Windows XP backup "Looking for backup devices" very slow

[Jonathan Sachs]

I'm sorry it took ME so long to get back to YOU. Classes resumed last week, and much of my time has been occupied by that.

I'm afraid I do not have any useful results to report. The script ran for less than a second, and appears to have failed. Following is the execution log (which I am transcribing, due to the copying problems which I noted earlier). I am using periods to represent leading spaces, since the bulletin board ignores actual spaces and treats " " as ordinary text, and I can't find anything that tells me what formatting codes it accepts.

Running script
...Reading script file
...Starting script execution of script DDSWriteRecalProc
...Script exception occurred
......Script exception occurred
.........I/O Exception (Sense Key: 6, Sense Code: 0x2800)
............SCSI Cmd I/O at 4/0.1.0: Read Buffer
......SCSIScript: auto abort triggered (Line: 401, Error: Auto abort triggered)
.........I/O Exception (Sense Key: 6, Sense Code: 0x2800)
............SCSI Cmd I/O at 4/0.1.0: Read Buffer
...Script execution completed

When I was preparing to run the script, L&TT warned me that there was a new version of the script and asked if I wanted to download it. I did not do so, since at the start of this process I had to fall back from current versions of scripts to older ones to make L&TT work at all.

[CA843380]

Jonathan,

The problem you are seeing is due to a current bug with LTT 3.5 that causes the DDS3 recal script to fail I think.

Please uninstall LTT 3.5 and download and install L&TT 3.4 SR1a to recalibrate the drive.

I have provided a link for you below, just select the 3.4 SR1a download link.
http://h20000.www2.hp.com/bizsupport/TechSupport/DriverDownload.jsp?pnameOID=21419&locale=en_US&taskId=135&prodSeriesId=42863&prodTypeId=12169&swEnvOID=228

Cheers,

Dave Dewar

[Jonathan Sachs]

I set aside this morning to try again, but when I opened the dialog box to save the 3.4 SR1 download on disk I found that I already had a copy of it.

I checked the version of L&TT that I used last time (installed then, still installed now), and realized that it _is_ 3.4 SR1a.

What now?

[CA843380]

Hi Jonathan,

Ok, so you are running 3.4 in which case the script should work.
I think I gave you bad info last time round, I should have paid more attention to the info you posted.
Looking in detail at the log you typed out from when you ran the recal script it appears that your drive reported a "unit attention" due to the media being changed, decoded from sense key 6 sensecode 2800. No idea what happened for this to occur, did you hit the eject button during script execution or something?

Execute the script with no tape loaded, it will prompt you for a tape, load one and the script should progress.

Cheers,

Dave Dewar

The script also generates an output under the test results tab (like the error rate info you posted) that provides more info.

A good output should look like:

Script Output:

Script version: V17.04.2003
HW-CONF = 0x00040001
Checking firmware file ../firmware/C1537A_L111.frm correct ...
Storing values from drive calibration table
Write Current values DDS3: 0x28
Write Current values DDS2: 0x23
Reading tape usage from serial EEprom
Tape usage is dds2 = 0 dds3 = 790
Prompt user to load NEW DDS3 media cartridge
Waiting for 'DDS3' tape to load ...
Checking for correct cartridge type ...
Using NEW media to Clean heads ...
Sending Servo Cmd : 18
Rewinding tape ...
Starting Write Current Recalibration ...
Checking for correct cartridge type ...
Set Buffered Mode: ON ...
Setting write current to 0x28 ..

will continue like this for awhile setting differnet write currents.

[Jonathan Sachs]

>...No idea what happened for this to
>occur, did you hit the eject button during
>script execution or something?

The script instructions were silent about whether to insert a tape or not. Considering what it was supposed to do, I assumed it would need one, so I put one in before starting it. My action was definitely influenced by the dire warnings about damage to the drive if I did not follow instructions; in this case there _were_ no instructions, so I did what I thought would be safest.

>Execute the script with no tape loaded, it
>will prompt you for a tape, load one and
>the script should progress.

Will try it again at the first opportunity.

[Jonathan Sachs]

I ran the script again tonight. It concluded by telling me that "The test has found that the drive is already set to the optimum write current."

I'm a little suspicious of the result because of the messages in the log. Using the same dot notation as before, it says:

Running script
... Running script file...
... Starting script execution of script DDSWriteRecalProc
... Script exception occurred
......SCSIScript: script aborted (Line: 640, Error: SYS_Abort() called to abort script)
... Script execution completed

Following is the entire script output:

Script version: V17.04.2003
HW-CONF = 0x00040001
Checking firmware file ../firmware/C1537A_L111.frm correct ...
Storing values from drive calibration table
Write Current values DDS3: 0x19
Write Current values DDS2: 0x23
Reading tape usage from serial EEprom
Tape usage is dds2 = 0 dds3 = 571
Prompt user to load NEW DDS3 media cartridge
Waiting for 'DDS3' tape to load ...
Checking for correct cartridge type ...
Using NEW media to Clean heads ...
Sending Servo Cmd : 18
Rewinding tape ...
Starting Write Current Recalibration ...
Checking for correct cartridge type ...
Set Buffered Mode: ON ...
Setting write current to 0x19 ..
Writing loop 10 of 120
Writing loop 20 of 120
Writing loop 30 of 120
Writing loop 40 of 120
Writing loop 50 of 120
Writing loop 60 of 120
Writing loop 70 of 120
Writing loop 80 of 120
Writing loop 90 of 120
Writing loop 100 of 120
Writing loop 110 of 120
Writing loop 120 of 120
Frames Written 1307
ERT values A: 0.031142 B: 0.052434
Setting write current to 0x14 ..
Writing loop 10 of 120
Writing loop 20 of 120
Writing loop 30 of 120
Writing loop 40 of 120
Writing loop 50 of 120
Writing loop 60 of 120
Writing loop 70 of 120
Writing loop 80 of 120
Writing loop 90 of 120
Writing loop 100 of 120
Writing loop 110 of 120
Writing loop 120 of 120
Frames Written 628
ERT values A: 0.081899 B: 0.117693
Setting write current to 0xf ..
Writing loop 10 of 120
Writing loop 20 of 120
Writing loop 30 of 120
Writing loop 40 of 120
Writing loop 50 of 120
Writing loop 60 of 120
Writing loop 70 of 120
Writing loop 80 of 120
Writing loop 90 of 120
Writing loop 100 of 120
Writing loop 110 of 120
Writing loop 120 of 120
Frames Written 1168
ERT values A: 0.105139 B: 0.177801
Setting write current to 0xa ..
Writing loop 10 of 120
Writing loop 20 of 120
Loop: 26 ERT value too high .. skip this write!
ERT: 0.041788 Current: 0x19
Min ERT: 0.041788 Min Curr 0x19

[CA843380]

Hi,

The logfile looks fine. The drive has tried to optimise (lower) the error rate measured but trying successively higher write current settings and has found that the setting you have at present is giving the best error rate performance.

However, this error rate performance is not great and I think it is at the root of your intermittment long verify issues. I suspect that you have a drive with significant head wear that cannot not be compensated for by write current adjustments. The recal script is not alway successful in recovering drive performance.

Your options are:
1. Do nothing.
2. Replace the drive, is it still in warranty? Rereading the thread you mentiond you bought it 2 years ago? We typically offer a 3 year warranty so it would be worth phoning up HP support and seeing if you can get an exhange.
3. If this is not possible your could try using DDS2 tapes instead. This would limit your backup size and transfer rate but the drive would have more margin in this situation.

Cheers,

Dave Dewar

[Jonathan Sachs]

The drive is 'way out of warranty, and doing nothing really is not an option. I have been running backups far less frequently than I should because of this problem, and I have doubts about how useful they will be in the drive restores as slowly as it verifies.

I noticed that you did not list repair as an option. Is this device considered nonrepairable? Considering what I paid for it, that would be rather disappointing, and would make me think twice about buying another HP drive as a replacement.

[Jonathan Sachs]

One person has suggested that before I replace the drive I open it up and try physically cleaning the heads. That seems reasonable to me if it is unusable and unrepairable in any case. Do you have any advice on how to conduct this operation with the maximum likelihood of success?

[Greg George]

I have used an exhaustive variety of older SCSI cards under XP, and in my experience, the 2910B and C are not well supported. It is telling that Adaptec does not offer drivers for this card for XP on its website, and does not include this operating system among those supported. These cards, and their European counterparts, were made largely for MACs.

The card has no boot BIOS for setting internal parameters at boot time. Instead, there is a utility available from Adaptec which must be run after booting from DOS with an appropriate DOS SCSI driver (attached). Suggest you create a DOS bootdisk with SCSI support (see bootdisk.com) and fiddle with the onboard parameters, particularly transfer rate through the card under XP. Also try changing termination from auto to manual if you have only external devices.

My bottom-line advice is to upgrade your SCSI card to an inexpensive 2930. While I realize you have no SCSI boot devices and don't need the BIOS, I have found this card to have rock solid support under XP. Further, it has an entirely different SCSI chip. I have run into problems with the 2910 under a variety of operating systems and have replaced mine with a 2930, to good effect.

I don't know anything about Adaptec's competitors, but I bet there are some cheap 2930 knockoffs out there. Take a look at XPs supported hardware list on the Microsoft website for alternate SCSI cards if you can't afford or don't want to spend money on the best, Adaptec.

[Jonathan Sachs]

>My bottom-line advice is to upgrade your SCSI card ...

I appreciate your effort to help, but I don't think this suggestion has anything to do with the problem. I'll summarize some facts which may not albeit apparent from the earlier messages.

The problem is read errors during the verifying step. It's not clear how that could be caused by the controller.

I used this card with this device and Windows XP for several weeks before the problem appeared, and used them with Windows 2000 four years before that. I have never had (or heard of) problems attributable to the card.

Adaptec does not publish a driver for Windows XP because it would be pointless to do so -- the driver is built into the Windows distribution.

Adaptec's manual for the 2910B states that it supports Windows 95, Windows NT, Windows 3.1, etc. (It does not mention later versions of Windows, of course, since it was published before they existed.)
It does NOT mention support for any Apple operating system.

At least one Adaptec technical note cautions against trying to use adapters intended for Wintel systems in Apple systems or vice versa; it lists adapters that are compatible with Apple systems, and the 2910 is not among them.

I have a problem with the drive. I want to stay focused on that.

[CA843380]

Hi Jonathan,

HP does not offer a repair service on out of warranty tape drives. There are 3rd party repair shops out there but I have heard very mixed reports about their success. I am not aware of a repair shop in the UK but I expect a quick hunt on google would find one.

With regard to cleaning the heads: this is possible, however I cannot guarantee it will fix your issue even if you manage to perform it successfully without damaging the heads.

However, it is certainly an option open to you as a "last ditch" thing to try. The write/read heads are contained in the horizontal gap in the silver drum within the tape drive, 2 read heads and 2 write heads. They are the very small black retangular objects you can see as you turn the drum with a finger at 90, 180, 270 and 360deg positions. The best way to clean them is to use a very soft cotton bud and Iso propol alcohol (IPA). You need to be very careful you do not break the heads or leave cotton strands on them. It's a very tricky process but it can be done.

Good luck.

Dave Dewar

[Greg George]

You've heard now. The question is what are you going to do about it?

XP does not support the card, but rather the chip on which the controller is based. This is a significant distinction. This card is considered outdated hardware by the XP operating system. You will find no 2910 driver under XP, only the corresponding chipset driver. This is because the Adaptec chip is in a variety of devices sold by a variety of venders, mostly crap.

"I used this card with this device and Windows XP for several weeks before the problem appeared, and used them with Windows 2000 four years before that. I have never had (or heard of) problems attributable to the card."

Its irrelevant that this card worked under 2000, or that it worked briefly with XP. It is well known that XP service packs contained updates for SCSI drivers.

"Adaptec does not publish a driver for Windows XP because it would be pointless to do so -- the driver is built into the Windows distribution."

Wrong, see above. Adaptec refuses to claim that the piece of crap 2910 card is supported by XP, check their website. Suggest you buy a real card, not a toy like this card is.

"daptec's manual for the 2910B states that it supports Windows 95, Windows NT, Windows 3.1, etc. (It does not mention later versions of Windows, of course, since it was published before they existed.)"

Yah, that's because the card was made before XP existed and isn't supported properly. Its based on a crappy controller chipset which is no longer in use. You cannot buy a 2910 card today with a box that claims XP support, and theres a darned good reason for that.

"t does NOT mention support for any Apple operating system."

"At least one Adaptec technical note cautions against trying to use adapters intended for Wintel systems in Apple systems or vice versa; it lists adapters that are compatible with Apple systems, and the 2910 is not among them.

See the Apple website. Its the chip, obviously, not the card. You can't but a PC card in an Apple. This controller chip was designed for the Apple and later migrated to the PC."


"I have a problem with the drive. I want to stay focused on that."

The 2910 card is for sissies. If you're going to spend $179 on an XP upgrade, you might as well get a real SCSI card that is properly supported by the operating system."

[Greg George]

Neither the 2910 card or its corresponding controller chip appear in the Windows Hardware Compatibility catalog for XP. There are tens of thousands of these cards around. Why do you suppose this card was not included for XP compatibility?

http://www.microsoft.com/windows/catalog/default.aspx?subID=22&xslt=search&qu=Adaptec&scope=2&btnSearch=Go

No, XP doesn't support the 2910 chipset properly. This may not have anything to do with your problem; you may have a broken drive too. We usually isolate the variables in solving a problem like this, however.

[Jonathan Sachs]

Well, my pursuit of a solution to this problem has led me to undreamt-of levels of bafflement. The only bright spot is that I have solved the problem.

First, I tried Greg's solution. I purchased an Adaptec 2940AU to see if that would help. (I was looking for a 2930, but I found a deal on the other card first.) It gave me exactly the same results as the 2910.

Then -- don't ask me why I tried this -- I attached the tape drive to my computer through an Adaptec USB2XCHANGE SCSI-to-USB 2.0 converter. That made the problem go away. Not only is the verify step a little _faster_ than write step, instead of five times slower, but both are about twice as fast as the write step was with either the 2910 or the 2940: 100 MB/sec compared to 50-55 MB/sec.

So it was the card all along. But the 2940, which IS certified Windows XP compatible, works just as badly as the 2910, which is not.

The doubled speed is a complete mystery. Why should a gadget which is essentially a hardware kludge, and adds another driver and another layer of protocol conversion to the I/O chain, be twice as fast as an adapter which was designed to work with the drive's native interface?