no tape drive in /proc/scsi/scsi

[ToddAndMargo]

Hi All,

I have an Exabyte VXA3 on an Adaptec 16160 scsi card. I am using CentOS 5.
The VXA3 BIOS probes on ID=1 at boot up (Adaptec's BIOS). The drive does
not show up in "cat /proc/scsi/scsi". In my messages log I get the following
boot messages:

Apr 16 20:14:55 server kernel: scsi1 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 7.0
Apr 16 20:14:55 server kernel:
Apr 16 20:14:55 server kernel: aic7892: Ultra160 Wide Channel A, SCSI Id=7, 32/253 SCBs
Apr 16 20:14:55 server kernel:
Apr 16 20:14:55 server kernel: (scsi1:A:1:0): parity error detected in Data-in phase. SEQADDR(0x1a6) SCSIRATE(0x0)
Apr 16 20:14:55 server kernel: Unexpected non-DT Data Phase
Apr 16 20:14:55 server kernel: (scsi1:A:1:0): parity error detected in Message-in phase. SEQADDR(0x1a7) SCSIRATE(0x0)
repeats 18 times.

How do I fix this?

Many thanks,
-T

[ToddAndMargo]

Hi All,

Mystery solved: I swapped a spare VXA 1 into the server got
the following:

cat /proc/scsi/scsi
Attached devices:
Host: scsi1 Channel: 00 Id: 05 Lun: 00
Vendor: ECRIX Model: VXA-1 Rev: 2A6A
Type: Sequential-Access ANSI SCSI revision: 02

The VXA 3 is still under hot swap. I will be exchanging it for a
another unit.

-T

[mtnbkr]

Any luck with the replacement VXA3? I Just got (2) VXA3 upgrade drives to put into my client's 1U Packetloaders and BOTH VXA3 drives are doing exactly the same thing you have described.

Did you ever get a resolution to this?

I am checking to see if a newer aic7xxx driver will fix this - ie: looking at lastest kernel source moving from the currently running 2.6.18.

--
Bill Arlofski
Reverse Polarity

[ToddAndMargo]

Hi Bill,

Yes I did. It turns out the problem is the stinkin' BIOS
on the tape drive. It is incompatable with Adaptec
SCSI Cards in a UNIX environment.

Exabyte's tech support sent me an "unpublished" BIOS
and that fixed it.

I had to put the turkey (VXA3) into a Windows 2000
server to update the BIOS. Alternately, Exabyte Tech
Support can send you a BIOS tape that is self installing
(But you have to keep after them, they have a hard
time following up on things.)

HTH,
--T

[mtnbkr]

Strange... On my client's server, there is an Adaptec 29160 single-channel LVD controller. Same as you, the Adaptec's BIOS recognizes the card properly, but when the Linux aic7xxx driver loads, I get those errors and no recognition of the VXA-3 device.


I have taken the drive to my office and have plugged into both an Adaptec 2940UW (ultra-wide, single-ended) controller and it works perfectly (not LVD of course), and just now into an Adaptec 2940U2W (ultra2 wide, LVD) and it also works perfectly.

As I was installing the Adaptec 2940U2W controller I just read your post about adaptec and VXA-3 BIOS issues and ALMOST plugged in the LSI Logic card I have on hand, but this VXA-3 drive seems to be happy with Adaptec controllers here, just not with the 29160 in my client's server.

I will be bringing the LSI Logic card with me (after I also test it here) just in case.

My issue might still turn out to be the pass-thru SCSI cable in the Packetloader. I will find out when I get on-site today.

Thanks for your reply.


P.S. And regarding your comment on Exabyte's "hard time following up on things..." I know all too well when last summer I installed 2 new VXA-2 1UPacketloaders at two of my client's sites. Neither worked - read errors on the tapes, sometimes they would work, then they would not... Required a complete power-off of the tape units to resume even partial functionality... This dragged on from July until September/October when both of my clients returned their packetloaders and moved to Quantum DLT autoloaders. Not a single SCSI or tape related issue since.

...And here I am again - fighting with an Exabyte tape drive. Sigh....