My server has always been on a thinly balanced edge with unRAID 6 and my AOC-SAS2LP card - long history of issues during the beta.  I got it to settle down for quite a while with a bunch of workarounds, though.  I'm seeing a recurance of problems lately - errors on the marvel driver, stack traces, and a hard reset of the link.  It feels like an uptick since 6.2.4 but that might just be a coincidence.  I think I have VT-x enabled in the BIOS but I'm not running an VMs - could turn that off if it would help.

 

Any ideas?  I'm starting to get uncomfortable with this again, even though it's generally recovering.  I did have a dropped disk during my first 6.2.4 parity sync which I had to rebuild onto itself (all the SMARTs are fine).

 

Diagnostics attached, thanks.

tower-diagnostics-20170115-1018.zip

Your motherboard BIOS is dated 2009. It might be worth looking for an update.

 

Since you're not using it disable VT-D to see if it helps.

Your motherboard BIOS is dated 2009. It might be worth looking for an update.

I'm pretty sure that's the most current but I'll give it another look.

 

 

Since you're not using it disable VT-D to see if it helps.

Thanks, the motherboard only suppports VT-x and most of the problems I've seen regarding the marvell drivers seem to refer to VT-d, but I suppose it can't hurt.  I turned VT-x on early in the unRAID 6 days thinking that Docker needed it but I've only played with a couple of VMs since then.

Misread, vt-x shouldn't be a problem. 

The card seems to be recovering after hard resetting the link - how worried should I be?

I wouldn't be confortable with so many errors, sooner or later you'll likely get a disk read or write error, but it may not even be a controller issue, e.g., a bad cable is always a possibility.