JorgeB

You best ask for help in the existing plugin support thread, I've never used it.

Enable the syslog server and post that after a crash.

Read speeds from a single device might be faster, write speeds to array with parity will likely max out around 1GB/s or less, using it as cache should have better writes versus array.

SMART looks OK, it was most likely a connection issue.

Unfortunately none of those contains a shutdown sequence, instead of shutdown/reboot press stop and see if the array stops and time how long it takes.

You upload the normal diagnostics after boot, also not the syslog from the syslog server, but that's not what we need, like mentioned above when Unraid does an unclean shutdown it saves new diags to the flash drive, logs folder, those are the ones we need.

Based on the btrfs trans id both images look brand new, see if there's another one on a different disk, post output of: find /mnt -name libvirt.img

Just so you don't get disappointed don't expect NVMe speeds for array assigned devices, Unraid isn't built for speed, I did a test some time ago and even without parity a read check was much slower than expected, it's not a bandwidth issue since it was done on am Intel Xeon with plenty of lanes, and all devices are using CPU lanes:

Disk1 dropped offline, this is usually a connection problem but since it dropped there's no SMART report, power down, check/replace cables and post new diags after array start.

From that thread:

Sorry, this is the serial

Yep.

Could be a device or GUI issue, but those values are known to be inaccurate anyway, even when using identical devices, I would just ignore, keep in speed mode, much more useful.

Jan 12 09:08:43 BadWolf kernel: mlx4_en: eth1: Link Down Jan 12 09:08:43 BadWolf kernel: mlx4_en: eth1: Link Up Jan 12 09:08:52 BadWolf kernel: mlx4_en: eth1: Link Down Jan 12 09:08:53 BadWolf kernel: mlx4_en: eth1: Link Up Jan 12 09:09:26 BadWolf kernel: mlx4_en: eth1: Link Down Jan 12 09:09:26 BadWolf kernel: mlx4_en: eth1: Link Up Jan 12 09:09:26 BadWolf kernel: mlx4_en: eth1: Link Down Jan 12 09:09:26 BadWolf kernel: mlx4_en: eth1: Link Up Jan 12 09:26:04 BadWolf kernel: mlx4_en: eth1: Link Down Jan 12 09:26:05 BadWolf kernel: mlx4_en: eth1: Link Up Jan 12 09:26:05 BadWolf kernel: mlx4_en: eth1: Link Down Jan 12 09:26:05 BadWolf kernel: mlx4_en: eth1: Link Up 10GbE link is constantly going up/down, try with a different cable, also note that max cable length for direct attach active cables is 10m (about 33feet).

It's a GUI issue, if you alternate to speed mode does it show correctly?

It should start writing to disk3 when there's less than 1TB free on disk1, and so on, assuming you're using the default high water allocation method: https://wiki.unraid.net/Un-Official_UnRAID_Manual#High_Water

Back up anything you need to another disk/device, when done let Unraid format the cache and restore the data.

Jan 7 18:01:35 Tower kernel: mce: [Hardware Error]: Machine check events logged Jan 7 18:01:35 Tower kernel: [Hardware Error]: Corrected error, no action required. Jan 7 18:01:35 Tower kernel: [Hardware Error]: CPU:7 (17:8:2) MC1_STATUS[Over|CE|MiscV|AddrV|-|-|SyndV|-|-|-]: 0xdc20000000010859 Jan 7 18:01:35 Tower kernel: [Hardware Error]: Error Addr: 0x0000000a2a4f1080 Jan 7 18:01:35 Tower kernel: [Hardware Error]: IPID: 0x000100b000000000, Syndrome: 0x000000005a020300 Jan 7 18:01:35 Tower kernel: [Hardware Error]: Instruction Fetch Unit Ext. Error Code: 1, IC Microtag or Full Tag Multi-hit Error. Jan 7 18:01:35 Tower kernel: [Hardware Error]: cache level: L1, mem/io: IO, mem-tx: IRD, part-proc: SRC (no timeout) Jan 7 18:28:54 Tower kernel: mce: [Hardware Error]: Machine check events logged Jan 7 18:28:54 Tower kernel: [Hardware Error]: Corrected error, no action required. Jan 7 18:28:54 Tower kernel: [Hardware Error]: CPU:9 (17:8:2) MC1_STATUS[-|CE|MiscV|AddrV|-|-|SyndV|-|-|-]: 0x9c20000000010859 Jan 7 18:28:54 Tower kernel: [Hardware Error]: Error Addr: 0x0000000001150800 Jan 7 18:28:54 Tower kernel: [Hardware Error]: IPID: 0x000100b000000000, Syndrome: 0x000000005a020300 Jan 7 18:28:54 Tower kernel: [Hardware Error]: Instruction Fetch Unit Ext. Error Code: 1, IC Microtag or Full Tag Multi-hit Error. Jan 7 18:28:54 Tower kernel: [Hardware Error]: cache level: L1, mem/io: IO, mem-tx: IRD, part-proc: SRC (no timeout) Hardware errors being detected, looks like CPU related.

You didn't specify the partition, it needs to be /dev/nvme1n1p1 These are irrelevant for recovery.

When the second device was first added filesystem crashed and went read-only, that would be a good time to make a backup if one didn't exist, since the device never finished adding pool is now in a messy state, there are some recovery options you can try here.

Like mentioned there's nothing to change to enable legacy boot in the flash drive, see if it boots in a different system if available, could be a BIOS issue.

Please don't cross-post, see reply on your other post.

Start here: https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=819173

New.

E.g. for typical Asus BIOS