OK, the thing that SEEMS to solve the problem is to run "mt -f 
/path/to/device status" just BEFORE the tape is loaded into the drive by 
networker. What happens then is that networker can't read the drive, since 
its already in use by mt. mt then returns the status as expected, and 
networker ejects the tape because it couldn't read it. The interesting bit 
is that when you load ANY tape after that has happened, it can read the 
tape just fine! And no, this is no across-reboot persistent. :(

So, who should I talk to? The guy who wrote the QLA HBA driver? Novell? 
EMC/legato? Spectralogic? My local pizza shop?

I'll probably try all of the above. :)

//Oscar

On Wed, 30 Nov 2005, Oscar Olsson wrote:

OO> We currently have a 7.2.1 backup server running on Solaris, with SAIT-1
OO> drives. And now we're adding a 7.2.1 storage node, connected to the same
OO> library. The storage node is running SuSE EL9 SP2, with kernel 2.6 (64-bit
OO> AMD). The drives are connected via a QLogic QLA2340. We are using the HBA
OO> driver that was installed per default by SuSE. The drives show up at boot,
OO> and are accessable through /dev/st0 or /dev/nst0 and so on. However, when
OO> mounting a tape in the drive, it says that it can't read the tape label!
OO> 
OO> This is what is seen in the logs:
OO> 
OO> With dmesg (also in /var/log/messages):
OO> 
OO> st0: Incorrect block size.
OO> st0: Incorrect block size.
OO> 
OO> And from daemon.log on the main solaris backup server:
OO> 
OO> 
OO> 11/30/05 18:01:59 nsrd: rd=sn01:/dev/nst0 Verify label operation in progress
OO> 11/30/05 18:02:56 nsrd: media warning: rd=sn01:/dev/nst0 reading: Success
OO> 11/30/05 18:03:15 nsrd: media warning: rd=sn01:/dev/nst0 reading: no tape
OO> label found
OO> 11/30/05 18:03:16 nsrd: rd=sn01:/dev/nst0 Eject operation in progress
OO> 11/30/05 18:04:13 nsrd: Jukebox 'Osato' failed: expected volume '006950S1'
OO> got 'NULL'.
OO> 
OO> 
OO> At first, I was thinking that it has something to do with the block size. So
OO> I mounted a volume on the Solaris server in order to check what block size
OO> that volume had. It was 64K, according to the networker admin GUI. So I
OO> configured the SAIT-1 drive on the storage node to use 64K block size, but I
OO> still had the same problem. Then I installed the mt_st package, so I could
OO> use stinit instead. I tried setting the block size to 0 and to 65535, and I
OO> got the same error as above. One interesting thing is that mt always reports
OO> the block size as 512 bytes, like this:
OO> 
OO> 
OO> sn01:/var/log # mt -f /dev/nst0 status
OO> drive type = Generic SCSI-2 tape
OO> drive status = 1073742336
OO> sense key error = 0
OO> residue count = 0
OO> file number = 0
OO> block number = 0
OO> Tape block size 512 bytes. Density code 0x40 (unknown).
OO> Soft error count since last status=0
OO> General status bits on (41010000):
OO>  BOT ONLINE IM_REP_ENs
OO> 
OO> The following is what I have in the st.conf on the Solaris box:
OO> 
OO> tape-config-list=
OO> "SONY    SDZ-100", "Sony Super-AIT1", "SONY_SAIT";
OO> SONY_SAIT       =       1, 0x36, 0, 0x19639, 4, 0x00, 0x00, 0x00, 0x00,0;
OO> 
OO> And this is what I *THINK* should be in stinit.def (I made it myself looking
OO> at EMC/Legatos AIT-3 entry from ftp-legato.com):
OO> 
OO> # SONY SAIT-1
OO> manufacturer=SONY model = "SDZ-100" {
OO> can-bsr scsi2logical drive-buffering
OO> timeout=800
OO> long-timeout=14400
OO> mode1 blocksize=0 density=0x00
OO> }
OO> 
OO> NOTE that I have tried doing this without stinit installed as well.
OO> 
OO> So all this leads me to believe that it might not be related to the block
OO> size after all, although it does complain about it. Any clues of why this
OO> could be? Bad HBA driver? All help is ofcourse appreciated! :)
OO> 
OO> //Oscar
OO> 
OO> To sign off this list, send email to listserv AT listserv.temple DOT edu 
and type
OO> "signoff networker" in the
OO> body of the email. Please write to networker-request AT listserv.temple DOT 
edu if
OO> you have any problems
OO> wit this list. You can access the archives at
OO> http://listserv.temple.edu/archives/networker.html or
OO> via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
OO> 

To sign off this list, send email to listserv AT listserv.temple DOT edu and 
type "signoff networker" in the
body of the email. Please write to networker-request AT listserv.temple DOT edu 
if you have any problems
wit this list. You can access the archives at 
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER