Hello

I'm hoping someone smarter can me can point out something obvious I haven't been able to track down.

I have 'recently' been trying to get my rig migrated to a new chassis and update some hardware and it has NOT been going well

I moved from a Chenbro chassis that had a port multiplier on it's backplanes to a new Hako-Core chassis which utilizes 4-disk backplanes (similar to a Storinator) and I have yet to have the system running in a stable fashion.

The current setup is a ASRock Rack ROMED8-2T with a AMD EPYC 7282
I'm utilizing 2x Broadcom branded LSI 9305-24i to wire into the backplanes (System Devices list them as: Serial Attached SCSI controller: Broadcom / LSI SAS3224 PCI-Express Fusion-MPT SAS-3 (rev 01))

All drives are visible in the system, I'm able to start the array without issue

Where I begin to see problems is a short time after starting up my docker containers

I have 2x Samsung drives in a BTRFS pool for my main cache, this is where my appdata lives

After some period of time, usually within 5-10 minutes my syslog begins being bombarded with various errors such as:

sd 34:0:15:0: No reference found at driver, assuming scmd(0x000000009270a119) might have completed

BTRFS error (device sdah1): error writing primary super block to device 2

There are others but my log did not scroll back far enough, one is related to a "Power-on or device reset occurred"
Admittedly, I had a previous forum post regarding this same error, but not the others

I have since then replaced both SSD drives that are a part of this cache pool, one at a time, allowing the pool to rebuild in-between

I have also replaced the HBA's I was using with new cards from Amazon in the event the original eBay cards were the culprit

I have tried moving the drives to different backplanes, which use different SAS cables to different ports on the HBAs

The issue persists

It seems to be isolated to the SSDs, I do have other SSD pools but they (at least so far) have not gotten enough I/O on them (due to docker containers failing/locking up during these occurrences)

I am frankly running out of ideas at this point and I can't keep throwing money at it hoping "maybe this will be the fix"

I'm curious if, at this point I could have possibly gotten a bad replacement for one (or more?) of the SSDs despite running preclear testing on the both

Also, I'm wondering if in my ignorance if there's something specific with the HBAs that may need to be configured? I've never had an experience where I'd had to do that so I'm not sure where to start. As best I can tell, these cards should be in IT mode as I received them?

I'm attaching diagnostics from today's latest foray into trying to make things work

If anyone can shed some light on my situation I will be eternally grateful

unraid-diagnostics-20250816-1750.zip

Try connecting the SSDs to the onboard SATA ports, they are usually better for SSDs, and retest.

10 hours ago, JorgeB said:

Try connecting the SSDs to the onboard SATA ports, they are usually better for SSDs, and retest.

I'm testing this out

Out of curiosity, is there any rhyme/reason as to why the SSDs usually don't work as well through the HBAs? Is this some make/model particularity.

I ask because for as long as I've been running unraid I've had my SSD drives going through an HBA. Though, on all my previous configurations I believe were utilizing port multipliers. Could there be something in that that may have prevented things from working appropriately?

I've seen issues before, especially with Samsung devices.