StarWind SAN & NAS Free Appliance software raid issue

[myrandor]

I started using Starwind VSan Appliance on VMware ESXi 7.0u2. It was able to detect and create a pool on our H810.

I also have 2 NVMe Samsung connected to 2 StarTech PCIexpress to NVMe. These cards and discs are detect in VMware, they are configured as passthrough to VSan Appliance and both SSD get detected.

However, they are seen as "Offline". After struggling a bit, I figured that it needed to be put in software raid to work, which I done. I configured them as Raid 0 using SSH and after that it's now visible in the "Manage Pools" tab.

Even if the "Usable Capacity" and "RAW Capacity" are calculated correctly, the "Free space" columns stays at 0B. I created a GPT label with one partition, formatted with XFS, nothing works. I tried to "edit" the pool to validate the calculation thinking it could help, but nothing...

I also found and used the script "/opt/VSAManagementConsole/Scripts/CreateMD0Pool.sh" which results at the same thing.

Is there a way for StarWind VSAN to see NVMe single drive (/dev/nvme0n1) as "Ready" so I can create a stripe directly in the Pool interface? If not, what do I need to do for StarWind VSan to be able to use the detected usable capacity??

[myrandor]

Here's how it appears when I don't create a software raid manually. There's no "bus protocol" or "slot number" for NVMe. Is it normal?

[yaroslav (staff)]

Hi,

Thank you for your questions. Your query appears to be related to StarWind SAN & NAS not StarWind VSAN.
It can be a GUI problem. Are you able to operate further with the disks?
Also, RAID 0 provides no redundancy. Given that, can you format individual NVME cards as XFS to avoid software RAID overhead?

[myrandor]

Yeah sorry, it's concerning SAN & NAS Free. I just updated the subject of the thread.

As for the disks, I just recreated a gpt label, a full xfs partition and formatted it with XFS. Still not detected in the Web Interface.

I really think something is missing in the detection of drives.

I know for the no redundancy of RAID-0. I wanted space and speed. Everything will be backuped/replicated accordingly.

[myrandor]

Another issue, after a couple reboot, I just saw that the bus address changed.

This make the "pool" not detect correctly. Also, if I "rescan", the "database" of StarWind doesn't remove the old disks. Last week I had to re-install SAN/NAS because the disk (HW Raid passthrough) could be detect correctly.

[Michael (staff)]

Hello myrandor,
Thank you for your input in improving our products. I will forward your findings to the responsible teams.

[Michael (staff)]

Hello myrandor,
It was suggested to assign disks to VM as RDM instead of pass-through the controller. Please let us know if it helps.

[myrandor]

Hello myrandor,
It was suggested to assign disks to VM as RDM instead of pass-through the controller. Please let us know if it helps.

That's what I just do... After deleting the first 1000 block using dd, the disks was able to be use. At last!

I also removed the raid hardware to use RAIDZ. I suppose it will be better.

Thanks!

[yaroslav (staff)]

Thank you for reaching back to us on this you can use Maintenance - Check State instead of dd.