11-04-2010 11:01 AM
Hi I'm attempting to perform a BMR dissimilar disk restore on SUSE Linux 10, using Netbackup 7.0.1.
My original client system has 1 disk drive and I'm attempting to restore the client system with 2 disk drives so I can partition the file system between the two disk drives. Is this considered dissimilar disk restore or dissimilar system restore? (I am aware dissimilar system restore isn't supported in linux.)
Whenever I attempted to perform a prepare to discover from the netbackup administrative console, I would boot my client with the boot media CD and my client would return the discoverable configuration to the master server. I would then initialize the configuration with the discoverable configuration and perform my disk mapping. Next, I issue a prepare to restore and reboot my client with the boot media CD again. As the client runs, I would receive this error: "Disk 'sda' too small, 11317761 KB needed, 0KB available" and the client would send another discovered configuration back to the master server and prompt for another disk mapping.
I'm positive my disks have enough space to hold my file system. I even performed a fresh reformat of the disks before it went into bmr restore. I've also tried reducing my parition and restoring only a limited amount of file systems such that it only uses at most 5gb of the harddrives and has around 54gb of space left over. The same error appears except the needed byte is around 5gb.
Any insights as to what the problem may be?
11-04-2010 03:44 PM
My original client system has 1 disk drive and I'm attempting to restore the client system with 2 disk drives so I can partition the file system between the two disk drives. Is this considered dissimilar disk restore or dissimilar system restore? (I am aware dissimilar system restore isn't supported in linux.)
Are you using a raid configuration for the two drives? Is the only piece of hardware that is changing the hard drives? (ie mother board, CPU etc...) If so make sure you have the necessary drivers.
The 2nd part, I believe the partitions are too small. IDE hard disks (well, devices, actually) are hd<drive><partition>. SCSI devices (and kernel-level emulation of SCSI devices, like USB devices or, in some cases, CD-RW drives) are sd<drive><partition>.
The standard partitioning scheme nowadays is (assuming your drive is hda):
Code: |
hda1 => bootloader and kernel(s) hda2 => swap space hda3 => root filesystem |
Check out
http://www.linux.org/docs/ldp/howto/Flash-Memory-HOWTO/basics.html
11-05-2010 11:31 AM
Are you using a raid configuration for the two drives? Is the only piece of hardware that is changing the hard drives? (ie mother board, CPU etc...) If so make sure you have the necessary drivers.
I actually have three drives in my system. The first two disks are configured with RAID 1 (/dev/sda). The third disk is stand alone (/dev/sdb). I'm restoring to the same system that I've performed a backup on.
The 2nd part, I believe the partitions are too small.
Assume /dev/sda is drive1 and /dev/sdb is drive2.
Before I attempted a dissimilar system restore, I took a full backup of the client that was in disk1. Then I did a restore of some of the file systems. This restore finished successfully.
Test case 1:
Backed up: Disk1 - filesys sda1, sda2, sda3, sda4, sda5, sda6, sda7, sda8
Restored: Disk 1 - filesys sda2, sda5
Next, I attempted a dissimilar system restore, I placed drive1 and drive2 into the system and attempted to perform the same restore, the only difference is the extra disk -- that's when the error about disk1 being too small occur.
Test case 2:
Backed up: Disk1 - filesys sda1, sda2, sda3, sda4, sda5, sda6, sda7, sda8
Restored: Disk 1 - filesys sda2, sda5 Disk 2 - empty
What I dont understand is, I was previously able to perform a full restore on disk1 without any errors. Why is it when I perform a dissimilar system restore, the error occurs on disk1?
EDIT:
I've attempted a full reformat of all my drives and have deleted all my partitions in the partition table and this error still appears. I'm certain it has something to do with the disk parition at this point. However whenever I do a df, it results in no filesystems on the drives. /dev/sda is consistently coming up as 0KB even though bmr discovered /dev/sda/ with 67GB.
11-07-2010 10:17 PM
11-07-2010 10:29 PM
Hi gumbo
Try the below VDO which covered the BMR restore to a vm but you may get some insight regarding your concern. Its a two parts presentation.
https://www-secure.symantec.com/connect/videos/netbackup-bmr-701-release-windows-2008-client-physical-machine-virtual-machine-restore-part1
https://www-secure.symantec.com/connect/videos/netbackup-bmr-701-release-windows-2008-client-physical-machine-virtual-machine-restore-part2
11-07-2010 10:47 PM
>>> Is this considered dissimilar disk restore or dissimilar system restore? (I am aware dissimilar system restore isn't supported in linux.)
I consider the destination hardware is exactly same machine as original expect the two disks of diff sizes. Then this scenario is DDR and not DSR.
Regarding the insufficient disk issue that you are facing, it would be great if you can provide below info:
1. restore log created on master server @ /usr/openv/netbackup/logs/bmrrst/<client_name>/*.log
2. Client configuration. On master, fire "bmrs -o list -r config" query and identify ID of your client config for which you did PTR.
Then run below command,
bmrs -o querytree -r database -table config -id <ID> -gui > cl.conf
Provide this cl.conf file along with the restore log.
Thanks.
-Mandar
11-15-2010 05:52 PM
Thanks for the tip Karam. My disks were already not restricted.
Thanks also for the video links Zahib. I've already watched them one too many times when I was attempting Windows BMR restores.
Mandar, I've attached the restore log and the cl.conf file. I'm using the configuration called 'DissimilarDisk' on a client named 'bsrv7'
Some extra information: I'm guessing the error has to do with the drives' partitioning. When I dropped to a command prompt during the prepare to discover process, I looked into the partitions on my disks. This is the output I got:
> cat /proc/partitions
Major= 8 Minor=0 Blocks = 71041024 Disk = sda
Major=8 Minor=16 Blocks = 71041024 Disk = sdb
> fdisk -l
Disk /dev/sda: 72.7GB, 72746008576 byte .... more info about /dev/sda
Disk /dev/sda doesn't contain a valid partition table.
Disk /dev/sdb: 72.7GB, 72746008576 byte .... more info about /dev/sdb
Disk /dev/sdb doesn't contain a valid partition table.
Which brings me to ask, does the disk have to be partitioned before every restore? If so, why didn't I have to perform a partioning when I was performing a general restore?
EDIT: I attempted another test where I created one partition on /dev/sda and another partition on /dev/sdb. This was done with the 'fdisk' command and I made sure I wrote out the partition table and saved it before I performed the restore. During the disk mapping, I mapped root "/" on /dev/sda and "/var" on /dev/sdb. It still errors out on the same condition: Disk 'sda' too small, # bytes needed, 0 KB available.
Looking forward to a response. Thanks.