Fiber channel tape drive config help

Hi all,
I am trying to get an FC tape library configured and I'm almost
there but for one issue - I can't access the tape drives very well.
Here's what I have.

Sun Enterprise 250 with Solaris 9
JNI FCI-1063 PCI HBA with JNIfcaPCI (v2.6.13 driver) package
Veritas Netbackup 4.5 MP6
Quantum/ATL P2000 library with FC230 card and 8 SDLT (220) drives
Brocade Silkworm 2800 switch

Solaris can see the drives and makes all 8 in /dev. Netbackup's
sgscan sees all the devices. robtest can move all the tapes around.
But I can't get much info back from the tape drives directly from
Solaris using mt or tar. After I first load a tape in a drive I get
this after a 5-10 second pause.

# mt -f /dev/rmt/4cbn status
QuantumSuperDLT tape drive:
sense key(0x6)= Unit Attention residual= 0 retries= 0
file no= 0 block no= 0

Then this until I unload and reload the tape.

# mt -f /dev/rmt/4cbn status
/dev/rmt/4cbn: no tape loaded or drive offline

Here is the st.conf entries and boot console output, etc. Any
thoughts what to persue next? I have no support on ANY of these
devices.

"QUANTUM SuperDLT1", "QuantumSuperDLT", "SDLT-data";
SDLT-data = 1, 0x38, 0, 0x19639, 4, 0x90, 0x91, 0x90, 0x91, 3;

/kernel/drv/sparcv9/fca-pci symbol ddi_model_convert_from multiply defined
fca-pci0: JNI Fibre Channel Adapter model FCI-1063
fca-pci0: SCSI ID 125 / AL_PA 0x1
fca-pci0: Fibre Channel WWNN: 100000E069C01B17 WWPN: 200000E069C01B17
fca-pci0: FCA SCSI/IP Driver Version 2.6.13, Feb 26, 2003 for Solaris 64-bit
fca-pci0: All Rights Reserved.
NOTICE: pci1242,46430 LOOP Initialization Complete, AL_PA=01
NOTICE: fca_link_down unschedule port_discovery
fca-pci0: Host: Port 000001 (100000E069C01B17:200000E069C01B17)
WARNING: intr status = 00000000 status = 980000a0 saved = 980001a0
NOTICE: pci1242,46430 LOOP Initialization Complete, AL_PA=01
fca-pci0: Host: Port 000001 (100000E069C01B17:200000E069C01B17)
fca-pci0: Port 0000EF (2000001086100C0D:2000001086100C0D) available.
fca-pci0: Port 0000E8 (2000001086100C10:2000001086100C10) available.
fca-pci0: Port 0000E4 (2000001086100C0C:2000001086100C0C) available.
fca-pci0: Port 0000E2 (2000001086100C11:2000001086100C11) available.
fca-pci0: Target 0 Lun 0: Port 0000EF (2000001086100C0D:2000001086100C0D) present.
fca-pci0: Target 1 Lun 0: Port 0000E8 (2000001086100C10:2000001086100C10) present.
fca-pci0: Target 2 Lun 0: Port 0000E4 (2000001086100C0C:2000001086100C0C) present.
fca-pci0: Target 3 Lun 0: Port 0000E2 (2000001086100C11:2000001086100C11) present.
fca-pci0: Target 0 Lun 1: Port 0000EF (2000001086100C0D:2000001086100C0D) present.
fca-pci0: Target 1 Lun 1: Port 0000E8 (2000001086100C10:2000001086100C10) present.
fca-pci0: Target 1 Lun 2: Port 0000E8 (2000001086100C10:2000001086100C10) present.
fca-pci0: Target 2 Lun 1: Port 0000E4 (2000001086100C0C:2000001086100C0C) present.
fca-pci0: Target 3 Lun 1: Port 0000E2 (2000001086100C11:2000001086100C11) present.

essex# ls -al /dev/rmt/?
lrwxrwxrwx 1 root root 49 Dec 1 12:55 /dev/rmt/0 - ../../devices/pci@1f,4000/fibre-channel@2/st@0,0:
lrwxrwxrwx 1 root root 49 Dec 1 12:55 /dev/rmt/1 - ../../devices/pci@1f,4000/fibre-channel@2/st@2,0:
lrwxrwxrwx 1 root root 49 Dec 1 12:55 /dev/rmt/2 - ../../devices/pci@1f,4000/fibre-channel@2/st@3,0:
lrwxrwxrwx 1 root root 49 Dec 13 08:29 /dev/rmt/3 - ../../devices/pci@1f,4000/fibre-channel@2/st@0,1:
lrwxrwxrwx 1 root root 49 Dec 13 08:29 /dev/rmt/4 - ../../devices/pci@1f,4000/fibre-channel@2/st@1,1:
lrwxrwxrwx 1 root root 49 Dec 13 08:29 /dev/rmt/5 - ../../devices/pci@1f,4000/fibre-channel@2/st@2,1:
lrwxrwxrwx 1 root root 49 Dec 13 08:29 /dev/rmt/6 - ../../devices/pci@1f,4000/fibre-channel@2/st@3,1:
lrwxrwxrwx 1 root root 49 Dec 14 15:56 /dev/rmt/7 - ../../devices/pci@1f,4000/fibre-channel@2/st@1,2:

essex# /usr/openv/volmgr/bin/sgscan all
Set DISPLAY to whaler:0 ([y],n)?
/dev/sg/c0t0l0: Disk (/dev/rdsk/c0t0d0): "FUJITSU MAG3182L SUN18G"
/dev/sg/c0t6l0: Cdrom: "TOSHIBA XM6201TASUN32XCD"
/dev/sg/c2t0l0: Tape (/dev/rmt/0): "QUANTUM SuperDLT1"
/dev/sg/c2t0l1: Tape (/dev/rmt/3): "QUANTUM SuperDLT1"
/dev/sg/c2t1l0: Changer: "ATL P2000 6318050"
/dev/sg/c2t1l1: Tape (/dev/rmt/4): "QUANTUM SuperDLT1"
/dev/sg/c2t1l2: Tape (/dev/rmt/7): "QUANTUM SuperDLT1"
/dev/sg/c2t2l0: Tape (/dev/rmt/1): "QUANTUM SuperDLT1"
/dev/sg/c2t2l1: Tape (/dev/rmt/5): "QUANTUM SuperDLT1"
/dev/sg/c2t3l0: Tape (/dev/rmt/2): "QUANTUM SuperDLT1"
/dev/sg/c2t3l1: Tape (/dev/rmt/6): "QUANTUM SuperDLT1"

--
- Kyle

In article 1U_vd.504004$wV.256298@attbi_s54,
kylet @ p a n i x . c o m (Kyle Tucker) writes:

# mt -f /dev/rmt/4cbn status
QuantumSuperDLT tape drive:
sense key(0x6)= Unit Attention residual= 0 retries= 0
file no= 0 block no= 0

If it helps, this is output to the console when this mt command
fails.

[ID 107833 kern.warning] WARNING: /pci@1f,4000/fibre-channel@2/st@1,1 (st65):
for Command: rezero/rewind Error Level: Fatal
[ID 107833 kern.notice] Requested Block: 1 Error Block: 1
[ID 107833 kern.notice] Vendor: QUANTUM Serial Number: . G
[ID 107833 kern.notice] Sense Key: Hardware Error
[ID 107833 kern.notice] ASC: 0xb (vendor unique code 0xb), ASCQ: 0x1, FRU: 0x0 [270]: [ID 336210 daemon.error] cannot determine ring status on QUANTUMSuperDLT10 (device 0, /dev/rmt/4cbn), received errno 5 on second open

--
- Kyle

In article 1U_vd.504004$wV.256298@attbi_s54,
Kyle Tucker kylet @ p a n i x . c o m wrote:
Hi all,
I am trying to get an FC tape library configured and I'm almost
there but for one issue - I can't access the tape drives very well.
Here's what I have.

Sun Enterprise 250 with Solaris 9
JNI FCI-1063 PCI HBA with JNIfcaPCI (v2.6.13 driver) package

Ok ok, have you ever tried to fill a bath tub by ****ing in it!?
Get a new server or be sorry.
If you want working FC_Tape support I suggest that you ask your tape library
vendor for certified, tested and working configurations.

If they can't provide any, dont buy their tape library.

/wfr
Fredrik

--
Fredrik Lundholm
dol @ ce.chalmers.se

In article ,
(Fredrik Lundholm) writes:

Ok ok, have you ever tried to fill a bath tub by ****ing in it!?
Get a new server or be sorry.
If you want working FC_Tape support I suggest that you ask your tape library
vendor for certified, tested and working configurations.

If they can't provide any, dont buy their tape library.

The library was free and I plan on putting it on support once I
am sure it will work. And I can attach this on a Sunfire server
but I don't see where that would help.

--
- Kyle

In article Tq2wd.658383$mD.476@attbi_s02,
Kyle Tucker wrote:

The library was free and I plan on putting it on support once I
am sure it will work. And I can attach this on a Sunfire server
but I don't see where that would help.

One problem is that there are no FC SDLT tape drives. You actually have
SCSI-tape drives behind a bridge. This is a known source of problems.
This usually means that the drives must sit on a loop, not in a fabric.
(which can lead to silent data corruption and/or disruptions of service
for all drives just because one unit acts up)

The other problem might be the FC card in the E250 and its driver,
do you have access to Sun branded cards you could try!?

The switch, in the 2800 days the firmwares were incompatible between
different OEM:s of the same switch even.

From where have you got the 2800!? Have you tried a Sun compatible firmware on
it!? Problem is that Sun started OEM:ing broacades with the 3800 series. Try
a storagetek firmwared 2800 firmware and a JNI 6460N FC card.

Can you access the drives with SCSI direct attach!? that might be a option
if you're not planinng on using all 8 drives simultaneously.

/wfr
Fredrik


--
Fredrik Lundholm
dol @ ce.chalmers.se

One problem is that there are no FC SDLT tape drives. You actually have
SCSI-tape drives behind a bridge. This is a known source of problems.
This usually means that the drives must sit on a loop, not in a fabric.
(which can lead to silent data corruption and/or disruptions of service
for all drives just because one unit acts up)

An ATL FC-230 Card Logs into the fabric as a fabric device.

Your JNI Fibre Card is set for loop, not fabric.
You should change the JNI card to fabric and make the appropriate changes in
the conf
file for establishing static bindings.

In article ,
(Fredrik Lundholm) writes:

One problem is that there are no FC SDLT tape drives. You actually have
SCSI-tape drives behind a bridge. This is a known source of problems.
This usually means that the drives must sit on a loop, not in a fabric.
(which can lead to silent data corruption and/or disruptions of service
for all drives just because one unit acts up)

The other problem might be the FC card in the E250 and its driver,
do you have access to Sun branded cards you could try!?

The switch, in the 2800 days the firmwares were incompatible between
different OEM:s of the same switch even.

From where have you got the 2800!? Have you tried a Sun compatible firmware on
it!? Problem is that Sun started OEM:ing broacades with the 3800 series. Try
a storagetek firmwared 2800 firmware and a JNI 6460N FC card.

Can you access the drives with SCSI direct attach!? that might be a option
if you're not planinng on using all 8 drives simultaneously.

The 2800 was given to me along with the P2000. No clue as to its history,
rev, etc. and I have no docs. I tried this whole setup using a Sun branded
switch, a Vixel Intralink 1000. This resulted in everything being recognized
exactly the same by the fca-pci driver and my being able to write and read
back small amounts of data to the tapes, but any larger amounts result in
this SCSI error.

Dec 20 14:54:01 essex scsi: WARNING: /pci@1f,4000/fibre-channel@2/st@1,1 (st65):
Dec 20 14:54:01 essex Error for Command: write Error Level: Fatal
Dec 20 14:54:01 essex scsi: Requested Block: 6555 Error Block: 6555
Dec 20 14:54:01 essex scsi: Vendor: QUANTUM Serial Number: . G
Dec 20 14:54:01 essex scsi: Sense Key: Hardware Error
Dec 20 14:54:01 essex scsi: ASC: 0xb (vendor unique code 0xb), ASCQ: 0x1, FRU: 0x0

Could this be the result of inaccurate settings in st.conf? Would the
settings for these tapes in SCSI mode be the same for them being used
via a fiber-SCSI bridge?

--
- Kyle

Kyle,

I've configured FC tape devices twice. In one instance a SCSI t oFC
router sat between the brocade switch and the scsi drives. In that case
I had to add st.conf entries to account for the multiple luns seen on
one target. Let me explain. The SCSI/FC router had 1 FC port and 4
SCSI connections. On the server I bound the HBA to the SCSI/FC router
port using a specific scsi target. All the tape drives were seen as
luns on the single target/port.

In another instance the library had 9940 drives with fibre channel
built into the drives. In that case the drives were connected directly
to the switch and the HBA was bound to each drive on different targets
which simplified st.conf requirements.

It is important to note that in both cases the drives were connected to
a fabric and not a loop. I mention that because I noticed "LOOP
Initializations" in your console messages above and I wonder if you
need to set your HBA's to fabric only.

Vic

victim of a drive-by shooting...

2 cups finely chopped very young human flesh
1 cup shredded cabbage
1 cup bean sprouts
5 sprigs green onion, finely chopped
5 cloves minced garlic
4-6 ounces bamboo shoots
Sherry
chicken broth
oil for deep frying (1 gallon)
Salt
pepper
soy & teriyaki
minced ginger, etc.
1 tablespoon cornstarch dissolved in a little cold water
1 egg beaten

Make the stuffing:
Marinate the flesh in a mixture of soy and teriyaki sauces
then stir fry in hot oil for till brown - about 1 minute, remove.
Stir-fry the vegetables.
Put the meat back into the wok and adjust the seasoning.
De-glaze with sherry, cooking off the alcohol.
Add broth (optional) cook a few more minutes.
Add the cornstarch, cook a few minutes till thick,
then place the stuffing into a colander and cool;
2 hours
Wrap the rolls:
Place 3 tablespoons of stuffing in the wrap, roll tightly -
corner nearest you first, fold 2 side corners in,
wrap till remaining corner is left.
Brush with egg, seal, and allow to sit on the seal for
a few minutes.
Fry the rolls:
325° if using egg roll wraps, 350° for spring roll wraps.
Deep fry in peanut oil till crispy golden brown, drain on paper towels.



Lemon Neonate

Turkey serves just as well, and