JorgeB

I believe it's a known issue with the UD plugin, but you should post there instead:

It has a SATA port multiplier, and yes, you should avoid those.

Don't know what those mean, looks like a plugin issue, best bet is to post in the existing plugin support thread.

NVMe device dropped offline: Dec 23 22:48:36 Zigplex2 kernel: nvme nvme0: controller is down; will reset: CSTS=0xffffffff, PCI_STATUS=0x10 Dec 23 22:48:36 Zigplex2 kernel: nvme 0000:08:00.0: enabling device (0000 -> 0002) Dec 23 22:48:36 Zigplex2 kernel: nvme nvme0: Removing after probe failure status: -19 Look for a BIOS update, the below also helps sometimes, failing that try a different brand/model device or board. Some NVMe devices have issues with power states on Linux, try this, on the main GUI page click on flash, scroll down to "Syslinux Configuration", make sure it's set to "menu view" (on the top right) and add this to your default boot option, after "append initrd=/bzroot" nvme_core.default_ps_max_latency_us=0 e.g.: append initrd=/bzroot nvme_core.default_ps_max_latency_us=0 Reboot and see if it makes a difference.

In this case the syslog is the main source, so it's the same, what I mentioned didn't help much was the array health email you posted earlier.

Difficult to say, it could last days or years, but that basically goes for any disk, just keep monitoring it.

XMP is basically overclocking, so you should avoid that, especially with Ryzen, but I checked before and RAM was @ 2133 MT/s, so XMP was disable.

Dec 23 13:16:33 Rudder2-Server kernel: mpt3sas_cm0: SAS host is non-operational !!!! Same issue.

Did you recreate the docker image?

Yep, I bumped the bug report a couple of times but after that stopped to not be a pain.

Not much point, you can run it with just one DIMM at a time to see if you can find the culprit.

Formatted with type 2 protection You need to remove this: https://forums.unraid.net/topic/93432-parity-disk-read-errors/?do=findComment&comment=864078

Documentation is in multiple places, i.e., about cache there's some info below, including how to replace a device, but like mentioned that part is currently not working. https://wiki.unraid.net/Manual/Storage_Management#Why_use_a_Pool.3F

Yes to both questions, but that controller is a bad option for Unraid, it uses SATA port multipliers and has very limited bandwidth, see here for a list of recommended options:

For these don't look at the raw values, normaluzed value is 100, worst is 100, threshold is 0, so it can't be better than it is, it's still 100 and it never went lower.

Please post the diagnostics.

Delete/rename all *.plg files in /boot/config/plugins, then re-enable one or a a few at a time.

If the problem was just the partition you could rebuild one disk at a time, like this: https://forums.unraid.net/topic/84717-moving-drives-from-non-hba-raid-card-to-hba/ But offset FFFF covers the first 127 sectors, and Unraid partitions start on sector 64, offset (8000), which is where you mentioned the LUKS headers were, and that makes sense, they should be right after the partition start, so you'll be missing that info in the damaged disks.

FYI pool device replacement is broken since v6.9.x

Nothing obvious logged, some strange SMB related errors, pool detected some corruption, which suggest possible RAM issues, crashes are likely hardware related.

Yes, from where you chose to save them in the syslog server.

v6.10-rc3 when released will be using kernel 5.15.x, or newer if it takes a long time which I doubt.

If it was only the MBR that was missing/damaged you could rebuild one disk at a time, since the partition info is outside parity it would be recreated, I never used encryption, but pretty sure LUKS headers are inside the partition, so not possible to rebuild just one or two disks since the info needed to rebuild the LUKS heathers is missing from other disks.

This means the disk dropped. With the same identifier? It will resume if you use the same log file.

Pool is showing 3 slots, if there was just one assigned it's still OK, but make sure you've started the array once before without any device assigned, there can't be a "all data on this device will be deleted" warning for the pool device.