2 hours ago, ich777 said:

As already said, maybe it‘s the filesystem.

 

What Kernel was the Ubuntu machine using?

5.15.0

59 minutes ago, peace-keeping-villa8590 said:

5.15.0

This is by todays standards a really old kernel.

 

What filesystem are you using on ubuntu?

2 hours ago, ich777 said:

This is by todays standards a really old kernel.

Comparing the vm kernel version to Unraids, I was really surprised. 

 

2 hours ago, ich777 said:

What filesystem are you using on ubuntu?

Ext4 I think but it also says "Contents FAT..."

4 hours ago, peace-keeping-villa8590 said:

Ext4 I think but it also says "Contents FAT..."

Have you yet tried another filesystem on Unraid?

Did you yet try the other things that I recommended?

4 minutes ago, ich777 said:

Have you yet tried another filesystem on Unraid?

 

Drive formatted as NTFS, xfs and no filesystem. 

 

8 minutes ago, ich777 said:

Did you yet try the other things that I recommended?

I am convinced it's a problem with the firmware of these drives. Thanks for actually helping me out trying to figure out what the issue but I don't have the hardware required to update the firmware and I'm pulling these drives. 

I have removed all but one of the Dell EMC drives and I was doing more searching around and found the a driver from dell for these drives from their website. @ich777 I think this might be what is needed. https://links.homeserver.co/vigWT 

On 1/31/2025 at 7:51 PM, peace-keeping-villa8590 said:

Did you yet try the other things that I recommended?

I used dd to do write test in Unraid. 

On 1/28/2025 at 12:41 AM, JorgeB said:

This is very strange since one of the devices is performing much faster than the other ones, I assume that if you swap cables/ports, the issue follows the devices?

I moved a drive to a different backplane and now the drive ^ is the only device connected to this HBA (9300-16i) 

The drive is Dell EMC branded but will not accept the firmware. 

During some research, some of which was saying SAS SSD cacheing is handled by the HBA. My theory is, the SSD's firmware isn't allowing the drives cacheing process to work since it's connected to a non-Dell HBA. Thought's?

It's a possibility, and it would make sense.

To test this, I bought https://links.homeserver.co/aCZSy It's already flashed to IT mode. 

I received a Dell H330 and no change in performance. I did some research on the controller these drives use and from what I gathered, it uses a Samsung proprietary in house controller. I thought it was interesting because newer models use sourced controllers. 

I would recommend trying with another Linux OS running a similar kernel, if it's the same, there won't be anything that LT can do.

On 2/17/2025 at 1:36 PM, JorgeB said:

I would recommend trying with another Linux OS running a similar kernel, if it's the same, there won't be anything that LT can do.

After initially using these drives and couldn't find what the problem was, I used them with TrueNAS Scale and they worked as expected for a few a while. When Unraid 7 came out, I decided to get it another try since I didn't like the lack of docker support in TrueNAS.

10 hours ago, peace-keeping-villa8590 said:

and they worked as expected for a few a while

Does this mean they then also got slow, and which TrueNAS release did you use?

On 2/19/2025 at 8:40 AM, JorgeB said:

Does this mean they then also got slow, and which TrueNAS release did you use?

The same dries were not slow. TrueNAS scale but I don't remember the version. It was up to date but prior to electric eel being released.

The prior release was Dragonfish, and that also uses kernel 6.6LTS, same as Unraid, and also the same as Electric Eel, how did you test the drives with TruenNAS? dd may not be a good test.

13 hours ago, JorgeB said:

The prior release was Dragonfish, and that also uses kernel 6.6LTS, same as Unraid, and also the same as Electric Eel, how did you test the drives with TruenNAS? dd may not be a good test.

I didn't save any screenshots of the testing I did. I will boot into TrueNAS again to capture the data and report back. 

Use Electric Eel, it's the one with the closest kernel to Unraid v7.0.0

On 2/21/2025 at 10:19 AM, JorgeB said:

Use Electric Eel, it's the one with the closest kernel to Unraid v7.0.0

I did this test a few times to fill the zfs cache afterwards this is the result. I made a one drive pool, test_pool, and did the fio tests against it. 

That will basically just test the ARC, though the results are far from great, I would recommend creating a single volume pool with one of the slow devices, and then just read/write a large file to it, using SMB or another local device, then compare with Unraid by repeating the exact same thing.

 

 

3 hours ago, JorgeB said:

That will basically just test the ARC, though the results are far from great, I would recommend creating a single volume pool with one of the slow devices, and then just read/write a large file to it, using SMB or another local device, then compare with Unraid by repeating the exact same thing.

 

 

I did the test twice writing 100G to fill the ARC cache then I did the test I posted.

 

While I was in my bios configuring it to boot into TrueNAS, I noticed that the slow drives appear differently in this area of the BIOS. I'm not sure what it means and it could be normal but maybe you know. Maybe the drives aren't being passed through the host the same way as the rest of the drives are.

 

Edited February 22, 20251 yr by peace-keeping-villa8590
Add information

18 hours ago, peace-keeping-villa8590 said:

I noticed that the slow drives appear differently in this area of the BIOS. I'm not sure what it means and it could be normal but maybe you know.

Also not sure, but it is a little suspicious, and it may be the result of firmware/model differences, assuming it follows the devices if you change ports.

 

Post the output from 

lshw -class disk

to see if shows more info