07-20-2019 07:11 AM - edited 07-20-2019 07:13 AM
Hi dear all,
Scalar i6000 library presented to more than 30 media servers and has 40 drives. one drive has issue and should be replaced with new drive. I need to replace it without any downtime or service affect. restart netbackup services is mentioned in all documents for drive replacement. Also I need to know which method is safest for replace a drive presented to many media servers?
Master server is 7.7.2 installed in linux and media servers are linux and Solaris
Thanks
07-21-2019 01:59 AM
Physically replace the drive, and ensure it is crrectly seen by the operating system.
The official way to then replace the drive in NBU is as the example below:
(1)
# tpautoconf -report_disc
======================= New Device (Tape) =======================
Inquiry = "QUANTUM DLT8000 0250"
Serial Number = PXB08P3242
Drive Path = /dev/rmt/119cbn
Found as TLD(6), Drive = 1
===================== Missing Device (Drive) =====================
Drive Name = QUANTUMDLT800014
Drive Path = /dev/rmt/9cbn
Inquiry = "QUANTUM DLT8000 0250"
Serial Number = PXB08P1345
TLD(6) definition Drive = 1
Hosts configured for this device:
Host = plum
Host = avocado
This shows the old dive as 'missing' and the new drive ...
Then by taking the same drive name as the 'missing device' and the <drive path> from the new device, run.
tpautoconf -replace_drive <existing drive_name> -path <path to replacement drive>
Example
tpautoconf -replace_drive QUANTUMDLT800014 -path /dev/rmt/119cbn
You will still need to stop/ start media manager / volume manager services on each media server that see the drive.
07-22-2019 03:12 PM
It might be not so easy cause Linux uses udev mechanism and it can be such a pain in the neck.
07-22-2019 04:11 PM
Providing new the drive is seen at OS level (and thus appears in scan command) it should be fine, I've done it many many times. We're not playing at OS level here, just effectivly swapping the old serial number and path for the new serial number and path against the drivename in NBU config.
07-24-2019 08:10 AM
@shahriar_sadm - first thing to remember is that NetBackup SSO works via "drive serialization" and by "serialization" we mean "keep track of tape drive serial numbers". i.e. MediaServerA might see tape drive serial AAAAAAAAA as device ID /dev/rmt99 - whereas MediaServerB might see the same tape drive via /dev/rmt100 - or whatever... but collectively across all media servers then NetBackup knows which tape drive is which because of the usually immutable serial numbers
Anyway, you might be lucky... if your i6000 is using device ID spoofing, then the true serial number of the tape drive might not be exposed to NetBackup - in which case then all you will need to do is down the correct drive across all media servers - get engineer to replace drive - and then up the drive again. i.e. if your i6000 is drive ID / drive serial spoofing then you won't have to inform NetBackup of the changed serial number.
However, if you do not have drive ID spoofing configured / available in your i6000 then you will simply have to follow the documentation. There are no shortcuts to this process, no hidden gems that only we know that are not in the documentation. Sorry, no magic steps for this one.
08-06-2019 11:47 AM
I have used the tpautoconf -replace_drive command for many years, but recently - after 7.6, it has not seemed to "stick" - after a few minutes the drive reverts to the old SN and commands fail.
I have to either use tpautoconf -a and recycle netbackup, or use the wizard.
08-06-2019 11:19 PM
I have heard tat before - not sure if it was from Genericus, or someone else. If this should happen, log a call.
I can't understand it - the config is held in NBDB, so it's not like a file doesn't get created or is accidently deleted or something. You cannot accidently delete a row from a database ...
08-08-2019 12:14 PM
Try and keep it simple.
Down the drive in NBU if it isn't already.
Replace the broken tape drive.
Make sure the O/S sees and plays well with the replacement drive.
UP the drive in NBU and run a backup against it.
NBU rides on top of the O/S driver and config...so if the O/S is happy, NBU will generally be happy with it. I've been managing as many as 1250 tape devices across 116 master backup servers for years. I live by the thought process of keeping it as simple as possible until deeper "digging" is required. All else fails, replace the downed drive in the library, and use the GUI Wizard to scan the drives. Try all the simply steps first before you have to get off the trail and into the weeds. :)