MFI driver - wrong connection speed

[as43031]

Good day!
My drives that are connected to the LSI-9260 controller do not have the correct connection speed (SATA-1), I see it in dmesg:

Code:

pass5 at mfi0 bus 0 scbus33 target 12 lun 0
pass5: <ATA Samsung SSD 860 2B6Q> Fixed Uninstalled SPC-3 SCSI device
pass5: Serial Number S5GANS0N503325D
pass5: 150.000MB/s transfers
pass5: Command Queueing enabled
pass6 at mfi0 bus 0 scbus33 target 15 lun 0
pass6: <ATA Samsung SSD 860 2B6Q> Fixed Uninstalled SPC-3 SCSI device
pass6: Serial Number S5GANE0N203948D
pass6: 150.000MB/s transfers
pass6: Command Queueing enabled
pass7 at mfi0 bus 0 scbus33 target 16 lun 0
pass7: <ATA Samsung SSD 860 1B6Q> Fixed Uninstalled SPC-3 SCSI device
pass7: Serial Number S5GANE0MB03905Y
pass7: 150.000MB/s transfers
pass7: Command Queueing enabled
pass8 at mfi0 bus 0 scbus33 target 17 lun 0
pass8: <ATA Samsung SSD 860 1B6Q> Fixed Uninstalled SPC-3 SCSI device
pass8: Serial Number S418NF0K301059V
pass8: 150.000MB/s transfers
pass8: Command Queueing enabled
pass9 at mfi0 bus 0 scbus33 target 18 lun 0
pass9: <ATA Samsung SSD 860 2B6Q> Fixed Uninstalled SPC-3 SCSI device
pass9: Serial Number S5GANC0N400796J
pass9: 150.000MB/s transfers
pass9: Command Queueing enabled
pass10 at mfi0 bus 0 scbus33 target 19 lun 0

the controller itself says that the drives are connected at SATA-3 speed
how can this be fixed?

 

[as43031]

TrueNAS-12.0-U5

 

[jgreco]

Yes, please don't use a RAID controller. Please run, don't walk, over to the following article:

https://www.truenas.com/community/r...bas-and-why-cant-i-use-a-raid-controller.139/

 

[as43031]

unfortunately I don't have another controller yet

 

[jgreco]

unfortunately I don't have another controller yet

And unfortunately the MFI stuff is known to be a bit flaky, so, a bit of a standoff there.

 

[as43031]

Im replace controller to LSI HBA 9207-8i, but I see a strange result on the speed of reading from the pool SSD (11111)
any idea?

 

[jgreco]

And what strange result would that be? It's not giving you a flat line?

In general, the more things that get introduced between the actual storage device and the consumer, the more variability you're going to get in reported performance. Even a SSD directly attached to a real PC generally doesn't give you "flat line" performance, and here you've got what I'm guessing to be a ZFS based NAS connected to a hypervisor over the network sitting in between.

If you believe that your setup should be capable of more, the traditional way to deal with this is to look at each subsystem individually and make sure the basic elements are all performing as well as they can -- a chain is only as strong as its weakest link. For example, using a RAIDZ1 pool to store VM block data is a bad idea;

https://www.truenas.com/community/threads/the-path-to-success-for-block-storage.81165/

you won't be getting all the benefits of parallelism and may be running into some other issues with RAIDZ. Then, when you've made sure that each subsystem is optimal, it becomes easier to look at the bigger picture to see if there are any meta-problems that may be causing performance issues.

 

[as43031]

my problem is in the blue line, it is only 200 megabytes per second, if I make a pool from one disk, then it shows 600 megabytes per second (44444), this is strange

 

[as43031]

hw.pci.honor_msi_blacklist=0 in loader.conf resolve my problem