RHEL 4 cannot see more than 8 tape drives per HBA

[MattS]

I have an issue where our RHEL4 media server cannot see more than 8 tape drives per HBA.  The only limits can find in regards to RHEL is something like 256 max scsi targets.  The tape library is a NetApp VTL.  All the virtual devices are configured the same but no matter how many times i rescan it only finds 8 per HBA.

 

[root@cntpabk1 ~]# ls -l /sys/bus/scsi/drivers/st/
total 0
lrwxrwxrwx  1 root root    0 Apr 15 12:00 0:0:0:0 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:09.0/0000:0d:00.0/host0/target0:0:0/0:0:0:0
lrwxrwxrwx  1 root root    0 Apr 15 12:00 0:0:0:1 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:09.0/0000:0d:00.0/host0/target0:0:0/0:0:0:1
lrwxrwxrwx  1 root root    0 Apr 15 12:00 0:0:0:2 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:09.0/0000:0d:00.0/host0/target0:0:0/0:0:0:2
lrwxrwxrwx  1 root root    0 Apr 15 12:00 0:0:0:3 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:09.0/0000:0d:00.0/host0/target0:0:0/0:0:0:3
lrwxrwxrwx  1 root root    0 Apr 15 12:00 0:0:0:4 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:09.0/0000:0d:00.0/host0/target0:0:0/0:0:0:4
lrwxrwxrwx  1 root root    0 Apr 15 12:00 0:0:0:5 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:09.0/0000:0d:00.0/host0/target0:0:0/0:0:0:5
lrwxrwxrwx  1 root root    0 Apr 15 12:00 0:0:0:6 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:09.0/0000:0d:00.0/host0/target0:0:0/0:0:0:6
lrwxrwxrwx  1 root root    0 Apr 15 12:00 0:0:0:7 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:09.0/0000:0d:00.0/host0/target0:0:0/0:0:0:7
lrwxrwxrwx  1 root root    0 Apr 15 12:00 1:0:1:0 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:01.0/0000:10:00.0/host1/target1:0:1/1:0:1:0
lrwxrwxrwx  1 root root    0 Apr 15 12:00 1:0:1:1 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:01.0/0000:10:00.0/host1/target1:0:1/1:0:1:1
lrwxrwxrwx  1 root root    0 Apr 15 12:00 1:0:1:2 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:01.0/0000:10:00.0/host1/target1:0:1/1:0:1:2
lrwxrwxrwx  1 root root    0 Apr 15 12:00 1:0:1:3 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:01.0/0000:10:00.0/host1/target1:0:1/1:0:1:3
lrwxrwxrwx  1 root root    0 Apr 15 12:00 1:0:1:4 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:01.0/0000:10:00.0/host1/target1:0:1/1:0:1:4
lrwxrwxrwx  1 root root    0 Apr 15 12:00 1:0:1:5 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:01.0/0000:10:00.0/host1/target1:0:1/1:0:1:5
lrwxrwxrwx  1 root root    0 Apr 15 12:00 1:0:1:6 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:01.0/0000:10:00.0/host1/target1:0:1/1:0:1:6
lrwxrwxrwx  1 root root    0 Apr 15 12:00 1:0:1:7 -> ../../../../devices/pci0000:00/0000:00:06.0/0000:0b:00.0/0000:0c:01.0/0000:10:00.0/host1/target1:0:1/1:0:1:7
-r--r--r--  1 root root 4096 Apr 15 12:21 fixed_buffer_size
-r--r--r--  1 root root 4096 Apr 15 12:21 max_sg_segs
-r--r--r--  1 root root 4096 Apr 15 12:21 try_direct_io
-r--r--r--  1 root root 4096 Apr 15 12:21 version

Unfortunetly i have no other media servers using this many tape drives.  I have the NetApp VTL configured to assign 10 tape drives per HBA port for this media server.

Any ideas?
Thanks!
Matt

[MattS · Accepted solution]

Ok, so i figured this out with help from NetApp support.

Apparently NetApp VTL's virtual tape drives are presented as SCSI2 devices.  So only devices 0-7 are scanned for.  To have RHEL4 scan beyond lun 7 you'll need to make some changes to /etc/modprobe.conf, specifically 'options scsi_mod'.

My original modprobe.conf file had this line before i began:

options scsi_mod max_luns=256



After following their KB article (sorry, not allowed to post it here) and some trial and error, the following line in the modprobe.conf file had fixed the issue for us:

options scsi_mod max_luns=256 default_dev_flags=0x20240

scsi_luns=x was actually the options recommened to me, but on a hunch i used the original max_luns=x by itself and it worked.

Hope this helps somebody someday.

:)
Matt