Read Open Error, I/O Error 2

[Arluin]

I have a Sun L100 tape library (ATL jukebox, 3x HP LT02 drives) and I am running Legato NW 6.1.3 on Sun E450 with Sol9.

I'm trying to get my support with Legato activated but in the meantime - when I turn on Auto Media Management and it goes out to label the media I get the error message: "read open error, I/O error."

I have "barcode reader" and "match bar codes" turned on.

Can anyone provide some clues to help me resolve this?

 

[605]

You are most likely talking to the wrong device.

When you installed the jukebox somehow you mixed up the drive name sequence - it must be the one that "inquire" will
discover. As a consequence NW loads one drive but talks to another drive. As a conseqence, it cannot read the label which causes the read I/O error.

Re-Install the jukebox properly should solve the problem.

 

[Arluin]

Given these results from inquire:

murcia: /etc/LGTOuscsi/inquire
scsidev@0.0.0:SEAGATE ST39204LCSUN9.0G4207|Disk, /dev/rdsk/c0t0d0s2
scsidev@0.1.0:FUJITSU MAN3184M SUN18G 1502|Disk, /dev/rdsk/c0t1d0s2
scsidev@1.6.0:TOSHIBA XM6201TASUN32XCD1103|CD-ROM, /dev/rdsk/c1t6d0s2
scsidev@2.4.0:HP Ultrium 2-SCSI F40Z|Tape, /dev/rmt/0cbn
scsidev@2.5.0:HP Ultrium 2-SCSI F40Z|Tape, /dev/rmt/1cbn
scsidev@3.2.0:QUANTUM SuperDLT1 4646|Tape, /dev/rmt/3cbn
scsidev@3.6.0:OVERLANDNEO Series 0417|Autochanger (Jukebox)
scsidev@4.4.0:ATL M2500 3.07|Autochanger (Jukebox)
scsidev@4.5.0:HP Ultrium 2-SCSI F40Z|Tape, /dev/rmt/2cbn
scsidev@6.0.0:FUJITSU MAN3184M SUN18G 1502|Disk, /dev/rdsk/c2t0d0s2
scsidev@7.0.0:SEAGATE ST39204LCSUN9.0G4207|Disk, /dev/rdsk/c3t0d0s2
scsidev@7.2.0:FUJITSU MAN3184M SUN18G 1502|Disk, /dev/rdsk/c3t2d0s2
scsidev@7.3.0:SEAGATE ST39204LCSUN9.0G4207|Disk, /dev/rdsk/c3t3d0s2

We've only configured the ATL JB and the Ultrium drives in jbconfig.

Are you saying we have to get the 0cbn order correct (e.g.: 0bnc would be incorrect) or are you saying that my devices as listed in nwadmin:
/dev/rmt/0cbn
/dev/rmt/1cbn
/dev/rmt/2cbn

don't match what inquire is discovering?

 

[605]

The question is whether the seguence inquire discovered is also the sequence of drives for the jukebox. The easiest way to find out it simply to load a media in each drive and really look which drive is used when you for example run an mt command. Then verify whether Networker will use the same drive if you do an inventory.

You may also diable the two other drives so NW does not have a chance to insert the media into another drive. Then use the "mt ... status" on each device until you find the one that responds.

Hope you can look at the drives in your jukebox ...

 

[wallace88]

Confirn that the drive ordering is correct.

Drive ordering means that 0cbn should be the first tape drive in the jukebox (j/b), 1cbn points to the second tape drive, and 2cbn points to the third tape drive.

To do this you should do this when you're next to the j/b, but not critical. Also, the following should be done on the backup server:

First, reset the j/b with: nsrjb -HEv. When it ask you which j/b, choose the ATL j/b.

Next run nsrjb. Again, choose the atl j/b. Confirm that the tape drives are unmounted. Then run the following to confirm that the drives are unmounted:

mt -f /dev/rmt/0cbn status
mt -f /dev/rmt/1cbn status
mt -f /dev/rmt/2cbn status

Go into the following directory: cd /etc/LGTOuscsi

Now we start to verify drive order. Load the tape in slot 1 into drive one, and then verify whether the o/s confirms the same:

./sjimm 4.4.0 slot 1 drive 1 ! this tells the j/b to move the tape in slot 1 to drive 1

mt -f /dev/rmt/0cbn status
mt -f /dev/rmt/1cbn status
mt -f /dev/rmt/2cbn status

One of the mt commands should report that it has the tape mounted. Which one? It should be 0cbn.

Unload the drive with: ./sjimm 4.4.0 drive 1 slot 1

Repeat the above with drive 2 and 1cbn, and again with drive 3 and 2cbn, each time recording what drive the mt command says the tape is loaded.

When you're finish, you should have a mapping of what 0cbn, 1cbn and 2cbn actually points to which physical drive in the j/b.

If 0cbn maps to the first drive in the j/b, 1cbn points to the second tape drive, and 2cbn points to the third tape drive, then the drive ordering is correct.

If it doesn't, then either reconnect the scsi cables so that the drives are in proper order, or delete the drives and j/b from networker, then rerun jbconfig. When jbconfig asks you what is the path of the first, second and third drive, then specify the real order as you determined from the above tests.

 

[Arluin]

D'oh! So the fact that Legato reports the 'physical' device names in the Devices box doesn't mean those are the actual devices that Legato is using!

In otherwords, just as you both said, the device order was incorrect. Thanks for the great posts. Currently successfully labelling my tapes...

 

[onefortheroad]

for what it's worth, and i know this thread is probably old but from the look of the output from the inquire the answer is yes, the devices are seen in the wrong order. the default config for this library is: Robot, D0,D1 etc
you have the robot on the second HBA:
scsidev@4.4.0:ATL M2500 3.07|Autochanger (Jukebox)
scsidev@4.5.0:HP Ultrium 2-SCSI F40Z|Tape, /dev/rmt/2cbn

the other 2 devices are on scsidev@2.....

a swap of the cabling should clear this if anyone else encounters the same.