Unsolved
This post is more than 5 years old
1 Message
0
27939
PE2950 - Unplug one of the two RAID1 harddrives as backup?
We would like to restore a harddrive image to our PE2950 by a disk-clonning software (Symantec Backup Exec System Recovery). However, since this is the first time we carry out restoration with this software on this server model, we afraid that the restoration would fail.
Therefore we wonder,
1. Our server is running with 2 harddrives and RAID level 1. Can we unplug 1 harddrive (call this HDD1) as a backup, carry out restoration on the remaining harddrive (call this HDD2), then if succeeds, plug back HDD1 and synchronize the content in HDD2 to HDD1? If this is feasible, what's the proper procedure to unplug, plug, and sychronize the harddrive? Any documentation?
2. Any recommendation on disk-clonning software that is sure to be compatible with Dell PowerEdge 2950 and RAID 1?
pcmeiners
4 Operator
4 Operator
•
1.8K Posts
0
December 24th, 2008 06:00
Use this method not only as an instant restore during imaging should something goes awry, but also if I suspect a major MS patch might cause issues. If something does go wrong, I pull the offending image disk, place the disk which was pulled as a safety in its place, boot in the OS, then place the disk with an image issue back in, it should then rebuild. Documentation, none that I know of, you might search the forum posts.
I use Acronis, make sure you run chkdsk on the source array before removing one of the mirrored disks, cloning/imaging software does not fair well with disk errors.
ahleungathome
2 Posts
0
December 26th, 2008 01:00
So you simply plug in a disk even when the server is booted up and Windows is running? Then the rebuild-process will automatically synchronize disk content from the running disk to this new disk? I know it's hot-plug, but I don't know it's really that convenient......
flortep
114 Posts
0
December 26th, 2008 04:00
Unfortunately there is no "restore" possible at all with only 2 drives.
Neither a Raid0 nor on a Raid1 disk arrangement will allow that.
Don't even take the array apart and supply power to the computer with one drive remaining !!
The very first write cycle (may it be Windows or Linux) will with good chance destroy tables or data so that you will loose probably both drives.
Even the immediate re-arrangement of the array may not cure that situation.
Only in a Raid5 or better cluster (50, 51) there is enough redundant information to rebuild a disk volume completely should 1 drive (and only 1) die or is removed..
Check Wikipedia for "Raid" to find a good explanation on the subject.
Have a nice 2009
flortep
pcmeiners
4 Operator
4 Operator
•
1.8K Posts
0
December 26th, 2008 07:00
Flortep...
No sure where you are getting your info from, but it is not how LSI based raid adapters work, done it it a number of times over the last 10 years.
Ahleung...
Perhaps a more reassuring method for you....
Another method would be to place an IDE/SATA drive in the server, clone the original raid 1 to it, make sure the OS boots into the system on the IDE/SATA drive; at that point carry out the restoration. Should the restoration fail, you would then reverse clone from the IDE/SATA back to the original array. Couple of issues, when the server restarts, depending on the cloning software, you may need to go into the bios and direct which disk the server boots to (eg Acronis as a default want to boot to the destination, not the source), and dependent on the OS, drive mapping can get messed up. I generally remove the original array or the IDE/SATA immediately after cloning and before booting into the OS, after the cloning, to negate the drive mapping issue. Again check the original array and the IDE/SATA for disk errors before cloning.
flortep
114 Posts
0
December 26th, 2008 08:00
yes you're right, sorry for having confused you, my fault. I misunderstood you had a Raid0 array rather than a Raid1.
ahleungathome
2 Posts
0
December 29th, 2008 07:00
We carried out a test. We unplugged one of the two RAID1 HDD, then tried to boot up the system with only one HDD plugged. It worked, both 2 HDD can boot up to Windows on their own.
Since we didn't really trust the hot-plug feature, all plug and unplug actions were done when the server is turned off. It required some manual reconfig in the SAS Config Utility, but it worked.
To test rebuild, we made the two HDD contents became different (by installing some softwares on one of them and not on the other), then we turned on the server with both HDD plugged. But the rebuild didn't take place. The server kept running with only one HDD. (The other HDD's LED is off).
So we tried hot-plug: boot up with one HDD first, then plug in the other when Windows is running. As soon as plugged, the new HDD's LED instantly light up, while Windows kept running smoothly. Everything seemed fine.
And here comes my question: Is the rebuild in progress?
If yes, how do we know when the rebuild has finished? We can't do this trick again until the two HDD contents become identical again. So we need to know when it's finished.