Ive posted a couple times related to issues with my LSI card entering a low power state and not coming back (d3hot to cold or something along those lines, once i get back to my laptop ill pull the logs) But its gotten to a point where I need to spend the time to ensure it gets solved, because its plaguing my main storage server.

The issue mainly seems to pop up after updating unraid (version doesnt seem to matter), after an update it can pop up after a few minutes to a day later, its random, and after so many times in a row doing it eventually it just doesnt do it anymore and stays on. This time after updating from rc2 to release it had the same behavior but this time there were so many array errors it disabled 2 of my disks.

I have disabled all PCIe power management in the bios, disabled everything power management except for basic c states even disabling c6 due to reports of errors. Updated the bios, updated to the latest IT firmware on the LSI card they dropped in the TrueNAS forums.

I have added in comments to boot "pci=noaer pcie_aspm=off"

I dont understand why its still dropping into a low power state.

System Specs:

asrock Z690 Pro RS w/ Intel i5-12400

LSI 9300-16i

The first couple errors before the system starts spitting out disk errors:

Bottom to Top -

Blockbuster	kern	kernel	mpt3sas_cm1: basefault_reset_work: Running mpt3sas_dead_ioc thread success !!!!
Blockbuster	kern	kernel	mpt3sas_cm1: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm0: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm1: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm0: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm1: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm0: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm1: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm0: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm1: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm0: SAS host is non-operational !!!!
Blockbuster	kern	kernel	mpt3sas_cm1: SAS host is non-operational !!!!
Blockbuster	kern	kernel	pcieport 0000:02:08.0: Unable to change power state from D3hot to D0, device inaccessible

Edited October 30, 2025Oct 30 by no1warr1or
added information

Make sure the HBA is well seated and sufficiently cooled; you can also try using a different PCIe slot.

23 minutes ago, JorgeB said:

Make sure the HBA is well seated and sufficiently cooled; you can also try using a different PCIe slot.

I've reseated the card a couple times over the last year or so with this issue. It has a 90mm noctua fan on top of the heatsink on full speed. The card can get hit hard transfer speed wise when doing rebuilds without issues and stay running for months is what gets me, but as soon as I do an update/reboot its a gamble if this pops back up.. I've considered maybe the aux 6pin power might need to be hooked up but I dont have a fully populated card ( only 3 sas connectors in use ). Ive also considered manual link speed for PCIe, but I hadnt thought of changing slots.

Right now I'm worried about changing anything, With the OS deciding to disable Parity 1 and Disk 1 I need to get the array back in order. Currently running Extended SMART tests to be safe per the guide for restoring disabled drives.. But I know its just because of this issue and I honestly cant believe I have to rebuild the drives to re-enable them.

Did you also try a different PCIe slot if possible?

11 hours ago, JorgeB said:

Did you also try a different PCIe slot if possible?

Not yet, currently in the middle of parity sync to restore parity 1 and disk 1. Once that's completed im gonna dig further.

75% into the parity rebuild and it crashed with the same error. I still cant believe the OS decided to disable 2 of my disks and theres no option to enable without rebuild. So here we go again starting with ANOTHER rebuild.

I reseated the card again, disabled ALL cstates, hooked up aux power to the LSI card and now I wait to see if it can manage a rebuild before a crash or before unraid decides to disable another drive for no reason and I lose 60TB of data

Maybe you already tried this, but I have messed around with these settings, and in my bios they are for either having the OS control the power management, or the motherboard bios itself.

So if you just disable them all (like the “Native ASPM power management” or whatever it’s called), that lets bios have control of power management instead of the OS. Since you have pcie_aspm=off set on the OS, maybe you want to switch back the appropriate bios settings to enabled or auto so Unraid has control. Because based on your description it sounds like it might be the motherboard bios doing some mindless energy saving.

Edited November 2, 2025Nov 2 by jornNER

On 10/31/2025 at 5:55 AM, no1warr1or said:

its random

I don't think it relate ASPM, it is something wrong on HBA. Both controller on HBA non-operational.

It try to reset the controller ( watch dog ), but not bring up.

mpt3sas_cm1: basefault_reset_work: Running mpt3sas_dead_ioc thread success !!!!

Edited November 3, 2025Nov 3 by Vr2Io

8 hours ago, Vr2Io said:

I don't think it relate ASPM, it is something wrong on HBA. Both controller on HBA non-operational.

It try to reset the controller ( watch dog ), but not bring up.

mpt3sas_cm1: basefault_reset_work: Running mpt3sas_dead_ioc thread success !!!!

The first error to pop up is d3hot to d0 which is why I think/thought it was ASPM related. After that error pops up the card spits out the rest of those errors and more.

I have IT firmware .12 that was released in the truenas forums that was supposed to help with the card going through resets. Which it did. But I still get these weird errors shortly after restarting the system. It can go many months without an error, but the minute I restart the systems it's almost a guarantee it'll pop up.

Seems like these 3 series cards are filled with problems, I see a lot of posts related to them. Looking at system drivers there are already tweaks that unraid made to it to fix an issue with other 3 series cards.

Ultimately I moved my array back to an old case and board with a 2 something series lsi sas card and rebuilt my array. Waiting for a unifi unas to come in along with a handful of 26TB drives to copy my data. Ill decide the fate of this server later.

3 hours ago, no1warr1or said:

Seems like these 3 series cards are filled with problems,

Series 3 16i seems little bit trouble.

I use one 93004i4e with expander to connect all disks. So I need 12gb and fade out all series 2 hba.