0 Replies Latest reply on Aug 24, 2012 2:54 AM by 957378

    configuring HP Ultrium 3000 LTO5 under Solaris 11

      Hi All,

      I got a new HP Ultrium 3000 LTO5 tape drive connected via a HP P212 controller and I am having some problems.

      Attempting to write to the drive gives I/O errors.
      I'm testing via:
      tar cvf /dev/rmt/0cbn
      and tar cvf /dev/rmt/0

      both result in "unexpected EOF"

      dmesg then shows:

      Aug 23 09:15:52 XXX scsi: [ID 107833 kern.warning] WARNING: /pci@0,0/pci8086,3408@1/pci103c,3241@0/st@41,0 (st0):
      Aug 23 09:15:52 XXX Error for Command: log_sense Error Level: Fatal
      Aug 23 09:15:52 XXX scsi: [ID 107833 kern.notice] Requested Block: 0 Error Block: 0
      Aug 23 09:15:52 XXX scsi: [ID 107833 kern.notice] Vendor: HP Serial Number: L $DR-1
      Aug 23 09:15:52 XXX scsi: [ID 107833 kern.notice] Sense Key: Volume_Overflow
      Aug 23 09:15:52 XXX scsi: [ID 107833 kern.notice] ASC: 0xfe (<vendor unique code 0xfe>), ASCQ: 0xca, FRU: 0xdd
      Aug 23 09:15:52 XXX scsi: [ID 107833 kern.notice] File Mark Detected
      Aug 23 09:15:52 XXX scsi: [ID 107833 kern.notice] End-of-Media Detected

      The drive is detected, cfgadm -alv lists it under /dev/rmt/0
      mt -f /dev/rmt/0 status says things appear correctly

      I originally configured st.conf using tape-config-list values from the HP manual but then found that the solaris self configure does the trick so no tape-config-list gives same results.

      The only thing I added to st.conf was the

      name="st" class="scsi" target=65 lun=0

      it's a SAS drive hence according to man st.conf the target value should start at 65.
      I've been following the st.conf changes with:

      update_drv st

      Is that sufficient or do I need to reboot -- -r every time?

      the hp tools that came with the drive report all the info about the drive and media hence that part seems ok. Device and media tests say all is OK except any write tests result in I/O fails.
      HP even sent me a new drive but same results on both.

      Is there any thing else I can try in terms of diagnostic etc?

      I'm fairly new to solaris so I'm completely out of clues.
      Any help would be much appreciated.



      I've found a bit of info relating to BLOCK SIZE and the ST_NO_RECSIZE_LIMIT flag for the drive config which is to do with variable block size.
      The literature suggests it should be on and I think it is.
      Not sure but it looks like a lead to the problem.

      From the HP LTO-5 documentation:

      +A READ, SPACE, WRITE or WRITE FILEMARKS command found EOT
      unexpectedly. This typically occurs when a drive cannot locate
      the target object on tape because the block count is too great.The
      EOM flag will be set.+

      I also found that the auto config is different to manual config:

      AUTO: 2,0x3B,0,_0x1018619_,4,0x46,0x46,0x58,0x58,3,60,1200,600,1200,600,600,18000;
      MANUAL: 2,0x3B,0,_ 0x18659_,4,0x00,0x44,0x46,0x58,3,60,1200,600,1200,600,600,18000;

      The flag in question there is BCR, also the default tape densities but the selected one (#3) is right.

      Edited by: pirator on 23-Aug-2012 19:45