[SOLVED] Raid error with disk clean

[ewok2]

Hello

I have install a proxmox server to replace an esxi server.
=> I am new to proxmox ;-) and it is much beter than esxi !!!

On this server I have some VM Ubuntu, a pfsense VM and an OMV VM with 2 disk for the data in raid (mirror).
But I get often error and the raid is KO.

cat /proc/mdstat
Personalities : [raid1] [raid0] [raid6] [raid5] [raid4] [raid10]md0 : active raid1 sda[1] sdc[0](F)      3906886464 blocks super 1.2 [2/1] [_U]      bitmap: 26/30 pages [104KB], 65536KB chunkunused devices: <none>

It can be sda or sdc that has the error.
On OMV the SMART state of the 2 disk are Ok

sudo smartctl -t long /dev/sdc => give no error

I have try to use the driver :
virtio-scsi-single -> error after some days
virtio-scsi-pci -> error after some days
lsi -> main disk (not the raid one) does not boot

The 2 disk are Ironwolf ans buy less than a year. Never had problem on esxi

Any idea how to solve?
Thanks

 

[j.theisen]

Hi @ewok2

thanks for posting in the forum!

Can you please share a few details on your storage setup i.e. are you using ZFS, LVM or something different for VM storage.

How are the two mentioned Seagate disks connected to the VM? Are you doing PCI passthrough of the storage controller or are you using regular virtual disk images?

Yours sincerely
Jonas

 

[ewok2]

Hello Jonas thanks for help
No ZFS
The OMV disk is in LVM-thin (on SSD NVMe Samsung 990 Pro 1To (local-lvm)
The 2 IronWolf 4TB are connected in passthrough (scsiblock=1) throught ID ATA, (No passthrough of the full PCIe controler)
With virtual controler virtio-scsi-pci
In the VM (OMV/Debian), disk are managed in RAID1 sofware with mdadm
If you have any idea for some tests I can try them

 

[j.theisen]

Ok i see. So first things first, this is more or less expected, see [1]
This kind of block device passthrough is prone to I/O errors.

If you're up for it, you could try the "regular" passthrough using this tutorial [2]
This uses a different SCSI driver which might improve the situation.

Can you please check your logs inside OMV for any specific errors on the mbraid failure mode.

[1] https://pve.proxmox.com/wiki/Manual:_qm.conf#:~:text=only.-,scsiblock,on host
[2] https://pve.proxmox.com/wiki/Passthrough_Physical_Disk_to_Virtual_Machine_(VM)

 

[ewok2]

Thanks for help

I have removed the scsiblock
And configuer the disk with the doc "2"

after rebuild of the raid it is Ok:

Personalities : [raid1] [raid0] [raid6] [raid5] [raid4] [raid10]
md0 : active raid1 sdc[0] sdb[1]
3906886464 blocks super 1.2 [2/2] [UU]
bitmap: 6/30 pages [24KB], 65536KB chunk

unused devices: <none>

Wait and see if it is stable

 

[ewok2]

Almost 3 weeks without any error, tahnks!!!