05-16-2013 06:32 AM
Hi all,
I have a wiered phenomenon on a NBU 7.5.0.3. media server.
i have connected 20 drives via FC to this host.
Two rives are missing, because they are faulty and have to be changed, those are turned off, so it should not show up at the host at all.
When I do a rescan in windows device manager, everything is fine, there a 18 drives.
When I now do a scan with netbackup!
Don't count the, these are 20!
As you can see 2 serial numbers show up twice!
Here are the details from 1 of those candidates:
What is this?
Port 2/0/11 is the working one, Port2/0/4 is dead, I can see this directly at the HBA!
Why does Netbackup think that the dead drive has the same serial than another drive in the same library?
During the last night I saw lots of these in the windows appl log:
serial number check failed on Drive002 (device 4, SCSI coordinates {2,0,4,0}, \\.\Tape3), drive serial number is F002547019, database serial number is F00256600D, DOWN'ing it
These are IBM LTO5 drives, driver is installed non-exclusive.
best regards
Volker
05-16-2013 06:53 AM
if there any Hardware change then update to Device file....
So!! on the server you ned to rebuild the OS Device Tree...Make sure Device file are pointing to currect Devices.
Run scan and match the output.there should not be any Duplicated drive.
run tpautoconf -report-desc..
delet all device path that are seen by above command...
05-16-2013 07:08 AM
scan only reports what is seen at OS level.
Check for ghost drives. See http://www.symantec.com/docs/TECH93292
05-16-2013 07:11 AM
Hi Ankit,
thanks for helping out.
The OS device tree is ok! Right now there are 18 devices connected to the OS and Windows device manager does show 18 drives! 2 are missing because they are offline right now and will be replaced in the next days.
So when the OS does see the rigth number of currently connected devices, why does the netbackup scan see two phantom devices, with the wrong serial numbers/wrong SCSI IDs?
As far as I know, the scan command does query the OS directly and has nothing to do with the NetBackup device management DBs.
When I do a tpautoconf -report_disc than the device database tells me that 2 devices are missing. wich is correct!
Volker
05-16-2013 07:12 AM
Check it with your SAN admin the zones of your drives.
You have 10 drives with dual path to the net backup host . This is not a problem, net backup can handle dual paths correctly.
on the other hand, 10 drives are not zoned with the net backup hosts.
EDIT: I do not read all your post corectly.
You must check it with your SAN admin, Adan when you are sure that the drives are ok at the OS level, delete all drives from net backup and rerun the configuration wizard.
05-16-2013 07:27 AM
HI all,
no the drives are not dual path, only one path to the host per drive.
Will follow Mariannes advice and look for ghost drives!
Will come back to you with the outcom.
Thanks!
Volker
05-16-2013 07:37 AM
Hmm ok,
deleted some ghost devices from device manager.
NetBackup scan has not changed, still 20 devices, not the 18 that device manager is aware off.
Even in Qlogic SAN surfer, there are only 18 drives connected to the HBA.
Is that a bug in Netbackup scan?
Volker
05-16-2013 07:46 AM
Did you reboot the server per the technote?
"Prior to re-configuring any devices for use with NetBackup, ensure no additional ghost entries have been created following the reboot of the server"
05-16-2013 07:50 AM
Hi wr,
no I can't reboot the box it's a production server.
This is confusing, I can't reboot a server every time a drive gets broken and we have to replace it?
Volker
05-16-2013 08:30 AM
what library you have?
Most new libraries mask the serial number of a drive, so it gives to the host the same SN for the same drive position regardless how many drives you change,
05-16-2013 08:43 AM
We don't have this issue with Unix servers (hint)