Hello Mikko, thanks for sharing all those details, we have some additional information on this. I am suspicious with hw and not with you. A quirk will affect lots of devices and we need to be sure that we have examined the error you get so that there is no need to revert things. FYI, I have a disk with a quirk for no obvious reason on all hosts because nForce had NCQ trouble with it. If a quirk gets applied usually it stays for a long time if not forever. Back to your case, on first look I can see that disks are in mint condition and are running the default fw, 82.00A82. Damien is there a newer version available? Please run a short test to each one to ensure that everything is OK after working for 10K+ hours and share their health log again. It is clear that both go south at the same time so RAID array dies as expected following that initial event. You just leave the PC on and then you get that event on both? Supposing you get no error at disks after all that but still disks go offline, I would suggest to update your board fw to version 3621 and retest. The controller-disk combo should work fine. Just to clarify things, upon turning off LPM, you get no error at all? Except the short form of lspci, let's get the fulll dump for SATA controller to see its active PCI settings with: sudo lspci -nnnnvvvvxxxxks 01:00.1 You can get it before touching anything and then with 3621 fw if you update your board. This will pinpoint if anything has been modified between different fw releases. If the problem remains no matter what, a quirk is inevitable. Do we wish to block LPM on Intel hosts while no one has reported errors? How about a new quirk called ATA_QUIRK_NO_LPM_ON_AMD500 for 500 series ICs or ATA_QUIRK_NO_LPM_ON_AMD similar to ATA_QUIRK_NO_LPM_ON_ATI blocking LPM for ATI chips? Best regards, Ioannis
