Link to home
Start Free TrialLog in
Avatar of Intern
Intern

asked on

Need to make and Image of Redhat PC

We would like to make an image copy of a Redhat Linux server, so we could use the image for disaster recovery.  We tried to use a Ghost, but it messed up the partition table and would not allow use to change partitions around after we re-imaged the PC.

We tried using FeatherLinux to make an image and it worked ok on a test PC, but could not detect the partitions on the server due to the raid configuration, or so we think.

Anyone have any ideas on how to make and image of a Redhat Server?  This is a Dell PowerEdge 2650 with hardware raid (2 logical drives).  Any ideas are appreciated.
ASKER CERTIFIED SOLUTION
Avatar of wesly_chen
wesly_chen
Flag of United States of America image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of Intern
Intern

ASKER

You are suggesting that I take one of the drives out and place in a clean one.  Then when the new drive is imaged from the existing drive I take out the new one and replace it with the original????

Not really sure what you mean?  I just want to make an image like what Ghost does.
> You are suggesting that I take one of the drives out and place in a clean one.  Then when the new drive is imaged from the
> existing drive I take out the new one and replace it with the original
Yes, so those two disks have identical partitions and content. Ghost can do the same thing, too

Usually Ghost should be able to do the image (option: Image All) but some RAID controller driver is not supported by Ghost.

After you do the image, the image disk can only work on the same hardware configuration PC (say another Dell Power
Edge 2650) due to RAID driver, NIC driver, VGA driver.....

Wesly
Avatar of Intern

ASKER

Thanks for the advice, but I do not think this is what we are looking for.  We have 2 servers each with 6 physical disks, 2 on RAID 1, and the other 4 on RAID 10.  So I would need 6 extra HD's just to create an image for these machines.  These disks are not cheap enough to make that an option, since they are the best you can get for the Dell servers right now.

I think I need sometype of a software option that will dump the image to a networked computer for storage.
> 2 on RAID 1, and the other 4 on RAID 10.
RAID 1 and RAID 10 have already do the image (mirror). You have already hard disk redunctancy. One disk failure will still
functioning or any data lost.
All you need is 2 spare hard disks for hot swap.

The rest is backup the network configuration and data to tapes or some other low cost IDE disks.

Regards,

Wesly
Avatar of Intern

ASKER

We want an image copy of the current server so that if a serious failure occurs, requiring a full-install of the OS and DB.  Then we could use this image to dump the OS and database back on the server, then recover the DB from tape
Ok, I use "dd" for Solaris disk image and it serves well. I think it should apply to Linux as well.

As root
Backup (create image)
# dd if=/dev/sda of=/backups/server1.dd   (/backups might be your network disk mounted through NFS)

Restore (from image)
# dd if=/backups/server1.dd of=/dev/sda

Wesly
> 2 servers each with 6 physical disks, 2 on RAID 1, and the other 4 on RAID 10
By the way, you only need to do 3 images, not six.

Wesly
Avatar of Intern

ASKER

3 images per server....  right?
> 3 images per server....  right?
You're right.
However, if your OS, softwares (including database softwares) are on RAID1, and databases are on RAID10,
then ,IMHO, do image for OS disk per server is enough. Because you are going to restore database from tape
anyway.
===> cut to 2 in total.

Wesly
Avatar of Intern

ASKER

yes our database is on RAID10, using a ASM (RAW) disk storage method.  So that would be a plausible way to recover the operating system.
Avatar of Intern

ASKER

After some discussion we are looking into the Disk swapping way of making an image.  I was wondering if you could give me some insight into the correct way to do this.  The DB server would only need one disk.  The application Server is another beast in itself.  The software and all the Oracle Application server information exist on the RAID10 device.

Disks: 2,3,4,5

If we were to make a backup of this server, I am thinking that we would at least have to backup the application server files on this image so we could do a restore from tape.  What we are trying to figure out right now is which disks would have to be copied in a RAID10 configuration?

2, and 3 are grouped together,
4, and 5 are grouped together

I was thinking that if we replaced 2, and 4 with fresh disks then it should work.  But I am not sure.  I know that RAID10 does mirroring and striping, so I don't know what disks are the correct ones to replace.  I know this is way off of the Linux forum subject, but any insight would be helpful.
> 2, and 3 are grouped together,              <======== striping
> 4, and 5 are grouped together               <======== striping
> I was thinking that if we replaced 2, and 4 with fresh disks then it should work
So you should replace (2, 3) or (4, 5). (4 and 5) are the mirror of (2 and 3).

> The software and all the Oracle Application server information exist on the RAID10 device
By the way, could that possible to move Oracle Application and server information to RAID1? (re-install Oracle or
copy over and make a symbolic link). Then this will save you the time for backup/image and restore.

Wesly
Avatar of Intern

ASKER

Thanks for the thoughts, no it would not be possible to move the app server info to the RAID1 because of space constraints.  It is only 36 gigs, and with all the software, programs, and backups of the app server I don't know if it would be enough.  I could however move the information to the RAID1 and store all the backups to the RAID10, kind of defeats the purpose of having the RAID10 though.

Anyway thanks for the thought.