Truenas import pool

Hi everyone.

So… Lets start from the beginning.
My Truenas Core 12 running since 2019 in a PowerEdge 1900 died (with a RAID CONTROLLER in IT MODE). The motherboard failed.
So not to loose everything (its a 8 disc RAID-Z1) I take the disks out and put in a HP P2000 JBOD.
After that passthrough the controller to an VM running a new installation of TrueNasCore 13.
The disks are detected, but I cannot import the pool. It tells the message No pools to import.
Tried setting a new VM and import the config backup file that I already have, but the pool show in offline, even with the disks showing.
Anyone can help? I tryied make at least 5 times new vms with the controller passthrough without sucess. The disks show but I cannot import the pool or see it in a restored config file.

Tryed to post images without sucess.

Welcome to the forums. I raised your trust level, you should be able to post images now.

Thanks!

Lets try…

import


dmesg.txt (14.1 KB)

What does the partitioning look like on one of those drives?

 sudo gpart show /dev/da1

It’s better to use SSH than the broken “Shell” in the web UI.

What if you assess available pools by specifying this command:

sudo zpool import -d /dev/gptid

Hi,

It show that information on the 8 drives.
The second part I dont understand.

You are telling to import for just one device? Like zpool import -d /dev/da1?

The result of the show:

truenas% gpart show /dev/da1
=> 63 4294967232 da1 MBR (7.3T)
63 4294967232 - free - (2.0T)

truenas% gpart show /dev/da2
=> 63 4294967232 da2 MBR (7.3T)
63 4294967232 - free - (2.0T)

truenas% gpart show /dev/da3
=> 63 4294967232 da3 MBR (7.3T)
63 4294967232 - free - (2.0T)

truenas% gpart show /dev/da4
=> 63 4294967232 da4 MBR (7.3T)
63 4294967232 - free - (2.0T)

truenas% gpart show /dev/da5
=> 63 4294967232 da5 MBR (7.3T)
63 4294967232 - free - (2.0T)

truenas% gpart show /dev/da6
=> 63 4294967232 da6 MBR (7.3T)
63 4294967232 - free - (2.0T)

truenas% gpart show /dev/da7
=> 63 4294967232 da7 MBR (7.3T)
63 4294967232 - free - (2.0T)

truenas% gpart show /dev/da8
=> 63 4294967232 da8 MBR (7.3T)
63 4294967232 - free - (2.0T)

You have no zfs_member partitions. :flushed:

How do you have MBR and not GPT partition tables?

I have a bad feeling you didn’t do this part properly or safely.

No, it’s to see if using the GPTIDs allows you to view available pools.

That’s moot now, because you don’t have proper partitioning nor a GPT partition table.

Your output does not look like something TrueNAS Core 12 would do.


@HoneyBadger or anyone familiar with using RAID cards: Is the only feasible solution to plug the drives back into the same exact card (in the same exact mode) in order to detect and access the pool?

The partitioning scheme is wrong. MBR table, no zfs_member partitions, no swap, and unusual “free” space at the end.

What does the following show?

zdb -l /dev/da1

Personally I’d want to remove the passthrough element as this could be causing the issue.

Is this passed through in IT mode not RAID mode?

@Johnny_Fartpants

Its an HP H122 HBA, its not even a raid controller, just a SAS controller.

And the command zdb -l returns “failed to unpack label 0” to 3…

Nothing else? No mention of version or pool name?

Do you mean the H221? That is an HBA (based on the SAS2308) but the core of the problem I don’t think lies there, it’s likely an issue at the source:

For clarity - this was the old PowerEdge 1900, with Xeon 5x00 generation processors and mostly likely a PERC 5i?

IIRC, the PERC 5-series cannot do IT/JBOD mode - so it would have required each of those drives to be an individual RAID0.

Correct. It may be possible to access the data with a newer PERC card like a 6i, but we need to be certain on what’s been used to create the partitions.

2 Likes

Well,

Returning to the client today.
I will test some things.

First the controller was not a PERC, was a RocketRaid 2680 I think. The disks all appears as legacy on that, not being configured. So “in theory” they were like passthrough mode!

I will try use the new setup phisically to eliminate any hypervisor problems with the passthrough. If not works I have an old Dell T410 laying around. I will try to put the old RocketRAID and the disk on in to see what happens.

I post the results later!

Thanks in advance for everyone for the clarification!

Just to everyone know.
Tryed restoring the disks in the same order and the same controller on another host and they appears clean, as in the new system.
In the controller all appears normal, as Legacy, so in “theory” in passthrough.

Anyway, as this was a backup unit, I redo all the system in the HP P2000 and configure all over.

It take more then 1 week to complete the backups again, but its working.

Thanks for all the help!