priapus

Recently, my NixOS system has started taking around 2 minutes to boot. The delay happens when stopping system-udevd and multiple USB errors are thrown during it. I've tried booting with zero usb devices connected, but the issue still occurs. I have also tried updating my mobo, disabling XHCI handoff, legacy USB support, and the builtin wireless chipset. I haven't even been able to identify what USB port could be causing the issue. Motherboard is an MSI X670E TOMAHAWK WIFI.

Hastebin dump with the journalctl, lsusb -t, lsusb, lspci, and usb-devices output

Full dmesg dump

I would really appreciate any help, I've been losing my mind trying to fix this.

Edit: I looked at some dumps on linux-hardware.org, it seems likely that its the bluetooth chipset causing the issue. I'd already disabled this in the motherboard settings, but that apparently isn't enough.

If anybody knows how to make systemd-udevd ignore this device, I would really appreciate it.

Edit 2: Stumbled onto this post from someone with a Gigabyte X670 which seems to use the same BT chipset, which led to this post which recommend unplugging the computer for a bit.

I left it unplugged overnight and lo and behold the next morning the issue is gone. What a bizarre problem.

I am upgrading my drives and have created a new pool. The original pool had a drive fail, but it has since been replaced. Each drive CKSUM value is at 4.04k and pool had two files with permanent errors. I deleted those, but it now shows this:

errors: Permanent errors have been detected in the following files:

        zfsa:<0x8220>
        zfsa:<0x8056>

When I try to create a snapshot and send it to the new pool. After a few terabytes, it fails with this error warning: cannot send 'zfsa@zpool_transfer': insufficient replicas.

CKSUM was always at zero until the first drive failed. The data is not important and I don't care about whatever is corrupt, I just want to get the data to the new pool.

Edit: Forgot to mention, I have scrubbed the pool many times.

Another edit: I've also tried using zpool clear as well. Despite the high checksum errors I've had no issues outside the two now deleted files. The pool is used for a media server which has been working perfectly.

After I scrub the checksum errors go to zero, but they start climbing when I run the send/receive.

Edit:

Finally remembered to come edit this post with the solution I got from the Practical ZFS forums. Thanks to Jim Salter over there for the solution!

You can bypass the checks by settings a ZFS module parameter. A way to do this temporarily is with echo 1 > /sys/module/zfs/parameters/zfs_send_corrupt_data, which will last until you reboot. After doing this, sending the data, and a scrub, everything was fine!

I was looking to upgrade my storage and was recommended to go with used SAS drives and an LSI SAS controller. I purchased an LSI 9211-8i HBA, 8TB Seagate Exos drives, and these cables. The drives are not spinning up at all when connected to the power supply. Are these cables not the right choice for this?

Edit: I have confirmed that a regular SATA drive works if connected with an SFF-8087 to Sata cable. Either I've somehow received 10 dead drives or I'm not powering them right.

Edit 2: I'm guessing its related to this: https://forums.unraid.net/topic/84038-so-your-new-sas-or-sata-drive-wont-start-spin-up/

What an irritating issue.

So my options seem to be:

• Tape mod the SATA power adapter or drives to prevent pin 3 from disabling the drive - cheap but tedious
• Use a molex adapter to power the drives - less cheap, I'd need new PSU and SAS cables
• Upgrade my PSU to one that supports SATA 3.3 - I'll probably do this. I just bought a new PSU for the purpose of powering these drives, so I can still return it.