Backup Central

On Saturday 06 March 2010, Jon LaBadie wrote:


Because the scsi buss is a transmission line, and demands a decent VSWR,
there are 2 things to remember when dealing with scsi.

1. A list of pre-requisites that must be met if it is to work:

1.a: termination power, it has to come from someplace, preferably the host
controller.

1.b: terminations can only be at the ends of the cable, one and exactly one
on each end. Leaving a 6" pigtail hanging out because you removed the drive
on the end of the cable, and which co-incidentally also was the terminating
device is a more common mistake than one would think.

1.c: when supplying scsi term power, there needs to be an isolation diode
between the supply and the term power line in the cable, pointed so that the
voltage can get to the cable ok, but cannot be fed back into an accessory
device after it has powered the terminators on that device.

1.d: Because this diode has a voltage drop, and the original idiots (yes,
that _is_ the right terminology) chose the resistor values based on std
values AND a 5 volt supply, so a common silicon diode virtually cannot be
used if the scsi bus is to be error free. The .65 to .7 volts of drop in an
Si diode reduces the logic one voltages noise margin by about 2/3rds. from
the target of 3.0 volts, but usually about 2.85 due to the resistors chosen,
throw in the .7 loss of the diode divided by the resistor ratio's (330 to
ground, 220 to the 5 volt line, and your logic 1 noise margin is whats left
after you subtract the 2.4 volts of a guaranteed logic 1. At 2.7 volts it
might work if the resistor tolerance are spot on. At 2.55 volts, you have
only a 150 millivolt noise margin and you may as well hold a seance, its
dead, Jim. Therefore this diode must be either a power germanium or a
schotkey rated for the current in order to get a decently low forward voltage
drop. Otherwise find a 6 volt supply for termination power. TBT, when the
psu voltage in the host machine is starting to sag in its old age, and is
already 200 millivolts low, I've been known to go get a 6 volt dc wall wart
from the shack and use that. It Just Works(TM). FWIW, I have yet to see a
Ge or Schotkey diode used on any of the major scsi vendors cards, so don't
assume that just because it says Adaptec on it, it can't be wrong. And it is
precisely those cards I've had to wall wart power several times.

If you've done all that & it still doesn't work, have a tech with a 100+ mhz
scope look at it to see why the signal echos (aka VSWR) are so bad. Maybe
you've got one bad resistor in the termination packs, so check all active
lines, 21 of them in a 50 wire scsi-II cable IIRC. Make sure he understands
exactly what the term 'VSWR' means, a great many don't, somehow thinking it
only applies to continuous wave radio/tv transmitters. VSWR is VSWR whether
its digital, or CW, its still VSWR.

2: Alternatively, you can advertise for a virgin to be sacrificed. It won't
help of course. And its a terrible waste considering how hard it is to find
one these days.

--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)

Reality is bad enough, why should I tell the truth?
-- Patrick Sky

Well, thanks for the long explanation. I doubt i will be able to test the setup as much as you have described but it's useful information. I've tried a couple of different cables, unplugged and plugged back in, all that jazz. These are the cables i've used previously without any trouble.

I'm more inclined then to think it is the drive. but who knows..

On Monday 08 March 2010, rory_f wrote:
Gene Heskett wrote:
[huge snip]
Well, thanks for the long explanation. I doubt i will be able to test the
setup as much as you have described but it's useful information. I've
tried a couple of different cables, unplugged and plugged back in, all
that jazz. These are the cables i've used previously without any trouble.

I'm more inclined then to think it is the drive. but who knows..

In your case, two things. Is it the last drive on the cable?, and is it
terminated properly? And I would certainly check the 5 volt line of the host
computer to see if its sagging in its old age. 4.85 volts won't cause the
host to reset itself, but it does put the noise margin of a scsi buss in
danger. You need every millivolt you can get there.

+----------------------------------------------------------------------

|This was sent by rory < at > mrxfx.com via Backup Central.
|Forward SPAM to abuse < at > backupcentral.com.

+----------------------------------------------------------------------



--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)

Accept people for what they are -- completely unacceptable.

On Monday 08 March 2010, Charles Curley wrote:
On Sat, 06 Mar 2010 18:16:09 -0500

Gene Heskett <gene.heskett < at > verizon.net> wrote:
Refresher course 101 on scsi here folks, bear with me.

Very nice write-up, Gene. Thanks.

This is a good candidate for the Wiki.

Be my guest if you have write privileges.

--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)

How should I know if it works? That's what beta testers are for. I
only coded it.
-- Attributed to Linus Torvalds, somewhere in a posting

[quote="Gene Heskett"]On Monday 08 March 2010, rory_f wrote:
Gene Heskett wrote:

In your case, two things. Is it the last drive on the cable?, and is it
terminated properly? And I would certainly check the 5 volt line of the host
computer to see if its sagging in its old age. 4.85 volts won't cause the
host to reset itself, but it does put the noise margin of a scsi buss in
danger. You need every millivolt you can get there.



It is setup as such:

Overland tape library, drive inside it. Drive connects to library via one cable, then another cable connects to scsi card in linux machine - It has been setup like this for a couple of years now.

It could very well be the cable degrading, or the card in the linux machine. That is something I am yet to check. Our other drive is being replaced today (For a NEW ibm lto 3 drive) so hopefully i can test that one - if i get th same issues ,we can say it is something else other than the drives.

I'll let you all know what i find.

Thanks.

On Mon, Mar 8, 2010 at 9:55 AM, Gene Heskett <gene.heskett < at > verizon.net> wrote:
Be my guest if you have write privileges.

It's a wiki - everyone has write privileges (including, to my dismay,
spammers .. but I try to keep up with them).

Please do add this to the wiki!

Dustin

--
Open Source Storage Engineer
http://www.zmanda.com

On Monday 08 March 2010, rory_f wrote:
[quote="Gene Heskett"]On Monday 08 March 2010, rory_f wrote:
Gene Heskett wrote:

In your case, two things. Is it the last drive on the cable?, and is it
terminated properly? And I would certainly check the 5 volt line of the
host computer to see if its sagging in its old age. 4.85 volts won't
cause the host to reset itself, but it does put the noise margin of a
scsi buss in danger. You need every millivolt you can get there.

It is setup as such:

Overland tape library, drive inside it. Drive connects to library via one
cable, then another cable connects to scsi card in linux machine - It has
been setup like this for a couple of years now.

It could very well be the cable degrading, or the card in the linux
machine. That is something I am yet to check. Our other drive is being
replaced today (For a NEW ibm lto 3 drive) so hopefully i can test that
one - if i get th same issues ,we can say it is something else other than
the drives.

I'll let you all know what i find.

Thanks.

Please do. It could be something I missed, or it could be a brand new
phenomenon to file away in my trivia file.

--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)

You can't have your cake and let your neighbor eat it too.
-- Ayn Rand

On Monday 08 March 2010, Dustin J. Mitchell wrote:
On Mon, Mar 8, 2010 at 9:55 AM, Gene Heskett <gene.heskett < at > verizon.net>
wrote:
Be my guest if you have write privileges.

It's a wiki - everyone has write privileges (including, to my dismay,
spammers .. but I try to keep up with them).

Please do add this to the wiki!

Dustin

I would Dustin, but 2 questions: Where do I put in, or can I make a new
category?

And 2, I don't see any edit buttons anyplace, security by obscurity I believe
they call that. So that seems to be a show stopper.

--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)

<netgod> Feanor: u have no idea of the depth of the stupidty of american law

On Monday 08 March 2010, Dustin J. Mitchell wrote:
On Mon, Mar 8, 2010 at 9:55 AM, Gene Heskett <gene.heskett < at > verizon.net>
wrote:
Be my guest if you have write privileges.

It's a wiki - everyone has write privileges (including, to my dismay,
spammers .. but I try to keep up with them).

Please do add this to the wiki!

Dustin

I would Dustin, but 2 questions: Where do I put in, or can I make a new
category?

And 2, I don't see any edit buttons anyplace, security by obscurity I believe
they call that. So that seems to be a show stopper.

--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)

<netgod> Feanor: u have no idea of the depth of the stupidty of american law

I started an amtapetype earlier on the new drive, and it's still running now.

Writing 1024 Mbyte uncompresseable data: 31 sec
Estimated time to write 2 * 409600 Mbyte: 24800 sec = 6 h 53 min
wrote 12845056 32 Kb blocks in 98 files in 11738 seconds (No space left on device)
wrote 3473408 32 Kb blocks in 53 files

that estimated time write is still high (i think) but the 'no space left on device' is roughly about double the amount of seconds as it was when the tape was only being filled 50%. So i think this is going to produce a proper tapetype, if not a slow one.

So it could be a combination of things.. i'll know more tomorrow.

On Monday 08 March 2010, rory_f wrote:

Please do. It could be something I missed, or it could be a brand new
phenomenon to file away in my trivia file.



gene,

amanda@backup tor]$ amtapetype -o -f /dev/nst0 -e 400G
Writing 1024 Mbyte compresseable data: 31 sec
Writing 1024 Mbyte uncompresseable data: 31 sec
Estimated time to write 2 * 409600 Mbyte: 24800 sec = 6 h 53 min
wrote 12845056 32 Kb blocks in 98 files in 11738 seconds (No space left on device)
wrote 12910592 32 Kb blocks in 197 files in 11840 seconds (No space left on device)
define tapetype NEW_IBM_DRIVE {
comment "just produced by tapetype prog (hardware compression off)"
length 402432 mbytes
filemark 0 kbytes
speed 34955 kps
}

comparing this tapetype to the last ibm drive (that broke and was replaced by this) the speed is nearly half.

define tapetype BROKE_IBM_DRIVE {
comment "just produced by tapetype prog (hardware compression off)"
length 402432 mbytes
filemark 0 kbytes
speed 71201 kps
}

This brings me to another question - what could be causing this ?

The system itself has not changed, nor has the cabling, really. unless the cables have become damaged (or the scsi card inside the machine is now having problems), what else is to check?

Theoretically, this new drive should be as fast if not faster.. but it isn't.

At least it seems to *work* now.

On Tuesday 09 March 2010, rory_f wrote:
Gene Heskett wrote:
On Monday 08 March 2010, rory_f wrote:

Please do. It could be something I missed, or it could be a brand new
phenomenon to file away in my trivia file.

gene,

amanda < at > backup tor]$ amtapetype -o -f /dev/nst0 -e 400G
Writing 1024 Mbyte compresseable data: 31 sec
Writing 1024 Mbyte uncompresseable data: 31 sec
Estimated time to write 2 * 409600 Mbyte: 24800 sec = 6 h 53 min
wrote 12845056 32 Kb blocks in 98 files in 11738 seconds (No space left on
device) wrote 12910592 32 Kb blocks in 197 files in 11840 seconds (No
space left on device) define tapetype NEW_IBM_DRIVE {
comment "just produced by tapetype prog (hardware compression off)"
length 402432 mbytes
filemark 0 kbytes
speed 34955 kps
}

comparing this tapetype to the last ibm drive (that broke and was replaced
by this) the speed is nearly half.

define tapetype BROKE_IBM_DRIVE {
comment "just produced by tapetype prog (hardware compression off)"
length 402432 mbytes
filemark 0 kbytes
speed 71201 kps
}

This brings me to another question - what could be causing this ?

The system itself has not changed, nor has the cabling, really. unless the
cables have become damaged (or the scsi card inside the machine is now
having problems), what else is to check?

If the card is wide scsi, perhaps a cabling issue or termination issue has
caused it to fall back to scsi-II width and speeds? I have read that some
cards do this, and a reboot once the problem is solved, might bring back the
full speed. Might be worth a try just to satisfy the curious cat in all of
us.

Theoretically, this new drive should be as fast if not faster.. but it
isn't.

Do you recall the block size used with the previous drive? If it was also
32k which is the default, some speed improvements can be had with larger
block sizes, but I doubt that could make a 2/1 ratio change in the drives
speed.

At least it seems to *work* now.

Which is nice, but I believe I would take up the speed problem with the
vendor after I had rebooted and checked to see if it persists. Perhaps there
is a spindle speed jumper that is miss-placed on this one? Possibly
automatically interlocked with the write clock speed since you did get the
exact same size, so the drive maintains the same bits per inch.

But I'm making SWAG's here, my knowledge of drives isn't nearly as deeply
ingrained as the scsi experience is.

Perhaps someone else here can help?

--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)

The control of the production of wealth is the control of human life itself.
-- Hilaire Belloc

On Tuesday 09 March 2010, rory_f wrote:

If the card is wide scsi, perhaps a cabling issue or termination issue has
caused it to fall back to scsi-II width and speeds? I have read that some
cards do this, and a reboot once the problem is solved, might bring back the
full speed. Might be worth a try just to satisfy the curious cat in all of
us.

Theoretically, this new drive should be as fast if not faster.. but it
isn't.

Do you recall the block size used with the previous drive? If it was also
32k which is the default, some speed improvements can be had with larger
block sizes, but I doubt that could make a 2/1 ratio change in the drives
speed.

At least it seems to *work* now.

Which is nice, but I believe I would take up the speed problem with the
vendor after I had rebooted and checked to see if it persists. Perhaps there
is a spindle speed jumper that is miss-placed on this one? Possibly
automatically interlocked with the write clock speed since you did get the
exact same size, so the drive maintains the same bits per inch.

But I'm making SWAG's here, my knowledge of drives isn't nearly as deeply
ingrained as the scsi experience is.

Perhaps someone else here can help?



I've rebooted the machine a few times. I'll try again, after i set the cables again.

One question; i don't have to run amtapetype every time i do this do i? For instance, say the speed problems are resolved, will amdump pick up on this automatically or does it stick to the tape speed given in amtapetype absolutely?

On Tue, Mar 9, 2010 at 9:55 AM, rory_f <amanda-forum < at > backupcentral.com> wrote:
One question; i don't have to run amtapetype every time i do this do i? For instance, say the speed problems are resolved, will amdump pick up on this automatically or does it stick to the tape speed given in amtapetype absolutely?

Amdump just runs the tape as fast as it goes. The speed estimate is not used.

Dustin

--
Open Source Storage Engineer
http://www.zmanda.com

On Tue, Mar 9, 2010 at 9:55 AM, rory_f <amanda-forum < at > backupcentral.com> wrote:
One question; i don't have to run amtapetype every time i do this do i? For instance, say the speed problems are resolved, will amdump pick up on this automatically or does it stick to the tape speed given in amtapetype absolutely?

Amdump just runs the tape as fast as it goes. The speed estimate is not used.

Dustin

--
Open Source Storage Engineer
http://www.zmanda.com

Thanks Dustin, good to know.

Thanks to everyone else for the help. Time to return this HP drive ;-]

On Tuesday 09 March 2010, rory_f wrote:
Gene Heskett wrote:
On Tuesday 09 March 2010, rory_f wrote:

If the card is wide scsi, perhaps a cabling issue or termination issue
has caused it to fall back to scsi-II width and speeds? I have read that
some cards do this, and a reboot once the problem is solved, might bring
back the full speed. Might be worth a try just to satisfy the curious
cat in all of us.

Theoretically, this new drive should be as fast if not faster.. but it
isn't.

Do you recall the block size used with the previous drive? If it was
also 32k which is the default, some speed improvements can be had with
larger block sizes, but I doubt that could make a 2/1 ratio change in the
drives speed.

At least it seems to *work* now.

Which is nice, but I believe I would take up the speed problem with the
vendor after I had rebooted and checked to see if it persists. Perhaps
there is a spindle speed jumper that is miss-placed on this one?
Possibly automatically interlocked with the write clock speed since you
did get the exact same size, so the drive maintains the same bits per
inch.

But I'm making SWAG's here, my knowledge of drives isn't nearly as deeply
ingrained as the scsi experience is.

Perhaps someone else here can help?

I've rebooted the machine a few times. I'll try again, after i set the
cables again.

One question; i don't have to run amtapetype every time i do this do i? For
instance, say the speed problems are resolved, will amdump pick up on this
automatically or does it stick to the tape speed given in amtapetype
absolutely?


Absolutely no, amanda ignores that. It is for your edification only.

One other question: Did you take note of whether or not the drive was
streaming steadily, or was it 'shoe shining' occasionally? One can normally
hear such goings on by the squawk of the steppers as they wind up and slow
down.
+----------------------------------------------------------------------

|This was sent by rory < at > mrxfx.com via Backup Central.
|Forward SPAM to abuse < at > backupcentral.com.

+----------------------------------------------------------------------



--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)

Painting, n.:
The art of protecting flat surfaces from the weather, and
exposing them to the critic.
-- Ambrose Bierce

On Tuesday 09 March 2010, rory_f wrote:

Absolutely no, amanda ignores that. It is for your edification only.

One other question: Did you take note of whether or not the drive was
streaming steadily, or was it 'shoe shining' occasionally? One can normally
hear such goings on by the squawk of the steppers as they wind up and slow
down.



Hey,
I couldn't tell you . I don't recall anything like that though.