07-12-2013 04:49 AM
Hi folks,
I have an issue in one of my environments where NBU GUI does not show any media in a particular drive but when I perform ROBTEST it shows as media being present. I then try the usual unload/move commands with no success. Please see below for output:
s d
drive 1 (addr 500) access = 1 Contains Cartridge = yes
Source address = 1040 (slot 41)
Barcode = LV3489
SCSI ID from drive 1 is 1
drive 2 (addr 501) access = 0 Contains Cartridge = yes
Source address = 1036 (slot 37)
Barcode = KP0375
SCSI ID from drive 2 is 2
READ_ELEMENT_STATUS complete
m d1 s37
Initiating MOVE_MEDIUM from address 500 to 1036
move_medium failed
sense key = 0x5, asc = 0x3a, ascq = 0x0, MEDIUM NOT PRESENT
unload d1
Opening /dev/nst0, on the local host, please wait...
Error - cannot open /dev/nst0 (Input/output error)
This happened on 05/07/13 and subsequently required an Oracle engineer onsite who instructed me that the drive needed replaced due to not being able to read. This was the response:
Drive HP LTO2 SCSCI elemnt ID 500 doesn't detect tapes.
Drive needs to bo replaced .
We had the drive replaced and it worked for around 2 days but now the same issue has arisen. Is it possible that its a tape or library issue?
I thought I would consult with you folks before I start trolling logs to see if there is a repeat offender tape....
Solved! Go to Solution.
07-13-2013 11:21 PM
@Mark - This particular environment does not have a web interface so required a callout as there is only 2 drives and both went down which resulted in complete loss of service and worryingly it has some rather important DB's on it.
The drive however was proven not to be DOA and after a power cylce of library, reboot from UNIX side (I dont have root access) and finally stopping nbpem then deleting drives, ran config and both drives were found and went into active state. It did however take around an hour for nbpem to start filtering jobs through but at least its operational...for the meantime!
Thanks for all you suggestions guys/gals
07-12-2013 04:59 AM
Sounds like a tape drive issue but there are other factors that can cause this.
Is this a shared drive and if so how many servers access it and what O/S are they?
I have seen Windows server lock drives like this, constantly forcing the tape to stay in the drive - usually when the TUR registry key has not been added
I have also seen it happen when there are orphaned bptm processes causing it - worth checking all servers accessing it for any bptm processes (while no jobs are running of course!)
Failing that if it is really locked you could try re-booting all media servers accessing it to ensure the lock has been broken and they try again
Let me know more details and i will advise further
07-12-2013 05:35 AM
Cheers for the response Mark.
Its not a shared drive but the O/S is Solaris.
Checked for orphaned bptm processes and none were present.
All media servers have been rebooted and the issue is still present.
So I'm starting to think its either dodgy tape/s or perhaps a dodgy drive which would be cause for concern as it was only replaced a few days ago.
Unless you have any further tricks up your sleeve?
07-12-2013 05:43 AM
unload d1
Opening /dev/nst0, on the local host, please wait...
Error - cannot open /dev/nst0 (Input/output error)
Solaris? Does not seem to Solaris device name?
Linux perhaps?
Are you 100% sure that d1 is indeed /dev/nst0?
Compare output of:
scan -changer (look for drive position and serial #)
and
tpautoconf -t (look for serial # and OS device name that corresponds with d1 serial # above)
07-12-2013 05:43 AM
You didn't say specifically, but *is* there actually a tape *physically* getting stuck in the drive? Have seen the front bezels on drives being mis-aligned before physically stopping successful ejects.
or perhaps a dodgy drive which would be cause for concern as it was only replaced a few days ago
Not a concern per se - quite frequently refurbished & can be DOA unfortunately.
07-12-2013 06:44 AM
Sounded like an NDMP drive to me to be honest with that path - any of those in use?
If so then the filer is holding on to it.
Check on you switch that nothing else is zoned into that drive too.
07-12-2013 07:28 AM
@Andy - Im not sure if there is definitely a tape stuck as I work remotely (grumble) it was purely an assumption from the robtest. After speaking with the Oracle Engineer he reckons it may be a DOA drive also so he is paying a visit on Monday...and I was unaware that so many replacements are refurbs so cheers. Lets wait and see!
@Mark and Marianne - I will look to try your suggestions later on today (unfortunately we have to access these digital environments through a jumhost and it only allows 1 person at a time) so will update you accordingly.
Hopefully one works and saves an Engineer visit.
07-12-2013 08:19 AM
The Tape library would usually have a web gui that may be able to tell you more as well as assessing of a tape is in the drive and re-setting it which causes an automatic unload with most drives
07-13-2013 11:21 PM
@Mark - This particular environment does not have a web interface so required a callout as there is only 2 drives and both went down which resulted in complete loss of service and worryingly it has some rather important DB's on it.
The drive however was proven not to be DOA and after a power cylce of library, reboot from UNIX side (I dont have root access) and finally stopping nbpem then deleting drives, ran config and both drives were found and went into active state. It did however take around an hour for nbpem to start filtering jobs through but at least its operational...for the meantime!
Thanks for all you suggestions guys/gals
07-14-2013 12:36 PM