2200 Orphan Mirror Repair - Error

[teek]

I've been trying to rebuild a mirror on a 2200 using the intructions on the Wiki "Mirror Repair for Orphan"

When I enter the command "co de config mirror 100SS 100DD" (in my case, co de config mirror 50000 10000) I get this error:

"Device Configurations require Public Partition Ids"

I've tried doing the "co de config mirror" command with every number I can think of (Partition, Logical Device, Physical, Unique Id) with no success.

Anyone have any ideas? Here's the devices info on my 2200- if that will help. Many thanks.

------------------
1/30/2007 21:58:49 Command: info devices

Logical Device: 10006 Position: 0 JBOD Size (KB): 32296 Free (KB): 22376 Private Mounted
Label:Private Contains system files only
Unique Id: 0x18429CDE7304E723 Mount: /priv Index: 12 Order: 0
Partition: 10006 Physical: 10007 FS Size (KB): 32768 Starting Blk: 515 Private
Physical: 10007 Drive Slot: 0 IDE Size (KB): 156290560 Fixed

Logical Device: 10016 Position: 0 JBOD Size (KB): 32296 Free (KB): 22496 Private Mounted
Label:Private Contains system files only
Unique Id: 0x2823785760A99CBA Mount: /pri2 Index: 13 Order: 1
Partition: 10016 Physical: 10017 FS Size (KB): 32768 Starting Blk: 515 Private
Physical: 10017 Drive Slot: 2 IDE Size (KB): 156290560 Fixed

Logical Device: 10000 Position: 1 JBOD Size (KB): 153281104 Free (KB): 153127512 Public Mounted
Labelrive1 Single disk
Unique Id: 0x18429CDE7304E723 Mount: /0 Index: 0 Order: 2
Partition: 10000 Physical: 10007 FS Size (KB): 155406176 Starting Blk: 110420 Public
Physical: 10007 Drive Slot: 0 IDE Size (KB): 156290560 Fixed

Logical Device: 50000 Position: 2 MIRROR_CRACKED Size (KB): 153281104 Free (KB): 85928912 Public Mounted
Label:Mirror HAL10000
Unique Id: 0x7928060419F03C3F Mount: /2 Index: 2 Order: 3
Partition: 10010 Physical: 10017 M 50000 Size (KB): 155406176 Starting Blk: 110420 Public
Physical: 10017 Drive Slot: 2 IDE Size (KB): 156290560 Fixed


------------
Command executed without error.

[Hallis]

You might have to break the raid into JBOD first before it will allow you to rebuild the mirror. Try these commands.

"co de predict individual 10000"

or

"co de config individual 10000"

Adding "/force" to the end of those lines might work if they dont work by themselves.

When i got my snap from ebay it was in a broken mirror config because one of the drives had been uninstalled and no settings changes made beforehabd (which the seller didint tell me and i was real pissed)

Shane

[blue68f100]

Your cmd line is wrong, 10000 is the new drive, where 10010 is the original raid.

co de config mirror 10010 10000 should be the correct cmd.

If you go the individual way. Move your original drive to master (top)

[teek]

If I do a "co de config mirror 10010 10000" I get yet a different error:

Unknown Error -4 in configuring a MIRRORRED device with members: 10010 10000

I will give Hallis's suggestion a shot and report back. When you say "If you go the individual way. Move your original drive to master (top)," do you mean if I decide to stick with JBOD? Or if I use JBOD and then try to rebuild the raid?

[teek]

Hallis,

My 10000 drive is already configured as JBOD. I believe this happens automatically after I did the first step of formating and restarting using "co de format 10000 /reint"

[teek]

Okay so I took a guess that what Hallis meant was to take the 10010 disk (the one that still has the data from the mirror) and make that one JBOD. I did so, which left me with two JBOD drives. I then swapped the drive positions (as per blue68f100's suggestion) and rebooted. I was then able to go to the Configure Disks page and choose data protection. It warned me that I would lose all data on drive 2 (the one that had been the orphan). It's now rebuilding the Mirror. Hopefully I will have my data still there when it is finished.

[blue68f100]

Your data will be there, the reason for moving it to the top position. It defaults mirroring 10000(top) to 10010 (bot). This can be changed using a debug cmd.

You should beable to access it while the mirroring is going on.

[teek]

Well it appears that all the data was lost, because now the only share on the new mirror is the default "Share1." I wonder where I went wrong.

Perhaps I should have gone back and used the "co de config mirror" instead of using the disk configuration page. Any thoughts?

I would like to get this broken raid repair instructions fixed so that I know that if I lose a drive in the future, that I can replace the drive and re-create the mirrored Raid.

My plan at this point is to create a new share, move a couple files to it, open the case, break the Mirror by removing a drive, replace it and start the steps over.

Anyone see where I messed up the first time? Thanks for your continued help.

[blue68f100]

Sorry to hear you lost your data.

The co de info gives drive info including Model & SN. When you unplug a drive check to see which one is missing.

These markers repersent the Array ID's
40000 = RAID0
50000 = RAID1
60000 = RAID5

The cmd line utility are very powerful and non forgiving. The graphical interface is easier for most.

[teek]

Not a huge deal, I had the data backed up. I just want to get this process correct so I can be confident in case the thing ever really fails. What good is having a RAID 1 if you cannot restore the mirror, right?

Anyway, I'll post the co de info as I try this again tonight and maybe someone will be able to see what I'm doing wrong. Thanks again for the pointers.

[teek]

Okay, so this is new...when I powered on the 2200 tonight, before re-attempting to break and repair the mirror, I performed a disk check and rebooted. Everything came back clean in the disk check log. When I opened up the default share that had been created yestarday (after the mirror fix) all my data was there in a subfolder! Perhaps I just missed it yestarday when I looked and saw that my share was gone and replaced with the default "Share1." So it looks like when you fix a broken mirror that the system OS removes the shares and user permission settings, but keeps the data after all.

Can anyone confirm that this is normal behavior? The losing of shares and user perms, that is?

[blue68f100]

That's news to me, I never did change the def name. What drive failed, top (10000), or Bottom (10010) closes to MB.

Detail your sequence of repair, maybe it will help use understand what may have happened.

[teek]

Ok sure...here's the background...

Got an old Snap 2200 that I wanted to put into action for file archiving. Always used to use it in JBOD, and was curious how the RAID 1 was implemented. So I set it up to mirror and put about 20 gigs of files on it and set up some shares, users and file security.

I wanted to test what would happen if one of the drives failed (and how I would go about retrieving data and re-establishing the mirror should I have to replace a drive). So first thing I did was open her up and remove the TOP drive (Logical Device: 10000, Partition: 10000, Physical: 10007). I threw that drive into a WXP box and then a YellowDog box just to see if those OSs could read the file system (in case the whole things ever crapped out I needed to retrieve files).

When I powered up the 2200, it showed that one disk had failed, breaking the mirror. All the data was still there on the bottom drive (Logical Device: 50000, Partition: 10010, Physical: 10017) and everything operated exactly the same as before (all shares, users, permissions intact).

I reinstalled the top drive that I had removed and powered it back on. The top drive was recognized as an orphan and much to my dissapointment, I could not simply go into the config page and rebuild the mirror. So I started looking around and found the wiki on this site "Mirror Repair for Orphan."

I followed the instructions but kept getting the error "Device Configurations require Public Partition Ids" that I mentioned in my first post. Not being sure if the instructions "co de config mirror 100SS 100DD" refered to the #s for the Logical Devices, Physical, or partitions, I tried combinations of all 3 to no avail. I looked at your post:

"Your cmd line is wrong, 10000 is the new drive, where 10010 is the original raid.
co de config mirror 10010 10000 should be the correct cmd."

but I got a "Unknown Error -4 in configuring a MIRRORRED device with members: 10010 10000" error.

I then did as Hallis suggested (well actually he had it backwards, but I figured out what he meant) and configured the BOTTOM drive (the one in the Cracked Mirror with the data still on it) as JBOD. So now I had two JBODs.

I then swapped the drive positions as per your suggestion and rebooted. I went to the Configure Disks page and choose data protection. It warned me that I would lose all data on drive 2 (the one that had been the orphan, now on the bottom).

When it was done, I did a reboot and a disk check before trying to connect to it. When I did connect, I saw that only a new default share was there (Share1) and assumed all my data was gone. In fact, it was not. My data is there, in a subfolder of Share1, but all my shares, users, and file permissions were gone. My Network Settings and master password were saved however.

The only new thing I’ve found is that whenever I go to the View File/Folder Security Status page, it says: "This page is not available because no disks were found." This sounds scary but it doesn’t seem to affect anything.

Hope that makes the situation clearer. Thanks again for your help.

[blue68f100]

OK, That is odd in some ways and ???? in others.

Yours is the first to report a problem with Mirroring. The problem you explained with the shares have never been reported before, at least on this forum. Which may explain why this message "This page is not available because no disks were found." is present. This indicates the system is corupted. I do not recall what OS you are on but, you may need to reinstall it. Then I would changed the 2200 back to JOBD, reformat to clearup any drive corruption, then back again. Then run your test again.

You discovered that only another snap can read the HD's. And depending on model it must be the same model and OS. The 2200 OS resides on the HD, It is the only 2 drive unit with the OS on the HD. So you could not move it to any other model.

[Phoenix32]

Yet one more reason to have a 4 drive unit where you can set a drive as a spare.

Speaking of which, and not to confuse or hijack this thead, but I have to ask. I am not a software guy and I have never used RAID 1 (Mirror) ever on anything (just RAID 0, 3, 5, and 6, and once a 0+1), so if this is just stupid and a laugh at me deal, just ignore me. But, with a 2 drive RAID 1, and when 1 drive fails (either drive), aren't you supposed to repair this the same way you do other RAIDs? Meaning, put in a new drive, format it, and then make it a spare, letting the system pick it up and rebuild your arrary? Maybe I am way off here, but this makes sense from what I know of other RAIDs in other type systems (and I know the SNAPs use spares also).

What I am saying here is. Shouldn't the process here been like this;

1. Working RAID 1.
2. A drive fails.
3. You remove bad drive and replace with new drive. (no swapping of the good drive arround either, just leave it right where it is)
4. You format the new drive.
5. You set the new formatted drive as a Spare.
6. The system automaticly picks this spare up and rebuilds the Mirror.
7. Done.

No?

Why not? Thats how it works on all other RAID arrarys (excpet RAID 0 of course) that are capable of spares.

What did I miss here?

[blue68f100]

NOT on the Snaps 2x00.

The snap will init and format, put the OS on it, but will not mirror the two. The resync/mirroring must be done through debug. Remember that SnapAppliance at the time wanted theses units returned to the factory for repair. It wasn't till the 4100 that they made them field servicable. So they did not add it to the OS.

[Phoenix32]

That just sucks.

A workaround maybe, would be to tell a 2000 it is a 4000 (BIOS/CMOS edit)? Then it would do the auto repairs with a spare like it should?

[blue68f100]

Will the 4000 do a auto repiar to a RAID1 or Only RAID 5 ??????

[teek]

I'm on OS v4.0.830. Never did the 860 patch, and not what benifits it has. I'm going to reinstall it over the weekend and re-test.

Still no thoughts on that original error, huh? I hope that it was caused by the OS corruption...we'll see soon enough.

[Phoenix32]

Originally Posted by blue68f100

Will the 4000 do a auto repiar to a RAID1 or Only RAID 5 ??????

Good question, but I don't know the answer (note the question marks on the end of each of those comments). As I said, I have never used RAID 1. But on other RAID systems (servers etc), I have had IT guys tell me RAID 1 will rebuild with spares. So I am making an assumption here and asking. I may have to try it sometime just to find out.

[wgong]

Here is what the Dell 705N (4100) manual has to say about spare drives:

The spare automatically will replace a failed drive in a mirrored pair or a RAID 5 disk array.

[blue68f100]

Let me know if it works in RAID1, it does not in the SNAP OS with a 2x00 unit. But a 4100 is different.

[teek]

Okay really dumb question but how to reinstall OS? I have the original OS 3.4 disk that shipped with the 2200 as well as the 4.0830 upgrade files that I got off the Snap site a few years ago (snap_hlp_40830.sup, snap_jvm_40830.sup, Snap_ServerSync_40830.sup, snap_sys_40830.sup)

Do I simply use the Assit.exe to re-apply the snap_sys_40830.sup update? Is there a specific order to the updates (i.e. sys first, then jvm). I don't need the ServerSync.

As to the other discussion, I've only seen one other self contained RAID setup that did not allow automatic rebuilding of a RAID 1. But in that one, due to the files system, you could simply hook the drive up to another machine and do a direct disk copy. Really nice that SNAP wanted to make a little extra $ by forcing you to use their service dept. if your raid broke.

[blue68f100]

You are not going to like what I'm going to say.
You can not back up major revisions. ( 4 to 3 )

You will need to backup all data on the snap.
Then you need to have a copy of a 2200 v3 image file.

Remove both HD's.
use DD to copy the v3 os to the HDs
Install the HD and start the snap.
run the co de format xxxxx /reinit
if large disk install 3.4.805/7
re format again.
Now you are ready to copy your data back.

There is no reported problem with v4 on a 2200, why do you want to backup?

[teek]

I don't want to backup to v3. You had suggested re-installing my OS due to the apparent OS corruption. I'm just asking how I do that. My current version is still v4.0.830. In order to reinstall 4.0830, do I simply re-apply the snap_sys_40830.sup patch with assit.exe, or do I have to downgrade first? Do I format first then do it? Just wondering procedure. I can't seem to find a search button on this forum to look through other posts on how to reinstall the OS.