Booting from Slave Drive

[Tim]

I have found that if I have two drives, each in it's own removable rack and
set to "cable select", and each with an instance of Windows XP installed,
then I can boot from either one just choosing whether to power on the master
drive. When the master drive is powered on (using the rack's key switch)
then the system will boot from that drive. When the master drive is powered
off then the system will boot from the slave. It appears to work fine. Is
there any harm in choosing a system this way? I am using an Intel D865PERLK
mobo and two WD12000JB HDs. TIA

 

[Rod Speed]

I have found that if I have two drives, each in it's own removable rack and
set to "cable select", and each with an instance of Windows XP installed, then
I can boot from either one just choosing whether to power on the master drive.
When the master drive is powered on (using the rack's key switch) then the
system will boot from that drive. When the master drive is powered off then
the system will boot from the slave. It appears to work fine. Is there any
harm in choosing a system this way?

Nope. You just have to be a bit careful about how you set it
up in the first place if you are cloning one of the XP installs.

 

[Timothy Daniels]

I have found that if I have two drives, each in it's own removable
rack and set to "cable select", and each with an instance of
Windows XP installed, then I can boot from either one just
choosing whether to power on the master drive. When the
master drive is powered on (using the rack's key switch) then
the system will boot from that drive. When the master drive is powered
off then the system will boot from the slave. It appears to work fine.
Is there any harm in choosing a system this way? I am using an
Intel D865PERLK mobo and two WD12000JB HDs. TIA

And if you want to save power and the wear on the "Slave" drive,
you can leave the "Slave" unpowered when you power up the
"Master". What you are doing is effectively putting one or the
other hard drive at the head of the BIOS's hard drive boot order
by "removing" or "replacing" the "Master" - which the BIOS
puts by default at the head of the boot order. You can change
that boot order by keyboard input to the BIOS - which you might
want to do if you find yourself without hard drives in removable
racks sometime. Another way to do the same thing is to run the
hard drives' power cables through DPST toggle switches, as I
do.

*TimDaniels*

 

[Tim]

And if you want to save power and the wear on the "Slave" drive,
you can leave the "Slave" unpowered when you power up the
"Master".

But does this cause any electrical anomolies - using one drive in the middle
of the cable while having an unpowered second drive attached at the end of
the cable? Is there any way I can test the data integrity of this "slave"
drive when it's used alone as a "master"?

Another way to do the same thing is to run the
hard drives' power cables through DPST toggle switches, as I
do.

Very interesting, are they mounted in a case bay?

 

[Tim]

But does this cause any electrical anomolies - using one drive in the
middle of the cable while having an unpowered second drive attached at the
end of the cable? Is there any way I can test the data integrity of this
"slave" drive when it's used alone as a "master"?

Sorry, I mean in situations where the "slave" is the only active drive.

 

[Timothy Daniels]

But does this cause any electrical anomolies - using one drive in
the middle of the cable while having an unpowered second drive
attached at the end of the cable?

Perhaps. I haven't noticed any, though. The problem of reflections
from an open circuit at the end connector was the reason for the
ATA specs requiring a device being put at the end connector if it's
to be the only device on the cable. As Rod Speed has pointed out
many times in the past, that is in an area that is outside the ATA
specs. But I've tried running an ATA/133 HD as a single device at
the end connector, and it *seemed* to run fine, but who knows if there
were intermittent errors or whether it would work with other HDs?
With a device actually plugged in at the end connector but unpowered,
it could be more benign as far as signal reflections are concerned,
but I don't really know. All that I *do* know is that thousands, if not
tens of thousands, of removable rack users turn off their removable
hard drives or remove them physically, and no one has been reporting
errors occurring as a result.

Is there any way I can test the data integrity of this "slave"
drive when it's used alone as a "master"?

No simple way that *I* know of. If it works satisfactorily for you,
wouldn't that be the ultimate test?

Very interesting, are they mounted in a case bay?

No, too messy. The switches have 1/4" necks, and I used
existing 1/4" square holes in the metal chassis underneath the
plastic fascia. I reach the switches' toggle arms by sticking a
bent paperclip in through one of the air vents. That keeps the
switches away from accidental bumps and brushes which
could be disasterous if the hard drives are running. The
switches are rated for 5 amps at 28vdc, and I use DPDT switches
wired up as DPST, made by GC Electronics, part no. 35-010.
The power connector blocks and pins are hard to find because
there is such a low demand for them, but electronic stores that
cater to hard-core electronic geeks will have them or can
order them for you. Otherwise, you can cannibalize existing
pre-wired power cables. I used the Pan Pacific Ent. clones of
the Molex hard drive power connector, 8981-04P male block (for
female pins), and 8981-04M female block (for male pins). I used
4-conductor cables with #18 gauge stranded wires - one cable
running from the power supply's power connector to the switch,
the other cable running from the power connector to the hard drive
connector. On the power connectors, the colored wires are "hot",
(one at 5 volts, the other at 12 volts) and the black wires are
"returns". I assumed that each black wire "returned" for its adjacent
colored wire. If you know what "DPST" means, you can figure out
the wiring topology. (But be careful. I fried a HD by putting 12 volts
on a 5 volt pin. Fortunately, I had recently cloned it to another HD.)

*TimDaniels*

 

[Tim]

With a device actually plugged in at the end connector but unpowered,
it could be more benign as far as signal reflections are concerned,
but I don't really know. All that I *do* know is that thousands, if
not tens of thousands, of removable rack users turn off their removable
hard drives or remove them physically, and no one has been reporting
errors occurring as a result.

Unless they haven't made the connection between whatever obscure symptom
arises and the underlying cause. I've submitted this question to several
hard
drive and rack manufacturers, I'll pass along any response I may get.

No simple way that *I* know of. If it works satisfactorily for you,
wouldn't that be the ultimate test?

I agree, but who knows how long it will take to see a problem manifest
itself, especially if it's sporadic and unusual. I was hoping for some sort
of exhaustive hard drive test that could flush out any real-time performance
problems. I suppose I could just run a diagnostic utility on the drive at
periodic intervals, checking for accumulating problems.

The switches have 1/4" necks, and I used
existing 1/4" square holes in the metal chassis underneath the
plastic fascia. I reach the switches' toggle arms by sticking a
bent paperclip in through one of the air vents. That keeps the
switches away from accidental bumps and brushes which
could be disasterous if the hard drives are running. The
switches are rated for 5 amps at 28vdc, and I use DPDT switches
wired up as DPST, made by GC Electronics, part no. 35-010.
The power connector blocks and pins are hard to find because
there is such a low demand for them, but electronic stores that
cater to hard-core electronic geeks will have them or can
order them for you. Otherwise, you can cannibalize existing
pre-wired power cables. I used the Pan Pacific Ent. clones of
the Molex hard drive power connector, 8981-04P male block (for
female pins), and 8981-04M female block (for male pins). I used
4-conductor cables with #18 gauge stranded wires - one cable
running from the power supply's power connector to the switch,
the other cable running from the power connector to the hard drive
connector. On the power connectors, the colored wires are "hot",
(one at 5 volts, the other at 12 volts) and the black wires are
"returns". I assumed that each black wire "returned" for its adjacent
colored wire. If you know what "DPST" means, you can figure out
the wiring topology.

Great info though, I never considered electrical switches to achive the same
result. I assume then that most racks only disconnect the power connector
with their key switches, and don't even affect the IDE connection.

 

[Timothy Daniels]

I've submitted this question to several hard drive and rack
manufacturers, I'll pass along any response I may get.

That would be interesting, perhaps even amusing.

I assume then that most racks only disconnect the power connector
with their key switches, and don't even affect the IDE connection.

Yup. Just follow the wires that lead from the removable rack's
key switch, and you'll see that there are only enough to control
the power.

*TimDaniels*

 

[Tim]

That would be interesting, perhaps even amusing.

How so?

 

[Timothy Daniels]

The replies from the manufacturers would have been
filtered (invisibly to us) through layers of corporate
legal policies that are designed to limit their liability
in case of a claim for loss due to false advertizing or
some other mal-description of their product. What to
expect is some vague claims of working according
to advertized specifications with a slant toward
"well-known reliability", "thousands of satisfied users",
etc. What you will get will be corporate Good-Speak
without much substance, e.g. no results of lab testing
with well-defined parameters and many units under
test.

*TimDaniels*

 

[Timothy Daniels]

How so?

By the way, it appears that all (or nearly all on the basis of
installed units) of the removable tray/rack makers are in China.
Unless you know who the manufacturers are and you can
speak Cantonese, all you will get will be the U.S. distributors
of their rebadged equipment - who are in the business of
distributing and selling, not designing or testing. It will be
like asking the saleman at Radio Shack if the stuff he sells
is any good.

*TimDaniels*

 

[Rod Speed]

Tim wrote

The replies from the manufacturers would have been
filtered (invisibly to us) through layers of corporate
legal policies that are designed to limit their liability
in case of a claim for loss due to false advertizing or
some other mal-description of their product. What to
expect is some vague claims of working according
to advertized specifications with a slant toward
"well-known reliability", "thousands of satisfied users",
etc. What you will get will be corporate Good-Speak
without much substance, e.g. no results of lab testing
with well-defined parameters and many units under test.

More likely they'd just say what I said, that that flouts the
ATA standard and that while it may work, all bets are off.

Thats what they should be saying, even if they dont.

 

[Timothy Daniels]

More likely they'd just say what I said, that that flouts the
ATA standard and that while it may work, all bets are off.

Thats what they should be saying, even if they dont.

Yes, that's what the disk drive makers will probably say,
but the OP also intended to query makers of removable
hard drive trays. His questions would basically ask if
their products reduce in any way the data integrity of
hard drive-IDE channel communication.

*TimDaniels*

 

[Rod Speed]

Rod Speed wrote

Yes, that's what the disk drive makers will probably say,
but the OP also intended to query makers of removable
hard drive trays.

As you pointed out, he's unlikely to get a useful response from them.

And they're unlikely to even have any
legals bothering to scrutenise anything.

His questions would basically ask if their products reduce in any way the
data integrity of hard drive-IDE channel communication.

And they are unlikely to even understand that question.

Let alone have anything useful to say in response to it.

 

[Timothy Daniels]

As you pointed out, he's unlikely to get a useful response from them.

Agreed. Especially if he can't speak Chinese.

And they're unlikely to even have any
legals bothering to scrutenise anything.

Agreed, again. But I referred to *policy* - such as an American
hard drive manufacturer might have laid out for its "Customer
Satisfaction" department's responses to the public inquiries.
Basically, they tell Customer Service (which fields all questions
from the public) not to make any claims or statements that
contradict or even extend or elaborate on anything that they have
already cleared for release - especially not in writing. What one
then gets is just Nice Talk that is worth nothing, technically.

And they are unlikely to even understand that question.

Let alone have anything useful to say in response to it.

Agreed once more.

*TimDaniels*