Freenas Volpool unknown

[ciphered23]

Hi

I have a HP Ml10v2 server with 8gb. I have been running freenas for a few years with no issues. I decided to chcuk in another drive and dual boot windows and ubuntu. Howver, to get the grub working i change the Sata mode to AHCI from raid controller. Everything worked and all until today I decided to power on the freenas.

The first boot was done using AHCI enabled mode which worked but no volpool appeared.
Then I recalled changing the raid controller setting, so i put that back on and powered on. The boot had errors, but after resetting the BIOS I was able to load freenas.
However, the volpool is still unknown.

I had a look at the forum, and it looks like setting hw raid, is basically setting yourself up for this disaster I am in.

What are my options, here as this data is very important and the raid was set up in mirrored mode.


Thanks

 

[PhiloEpisteme]

Hi @ciphered23, welcome to the forums.

Can you clarify your disk setup a bit better? On your freenas setup I imagine you have disks as follows

Code:

% zpool status
freenas-boot
  boot disks here
<data pool>
  data disks here

I decided to chcuk in another drive and dual boot windows and ubuntu. Howver, to get the grub working i change the Sata mode to AHCI from raid controller.

On which disks did you install Windows and Ubuntu? Did you use one of the new disks or are you trying to use the same boot disks as for FreeNAS?

Then I recalled changing the raid controller setting, so i put that back on and powered on. The boot had errors, but after resetting the BIOS I was able to load freenas.

Can you tell us exactly which RAID card you're using, what version of FreeNAS you're using, exactly how each of your drives are connected to your system and exactly what each is used for? For example, are any of the drives used for both windows, freenas, and ubuntu or any combination thereof? If you can boot into FreeNAS at all, what is the output of both zpool status and zpool import?

Lastly, I will add that generally speaking dual booting a FreeNAS box is not advised. First because rebooting spins up/down your disks which adds wear and tear. Some folks do dual boot and do so successfully, I'm sure. In such cases I highly highly recommend that the data disks and boot media used by FreeNAS NOT be used at all whatsoever by any other OS you may have installed. The reason is that doing so risks your data; unless you REALLY know what you're doing you don't want any other OS touching your FreeNAS disks.

 

[ciphered23]

Hi PhiloEpisteme

Thanks for helping out.
So I have the following disks
1. USB flash drive for freenas--This is unharmed and working properly
2. 300gb Boot drive for ubuntu-- This drive has been disconnected
3. 2x1tb nas drives, not used for any boot.-- These were plugged in but never used in any other OS.

My idea was to move the freenas to a KVM inside Ubuntu. but never got to passing through these drives.


I have Hp b120i raid controller that comes by default with HP ml10v2 server.

Here's the status

root@freenas:~ # zpool status
pool: freenas-boot
state: ONLINE
scan: scrub repaired 0 in 0 days 00:01:42 with 0 errors on Fri Apr 24 03:46:42 2020
config:

NAME STATE READ WRITE CKSUM
freenas-boot ONLINE 0 0 0
gptid/f722ad79-b9c6-11e4-a379-3ca82a4b6368 ONLINE 0 0 0

errors: No known data errors


root@freenas:~ # zpool import
root@freenas:~ #


No output of the import command.

 

[PhiloEpisteme]

So, from your output your system is not recognizing the data pool you had. Does your system recognize your data disks at all?

 

[ciphered23]

Yes, the sytem recognizes the disks.

 

[PhiloEpisteme]

My guess is your RAID card is somehow obscuring the zfs partitions from FreeNAS. The fact that the system recognizes the drives but zpool import does not list your pool suggests that ZFS does not see the zfs partitions it expects. Do you see any partitions on the disks?

 

[ciphered23]

I think i know a little more about the issue.

I noticed theres a drive failure log from raid controller. while I was booting up.

The issue seems to me, is that when i reneabled raid, it had 3 drives plugged in, including the OS drive, which was not in the original Raid 1 config
So i plug it back in and boot and i get the following errors.

You are right, the issue is raid configuration, so now i will try to reconfigure the raid as original raid 1 config, with just the 2 1xtb drives.

 

[PhiloEpisteme]

I hope you get it sorted. Once you do, you may want to consider not using a raid card anymore or flashing it to IT mode.

 

[ciphered23]

no luck even after configuring the raid controller.... any surface scanning options to recover the data ?

 

[PhiloEpisteme]

I'm not sure. Just to get this straight, you used RAID1 with your two data disks on the raid controller or did you expose each disk individually? What was your pool designed like? How many disks did FreeNAS think was in the pool and what vdev type did you use?

 

[ciphered23]

To be honest, I don't remember which config I used previously. There were 2 drives in the pool, in raid 1 config in freenas. But im not sure what was the hw config i had used. I have nw tried raid 0 on hw as well as raid 1 , but none of them work.

 

[Heracles]

I have nw tried raid 0 on hw as well as raid 1 , but none of them work.

Hey @ciphered23,

You probably destroyed your tags while playing with this... When creating a virtual hard drive in a raid controller, it usually initializes that virtual drive, which is marking its physical drives with some unique IDs. These ones removed any previous ones that may have been there.

Unfortunately, ZFS is a very complex filesystem and if ZFS can not do it by itself, it usually means that nothing else will. From what you described, you were in a very unstable situation and by playing around it with trial and error, you may well have push it beyond recovery.

Hope you did your backups as you should always do no matter how good your main server is ? Now is time to do a restore... Remember that a backup is not working until it has been restored successfully at least once...

In case you had no backup, or one that you can not restore, lets hope it will be lesson learned for your next data storage system...

 

[ciphered23]

any surface scanning software or data recovery company, i can use to recover the data ?

 

[Heracles]

any surface scanning software or data recovery company, i can use to recover the data ?

Yes and No... Any software that can read the drive at low level can return the raw data the is written on it. The thing is, that data is almost as good as a cryptogram despite being clear text. The ZFS structures, all its checksums and everything, that is so complex and obscure to read that if ZFS itself can not read it, you will not be able to read it either. ZFS is not like FAT where one file is often in a single stretch on the drive and can be extracted as it. To rebuild a file in ZFS, you need to point to multiple blocks on multiple drives to get some bits from each. Which blocks, on which drives, for which bits, ...

So Yes, you can extract the content but No, you will surely not make any sense of it.

 

[no_connection]

I am not sure how you had set up the drives at the beginning when you made the FN build.
But you need to find out if switching to RAID or make a RAID array or whatever you did, if that change or destroy any data on the drive and if so how much.
One way to try is to fill a drive(not one with data but a scrap one) with a pattern from start to finish, add the drive like you added the drives to run FN and then check what sectors if any changed.
As far as I know there is a 2GB swap area at the start of the drive, so if anything got wiped like first 100 sectors or something then it might just be pool info and identifiers etc, I don't know enough about ZFS but since it stores master block littered everywhere it might just pick something up as long as you have "pool" data from a new pool spliced with bulk data.
It's a long shot so throwing it out there for last.

 

[ciphered23]

Is it possible to get some expert to have a look at this using webex or zoom ? I can pay for the time involved just to ensure I'm not missing anything.

 

[Heracles]

Hey @ciphered23

First, from what you did and the fact that FreeNAS can not see the pool anymore, it is almost 100% sure that you destroyed your pool. As you wrote yourself, what you did is highly discouraged, very high risk, with the very consequence of ending up with a destroyed pool.

As for working on that interactively, not only it is almost sure there is nothing to do, but trying to do it would expose us to professional responsibilities. Doing anything ourselves directly in the pool would expose us to gigantic risks.

Sorry about your situation, but you really did it : you created a highly unstable situation, you tried to make it even less stable by virtualizing it even more and at the end, everything collapsed. On top of that, you had no backups despite no single server, FreeNAS or other, can be more than a single point of failure.

Last thing will be for you to learn from that one and do better next time, biggest lessons to learn here being to follow guidance and do backups.

 

[ciphered23]

Thanks for the response, that's a lesson that i wont forget.

 

[ciphered23]

I am not sure how you had set up the drives at the beginning when you made the FN build.
But you need to find out if switching to RAID or make a RAID array or whatever you did, if that change or destroy any data on the drive and if so how much.
One way to try is to fill a drive(not one with data but a scrap one) with a pattern from start to finish, add the drive like you added the drives to run FN and then check what sectors if any changed.
As far as I know there is a 2GB swap area at the start of the drive, so if anything got wiped like first 100 sectors or something then it might just be pool info and identifiers etc, I don't know enough about ZFS but since it stores master block littered everywhere it might just pick something up as long as you have "pool" data from a new pool spliced with bulk data.
It's a long shot so throwing it out there for last.

I was searching for this, and came across another thread, which referred to the GPART boot record issue. One of the recommendation was to add identical new drives in this NAS installation and then, copy the raw data from these old drives. Do you think that can work ?

 

[Heracles]

Hi again @ciphered23,

One of the recommendation was to add identical new drives in this NAS installation and then, copy the raw data from these old drives. Do you think that can work ?

You can always try; you will not drop to any lower than loosing it all, which is where you are right now. Some raid controllers initializes the virtual drive when they create it, overwriting the drives with their own stuff. If the drives have been overwritten, it will not work for sure. But if the controller left the disks alone, it may be of help.

Good luck,