06-04-2012 04:19 PM
Hi All,
I am having ockward scenario :
I am having a media server (B : Not a robot control host) which is showing all the drives MISSING PATH in tpconfig o/p.
Now I go to device configuration wizard and I can see the drives (shred) in the screen. Here, A (consider) is robot control host which can see all drives wih proper path.
Now i go to server B and use tpconfig -multiple_delete -drive to delete all the drives which are having MISSING PATH and having success (The drives are deleted from the media server only and not from whole drive databse as they are seen by robot control host A).
I go backup to device configuration wizard to scan the drives in order to reconfig them but i am not able to see any drive over there now (Server A can see them in wizard ).
My question is where those drive gone ?? I am not having a single drive for that particular robot now? How could I bring them back?
I looks abnormal to me....please help
06-04-2012 04:21 PM
correction......
I am having a media server (B : Not a robot control host) which is showing all the drives MISSING PATH in tpconfig o/p for a particular TLD (other robots can see drives with proper paths.)
06-04-2012 04:40 PM
If volmgr/bin/scan is not seeing the devices (which effectively is what the device wizard runs) then if the os is NOT solaris, then almost certainly, it is a problem between the operating system and the devices, that is, the os cannot see them.
If the OS is Solaris, and scan isn't working, it can be the same as above (os cannot see devices) but it can also be the the sg driver needs rebuilding.
However, first, prove for 100% that the OS is seeing the correct drives AND all the paths work (eg, use mt -f /dev/rmt/xxx stat )
The majority of cases like this have nothing to do with NBU (well, yes, it might need reconfigering) as the root cause is the os missing the drives, or the paths changing. NBU doesn't suddenly lose drives, something has happened at the os level.
So, remember, forget NBU first, and concentrate on the os.
For Solaris to rebuild sg driver , first you need to work out if the drives are configered via WWN or target/ lun (or combination of both)
Example
sgscan shows just x4 tape drives, there should be x6 ...
06-04-2012 04:48 PM
Hi,
The OS is able to see drives but Netbackup sees them with MISSING PATH for that TLD.
Please consider the sequence of actions i have gone through.
Thanks,
Giri
06-04-2012 11:23 PM
can you plz post o/p of below commands.
run on master --- /usr/openv/netbackup/bin/admincmd/nbemmcmd -getemmserver
run on server B --- /usr/openv/volmger/bin/scan
06-04-2012 11:28 PM
OK, if you run /usr/openv/volmgr/bin/scan, on the media server, do you see the drives ?
If you do, delete the drives in NBU, and re-run the wizard.
If not, report back here.
Do the drives work if you use mt -f /dev/rmt/xxxcbn stat (load a tape into the drive using robtest first)
Martin
06-05-2012 12:28 AM
The OS is able to see drives
Please tell us WHICH OS and what makes you confident that the OS can access and use the devices.
Please remember that most OS'es do not remove device files when access is lost.
The test is to remove OS devices, rescan at OS level and see if devices are re-added.
06-05-2012 11:17 AM
@mph999
U mentioned....
OK, if you run /usr/openv/volmgr/bin/scan, on the media server, do you see the drives ?
If you do, delete the drives in NBU, and re-run the wizard.
Exactly what i performed as i have mentioned earlier....check below...
Now i go to server B and use tpconfig -multiple_delete -drive to delete all the drives which are having MISSING PATH and having success (The drives are deleted from the media server only and not from whole drive databse as they are seen by robot control host A).
I go backup to device configuration wizard to scan the drives in order to reconfig them but i am not able to see any drive over there now (Server A can see them in wizard ).
Thanks.
06-05-2012 11:45 AM
We STILL don't know which OS and how/where you see the devices in the OS and how you are sure that they are USABLE at OS level.
As per my previous post:
Please remember that most OS'es do not remove device files when access is lost.
The test is to remove OS devices, rescan at OS level and see if devices are re-added.
NBU can only discover devices that the OS presents to it.
06-05-2012 12:28 PM
You need to follow Mariannes advice in the post above.
Now, when you run the wizard, it effectivly run scan -all.
If the device wizards doesn't show the drives on the media server, usually, this means that if you run scan on the media server, it doesn't show the devices - which is why I am confused when you say scan works, but the wizard doesn't.
We also need to know the os type of the servers.
Martin
06-05-2012 12:30 PM
Maria,
Its Linux.
When I seen the drives in MISSING PATH in NBU...I was able to see them in scan
But when i deleted from server by tp command then they were not found in config wizard too (Before deleting i was able to see them there in wizard as shared drives).
Thanks.
06-05-2012 12:51 PM
Please confirm that devices are accessible/usable at OS level.
Check device names in /dev/rmt.
Next, issue 'mt -f /dev/nstX status' for each of the device names seen in /dev/rmt.
You can also follow the steps in the Linux chapter of NBU Device Configuration Guide http://www.symantec.com/docs/DOC3656 to check/verify devices at OS level.
Extract from p. 76:
If the operating system detects the SCSI devices, NetBackup can discover them.
PS: MISSING PATH status is seen when OS has lost connectivity to devices.
tpconfig command only removed devices from NBU config, which was necessary because of the MISSING PATH status.
06-05-2012 02:29 PM
OK...I got confused between what tpconfig can see and what OS can see.
Now one thning is clear "If OS detects SCSI devices , Netbackup Can discover them".
Means if MISSING PATH comes in future OS should not be able to see the devices attached.
OK i got it...but how i can get them back now (Is there any way from robot control host to share these lost drives to be seen by media server A) or i hv to reconfig from OS only (if yes then how...pls tell).
And is there any situation where OS sees the drives and NBU unable to see.??
Thanks.
06-05-2012 03:16 PM
Have you tried the Device Config Guide as suggested?
Try to reboot the media server.
Check physical connection.
Check zoning.
Check /var/log/messages for errors.
Verifying OS device configuration: cat /proc/scsi/scsi
And is there any situation where OS sees the drives and NBU unable to see?
Only seen this on Solaris where sg driver needs to be rebuilt (as explained by Martin above) but this does not happen 'out of the blue' - only needed as initial step.
06-05-2012 03:20 PM
(1)
OK...I got confused between what tpconfig can see and what OS can see.
No problem ...
(2)
tpconfig is a real bad command for troubleshooting why things are missing.
tpconfig shows what is configered, NOT, what can be seen.
For example, I can configure a drive on my server that does not exist, tpconfig will show it, because it is in the configeration - that does not mean it will work.
(3)
Now one thning is clear "If OS detects SCSI devices , Netbackup Can discover them"
Usually yes.
(4)
Means if MISSING PATH comes in future OS should not be able to see the devices attached.
Yes
(5)
OK i got it...but how i can get them back now (Is there any way from robot control host to share these lost drives to be seen by media server A) or i hv to reconfig from OS only (if yes then how...pls tell)
THe robot control host does prevent other servers seeing the drives
(6).
And is there any situation where OS sees the drives and NBU unable to see.??
Yep, but not usually on Linux, in fact, I've only seen it on Solaris when you first need to rebuild the sg driver.
I have also seen a case where the OS could see the device (it was solaris as it happens) over the san, but was unable to communicate correctly - due to a SAN issue.
Next steps. Back to basic - forget NBU
As per mariannes post :
Check device names in /dev/rmt.
Next, issue 'mt -f /dev/nstX status' for each of the device names seen in /dev/rmt.
You can also follow the steps in the Linux chapter of NBU Device Configuration Guidehttp://www.symantec.com/docs/DOC3656 to check/verify devices at OS level.
For each drive, prove that he OS can see and use it using mt command (you may have to install this as a package if not available, it is a Linux command)
Once you have proved the OS can see the devices, try running scan - if this works, the wizard should then work.
If scan does not work, try this ...
scsi_command -d <device file>
martin
06-05-2012 03:49 PM
Can u help me in below now
Drive with serial no 1304927336 is visible at OS.
[root@AAA ~]# /usr/openv/volmgr/bin/scan |grep 1304927336
Serial Number: "1304927336"
Device Identifier: "IBM ULT3580-TD4 1304927336"
Drive 16 Serial Number : "1304927336"
Now NBU not showing Path for the same serial number
[root@AAA ~]# vmoprcmd -h AAA-autoconfig -t |grep 1304927336
TPAC60 IBM ULT3580-TD4 4C17 1304927336 -1 -1 -1 -1 - - -
[root@PG997 ~]#
other drive shows path well
TPAC60 IBM ULTRIUM-TD3 8711 G353880066 -1 -1 -1 -1 /dev/nst14 - -
Thanks,
06-05-2012 04:04 PM
to be more specific...
tpconfig -d o/p
98 Drive572 hcart TLD(8) DRIVE=16
MISSING_PATH:3:0:5:16:1304927336 DOWN
Thanks
Giri
06-05-2012 09:04 PM
Please show us FULL output of scan.
06-05-2012 11:09 PM
As marianne asked ....
I am also waiting for output of scan command. (i already requested to you)
06-06-2012 12:40 AM
Giri
The reason why we need FULL scan output:
If media server can see the robot, the robot itself will report all the drive serial numbers.
This only proofs that robot can see drives via the control path. So robot knows about the drives and positions so that it can mount and dismount tapes.
That does not proof that OS st driver can see and use the tape drives.
We call this the data path. We need to see if scan reports the data paths, e.g.
Device name: /dev/rmt/xxxxx
and
Passthru Name: /dev/sg/xxxxx.