SOLVED TrueNAS Scale - unable to boot GRUB unaligned pointer

[trilepton]

I was running TrueNAS scale Bluefin (not sure of exact version but it should be latest one). I'm running the boot pool on a usb flash drive which I know is not recommended but had been working fine for me so far. I tried mirroring the flash drive with a second new flash drive from the TrueNAS UI. As far as I can tell the new device was attached successfully. I also took the opportunity to upgrade an older data pool to use new ZFS features (I don't think it's relevant but mentioning just in case). Also, I have an LSI 9211-8i HBA card in the system.

After doing the above, I tried to restart the system to verify everything is working. However, booting the first time led to endless cycles of the HBA initializing and detecting the disks attached to it, and the Intel Boot Agent running and never getting to the boot options menu. I powered off the machine and tried booting again, leading to the message (after HBA is initialized):

GRUB loading..
Welcome to GRUB!

unaligned pointer 0x7470672c
Aborted. Press any key to exit.

Further attempts to boot result the same error. How can I resolve this issue?

System details:
Supermicro MBD-X10SLL-F-O
Intel Pentium G3220
8GB Single DDR3L 1600MT/s PC3-12800 DR x8 ECC

 

[jlpellet]

Not a scale user so ymmv, but I'd try the following:
1. Power down, remove the new boot drive & boot with just the old boot drive.
2. put the old boot drive on a mothertboard port (not HBA) & boot
3. Assuming you've saved config, install to a fresh boot drive.
4. remove the hba card & boot.
Good luck.
John

 

[Etorix]

I also took the opportunity to upgrade an older data pool to use new ZFS features (I don't think it's relevant but mentioning just in case).

This is VERY relevant because if you upgraded the boot pool you just broke GRUB.
Reinstall TrueNAS.

 

[trilepton]

This is VERY relevant because if you upgraded the boot pool you just broke GRUB.
Reinstall TrueNAS.

The pool that was upgraded is a normal data pool, not the boot pool.

Not a scale user so ymmv, but I'd try the following:
1. Power down, remove the new boot drive & boot with just the old boot drive.
2. put the old boot drive on a mothertboard port (not HBA) & boot
3. Assuming you've saved config, install to a fresh boot drive.
4. remove the hba card & boot.
Good luck.
John

I tried removing the new boot drive and just booting the with the old one. Same error is shown.
The HBA is only connected to data disks. The boot drive(s) are directly connected to the motherboard.

How is a mirrored boot pool supposed to work in terms of device selection in BIOS? Can any of the drives in the boot pool be the first one in the order?

 

[jlpellet]

Regarding the boot pool drive order, my experience is too old to be useful. After the now long-ago change to regularly write to the boot pool, I moved all of my boot drives to SSD (via internal SATA, if available, or USB external via USB-SATA case) and, with SSD, reimaging the boot drive & restoring from saved config takes <10 minutes. As always, YMMV.

I coninute to recommend 1) tryying to boot without the HBA (or at least with no drives on it powered) or 2) fresh install, assuming you have a saved config.

Good luck.
John

 

[trilepton]

It turns out the original flash drive used for the boot pool was just bad. It was reporting checksum errors and sometimes hangs the entire system during scrubs (which is why I wanted to mirror it). I think when I attempted to mirror it with a new drive, whatever was written to the original drive was corrupted leading to the grub error. This is supported by the fact that when I attempted a fresh install on the original drive (as a single drive, not mirrored), the installation failed during the extraction process. However, attempting the fresh install on another drive (the one I initially was going to add as a mirror) worked.

TLDR: I performed a fresh install and it works. Thanks @jlpellet !

 

[PhilD13]

You just found out why you don't use USB flash drives as boot drives... or any attached USB drives configurations as boot or pool drives for that matter. They work fine until the suddenly without warning don't and you will have a bad day, and also risk losing your data if not outright making it go "POOF it's Gone".

I suggest to either get 2 DOM's if you have SATA DOM ports on the motherboard and mirror those or get 2 small SSD drives and mount them somewhere in the case preferably where there is some airflow if you don't want to use space in the main chassis where the data drives are mounted. If you have a second HBA port on the card you can connect them there or if the motherboard does software RAID/JBOD (in many cases works too), then use 2 SATA ports on the motherboard.