUCP :...!uiucuxc!uiucux1!lemson 
NeXTMail accepted                                   BITNET : LEMSON@UIUCVMD

In <1992Jul27.025718.22038@umbc3.umbc.edu> brian@umbc4.umbc.edu (Brian Cuthie) writes:
| In article <4455@rosie.NeXT.COM> amm@NeXT.COM (Alan M. Marcum) writes:
| >Bob Peirce #305 writes
| >> I have an HSD scanner at the end of my SCSI chain.  Since I don't
| >> use it all the time ... I like to turn it off...
| >> Something I haven't tried is to actually disconnect the scanner
| >> from the CD drive and plug the SCSI terminator in.  That doesn't
| >> seem like a good thing to do on a running system...
| >
| >The SCSI bus is actually an active bus.  To function correctly,
| >all devices on the bus should be powered on, the cables should
| >have the proper impedence, and the bus should be terminated at
| >both ends.  A *very* short bus can be run with termination only
| >at one end (e.g., just the on-board disk, which has a terminator
| >on it).
| 
| This is bunk.  The SCSI bus is driven by open-collector devices.  A 
| properly behaved SCSI peripheral may be left in the powered down state without
| any ill effects.

Ah, thus speaks someone who hasn't dealt with the real world.  Many
(most?) SCSI devices will pull all the bus lines into the active state
when powered off.  In fact SCSI 2 only has an implementors note (NOT a
requirement!) suggesting that powered off SCSI devices tristate the bus
rather than pulling the signals one way or the other.  In my experience
(and I have looked at a *LOT* of different SCSI devices), almost none
do this yet.  Yes, there are some, but not many, for a number of
reasons (cost is one of them).

| Nothing personal, Alan, but is this the kind of answer I can expect from
| NeXT tech support at $225/Question ?

Alan is right, you are wrong.  Of course, if you know it all anyway,
why would would you want to call for support ;)

PS: Hi Alan; ain't supporting customers grand!
--
Let no one tell me that silence gives consent,  |   Dave Olson
because whoever is silent dissents.             |   Silicon Graphics, Inc.
    Maria Isabel Barreno                        |   olson@sgi.com

In article <nqab7h8@zuni.esd.sgi.com> olson@anchor.esd.sgi.com (Dave Olson) writes:
>In <1992Jul27.025718.22038@umbc3.umbc.edu> brian@umbc4.umbc.edu (Brian Cuthie) writes:
>| In article <4455@rosie.NeXT.COM> amm@NeXT.COM (Alan M. Marcum) writes:
>| >Bob Peirce #305 writes
>| >> I have an HSD scanner at the end of my SCSI chain.  Since I don't
>| >> ...
>| >The SCSI bus is actually an active bus.  To function correctly,
>| >all devices on the bus should be powered on, the cables should
>| >....
>| This is bunk.  The SCSI bus is driven by open-collector devices.  A
>| properly behaved SCSI peripheral may be left in the powered down state without
>| any ill effects.
>
>Ah, thus speaks someone who hasn't dealt with the real world.  Many
>(most?) SCSI devices will pull all the bus lines into the active state
>when powered off.  In fact SCSI 2 only has an implementors note (NOT a

    The problem is that nearly all IC's have protection diodes on their I/O
    lines.  That is, a diode to VDD as part of their internal circuitry.
    When the unit is powered down VDD (the power bus) goes to 0V thus
    forward biasing the input protection diodes and forcing the bus line to
    0-0.7V.  ANY INPUT, INPUT-OUTPUT, OR TRI-STATE LINE HAS THIS.  Many
    output-only pins also exhibit similar behavior depending on the
    process.

    Simply put, you have to use special IC's that do not have the
    aformentioned protection diodes or use the SCSI power pin to keep part
    of your board powered up to keep the input protection diodes out of the
    way.

    And before anyone asks why it is necessary to have input protection
    diodes... without them you blow up the IC when you touch it.  Even
    *with* them it's relatively easy to blow up IC's through static
    discharge.

					    -Matt

>--
>Let no one tell me that silence gives consent,  |   Dave Olson
>because whoever is silent dissents.		 |   Silicon Graphics, Inc.
>    Maria Isabel Barreno			 |   olson@sgi.com

--

    Matthew Dillon	    dillon@Overload.Berkeley.CA.US
    891 Regal Rd.	    uunet.uu.net!overload!dillon
    Berkeley, Ca. 94708     ham: KC6LVW (no mail drop)
    USA