Issues with Ultrium-3 Tape drive

[Joshua Small_2]

Hi,

Hot off the back of my 100+ post thread on the server forums, here's my newest issue.

I am running Windows 2003 R2, 64 bit edition, on a DL320 G5. Our tape drive is connected to an HP dedicated SCSI card.

We have of course checked drivers and firmware of both the tape drive and SCSI card as being the latest. The server itself has PSP 7.9 applied.

Using ntbackup, I'm seeing logs that error like this:

Error: The device reported an error on a request to write data to media.

Error reported: Hardware failure.

There may be a hardware or media problem.

It obviously looks to be a hardware problem, so we arranged for the replacement of the tape drive and then then SCSI card, and call cabling. This took quite some insisting, threatening and pushing, due to HP's insistance that it simply couldn't be hardware.

Anyway, we've had this issue with the "backlevel" engineers, which is allegedly the highest level of support, since the 24th July, who called today and stated they had reviewed the event viewer and satisfied themselves that the case could safely be resolved.

Since we purchased this server we haven't had a good backup. HP are vehemently refusing to provide any further support on the basis of this article:
http://support.microsoft.com/kb/817688

The article is quite clear however that it refers to an issue corrected in SP1. I called Microsoft for giggles and they informed us the patch isn't even available for a 64 bit server, given that Windows x64 always came wtih Sp1. Three different technicians, many harsh phone calls and eveyr manner of blame shifting has assured us that the above Microsoft article makes this not HP's fault.

Any assistance on this would be most appreciated.

[Joshua Scott]

What model of SCSI card is this? You should have and Ultra 320 speed SCSI card for the Ultrium 3 drives. What brand/model of tape drive is it?

[Joshua Small_2]

It's an HP branded tape drive supplied with the server, Ultrium 448.
The SCSI card is the LSI card that HP refer to only as "Dual channel Ultra 320 SCSI".

[CLEB]

Hi

Have you tried using another backup product?

I wouldn't trust ntbackup to protect a floppy disk.

I know you can get a fully working 60 day evaluation of Brightstor ARCserve from www.ca.com

Sounds like you've spent too much time on it and should be looking at alternatives.

I use Ultrium 3 tapes drives, standalone models and MSL6000 models with no problem.

[Joshua Scott]

I will just point out that the Ultrium 448 is an Ultrium 2 drive, not Ultrium 3, so make sure you are using the appropriate media.

Josh

[Joshua Small_2]

My error there..
The tape drive is an Ultrim 960, and yes a Ultrium 3- I made a post with reference to an incorrect drive.

The drive FAILS with IO ERROR the HP Library and Tape Tools test, but HP still are of the position that it's an OS bug.

[Daniel McLaren M.]

Hello Joshua, I was wondering if you could post the following:

Go into the device manager look at the SCSI controller right click it and see if in the description of the device you find the following (storport) or (scsiport), also if you could please place the driver version you have. If you have the product number of the controller paste it too.

This may be a long shot, however several issues have been reported with the storport driver, and maybe you will have better results with the scsiport.

Please try the following if you find (storport) in the product description.

This is the link for a scsiport driver that should be correct based on the controller you described (restart the server after you install it):

http://h18023.www1.hp.com/support/files/server/us/download/22560.html

After you install that driver, try to run the LTT Installation Check, this is a seperate application from LTT you can find it as an additional icon the program group folder, try all options except the final 30 minute one, but before make sure you have removal storage disabled if not it will fail the tests. (remember to turn it back on after the tests if not ntbackup won't work well).

That test will show you any problems related to cabling, damaged termination, etc which may have been overlooked during troubleshooting.

If it does not return any errors try to run a simple backup using nt.

[Joshua Small_2]

Thanks for the assistance. I will be trying that driver tonight and advising. Looking at the publish date and supported servers, am I right in understanding this is just an older edition of the driver?

The "supported servers" only appears to include an earlier edition of our server.

We are currently running:
LSI Adapter Ultra 320 SCSI 2000 Series (storport)
version 1.24.4.0


We have also applied the latest MS storport update, as an earlier attempt at fixing the issue.

Running the LTT Installation test actually passes (done that already). It's only the "read/write" test that ever causes a fail, and then it usually gets through 30-40 GB data before failing.

I am vigorously persueing this with HP, but their last communication was only to recommend disabling shadow copies.

[CLEB]

I remember the SDLT drives used to have an issue where they would overheat after being used for too long. Perhaps some of the Ultriums have this issue too?

[Daniel McLaren M.]

Yes Joshua, it is a older driver, but it is a scsiport driver, changing the driver to a older release might do the trick.

[Joshua Small_2]

I have tested and can confirm the issue is not related to the driver.
I had swapped the tape drive and SCSI card and still found the issue persisted.
However after swapping a riser board, the issue appears resolved.

I am satisfied the riser board is the cause of the issue, of course the difficulty now is getting HP to accept that.

[Andrew Mackay(CA)]

I have just found this post.

I am having the exactly the same issue but with a couple of differences.

1. I have HP DL380 G5's
2. Windows 2003 R2 32bit edition
3. HP LSI Ultra 320 2000 Series PCI-E SCSI card
4. HP Ultrium 960 direct attached drive.
5. L&TT reports I am having a SCSI cabling issue
6. HP Data Protector reports an issue writing data to media.[90:51] Cannot write to device because of an I/O error.

Since we purchased this server we haven't had a good backup.

I noticed that the PSP(Latest) I installed does not even install the latest PCI controllers etc for the mainboard.

So my question is there a problem with this new 'storport' scsi card driver?

[Joshua Small_2]

Hi Andrew,

I don't believe it's a storport issue at all.

I should have followed this up more, but I got distracted.
The issue resurfaced and eventually HP loaned us a completely different tape drive (SDLT) and the issue went away.

So in summary either we've had three different bad LTO drives (possible but unlikely) or there is simply something wrong with the LTO drives or the LATEST driver shipping with them.

Perhaps it wasn't ported to a 64 bit OS correctly, or perhaps it just is too generic and hasn't been customised for this drive.

We've been waiting several months for a further result but so far HP still tell us they are working on putting together a test lab so they can replicate it.

[Andrew Mackay(CA)]

I working on this issue at the moment as we are at a critical stage due to this being a production server in a remote office which is making it very hard.

Currently results
- I having the problem in a 32bit environment
- I can't backup anything over 30GB
- If I run smaller backups under 30GB they are successful.
- I was going to start changing the block size on the tape as a test.


Thanks for you quick reply Joshua
I will keep you posted, I have just logged my call with HP

[Joshua Small_2]

Well it's interesting to see it's not a 64 bit OS issue.

We too have a production server in a remote office.. it's a nightmare to work through this.

Do you have a copy of the HPSReport tool from HP? It's their tool that collects server information for their diagnostics. If you want to call them, I would strongly suggest you run it first. Then with them on the phone, get the email address and send it to them, even if they say they don't want it. Don't let them off the phone until they acknowledge they have it.

Then spam it to them every couple of days just to make sure they got it.

It seems to be the only way to stop them wasting your time by repeatedly asking for it.

[Soren Nicolaisen]

Hi,

We now have 2 cases with DL380G5's. One is at a remote office and the other is luckily at our main location.
Our remote site is using a MSL2024 library and the main facility uses a MSL6060.
Both DL380G5's uses the SC11Xe HP SCSI card and the 1.24.4 storeport drivers from the psp 7.90. Two of those cards are used in the backup server in our main facility. One of those cards run the one drive and robot and the other runs the second drive. Terminators all have green lights and cables work well with the old server.
OS is Windows 2003 R2 SP2 32bit installed on the P400 smart array controller
The Tape drives and Robot runs the latest firmwares supplied with the latest release of the HP tape tools.
Our corporate backup tool is EMC Networker 7.3 SP3 and when setting up the Jukebox we receive the following error running the jbconfig jukebox configuration tool that add's the robot to Networker.

Scanning SCSI buses; this may take a while ...
In update_scsi_bus_map: open_bus invalid on nadapters = 5
jbconfig: Jukebox error: scsi command MODE_SENSE failed

HP Tape tools states an error on the juke box, but not the drives.
If we revert to our old backup server that runs Windows 2003 SP1, the robot, HP Tape tools and tape drives + networker function perfectly.
We tried changing the riser card with one that supports PCI-X on the G5 and the old SCSI card from the old backup server that we know works, and the results was the same.

We are starting to run out of good ideas of what is causing our problems.

Thanks
Soren

[Joshua Small_2]

I've still gotten nowhere with this.
The tape drive works fine on other servers.
Different drives of the same type do not work on our server.
Different drives of different type do. The drive is supposed to be HP supported, so far we're waiting on a November 7th date for someone to start looking at it.

[HKoenig]

Hallo Joshua,

had anybody checked the cabling, regarding the differences for the LTO3 Tape drives in the MSL. Youre termination should be on the drives,
not as in the past on the Robotic Controller. You got problems with the old termination, depending on speeed and amount of data.

http://h20000.www2.hp.com/bc/docs/support/SupportManual/c00873645/c00873645.pdf

[Joshua Small_2]

We've had HP technicians onsite observing the cabling (which had already been changed) and there was no issue anyone could locate.

[Andrew Mackay(CA)]

Just an update on my issue?

I still have not got anywhere with this, HP don't have a clue whatâ s going.

The diagnostics they want me to perform are so unrealistic and the HP technician I have assigned refuses to believe this thread is related to my problem.

This is an evil problem that has me stumped

[ynotravid]

Soren, we had a similar issue with one our remote sites where everything seemed to be working but jbconfig would give error:

jbconfig: Jukebox error: scsi command MODE_SENSE failed

We changed the windows driver to "Unknown Media Changer" and then jbconfig worked.

[Marc Hemmett]

We have similar issue with a Ultrium 960, HP DL380G5, PCI-E Ultra SCSI 320 controller with the 1.24.4 StorPort driver Win 2003.
We moved the TBU, cable, to a HP DL 380 G3 with a HP LSI 1020 SCSI controller and 1.8.23 driver and all works fine. What appears to be the issue is the controller / driver combination as at the Windows level it appears to work, but will not been seen in the device list in Brightstor (same version both servers). We are going to replace the controller with a different version and then advise HP what happens..

[Joshua Small_2]

Well we finally got somewhere on this.
HP gave up on the SCSI drive and replaced it with a SAS model (and provided a SAS card).

It runs at half the speed and it's actually worth quite a bit less, but it does work and that's more than we've had previously.

[MarkNg]

I am actually encountering same problem with yours. I have a DL580G5 attach to 1/8 G2 Autoloader with ultrium920 drive and LSI ultra320 PCI-e card. I am using Arcserve 11.5R3 for backup. Arcserve always encounter hardware error when backup reach aroung 30GB.

I have did some experiment by attaching a SDLT drive to the server and backup went through sucessfully. I get HP to bring in a new 1/8 G2 autoloader and it is still encountering same problem either with Arcserve or NTbackup.

[Joshua Small_2]

I'm about to give up on SCSI tape drives altogether.

I have a second server with a similar problem, and haven't been able to get anywhere with it.