SYNCHRONIZE CACHE command timeout error

[Tangeek]

Hi,

Bumping this thread as well becuse I have the exact same issue as you all (Ironwolf 8TB in my case). While it gives me hope to know there's a "hidden updated firmware", I couldn't find it, either here or on the Seagate website. Did someone share it somewhere and I missed it ? Or is there something preventing to do it publicly ?

I would be really grateful if someone did. Right now I have to reboot my NAS every few days just to get my disks back up again (I don't use FreeBSD so I don't have the camcontrol command, and I have no idea how to do the queue thing on Linux)

Still, the research you did, Gcon, is amazing. I was going nuts moving the disks around in the backplane, cables, and SAS cards, and always having the same disks going down. Thanks a lot ! At the very least knowing where it's going wrong makes me sleep a bit better.

 

[Tangeek]

I know it's outside the scope of this forum called truenas, but just in case a Linux user is googling down here by accident like me, add libata.force=noncq to the kernel options at boot time. Seems to have done the trick for me, I have been stressing the s*it of my disks for a few hours now without an incident (two dd's writing from urandom, three copies through SMB, multiple useless backups running at once, ... Usually so much random I/O gets my disks panicking).

 

[Bearded]

Hi,
Despite having one version of "secret" firmware for the disks, it unfortunately does not solve the issue for me. After a period of heavy use, a disk with a serial # I definitely double-positively have flashed, did the same error again. I'm so sad to have 14 of these absolutely useless drives. Seagate denies to know of any firmware fixes, and (initially) they offered to just replace the drives. If anyone here reads this and still has their drives under warranty, i cannot stress this enough: Get Seagate to replace them for you.

Goint to test the disabling of NCQ as @Tangeek suggested.

 

[Alex_K]

Hey guys

It seems there is another flavor of 16.00.12.00 fw abailable on Broadcom website for 9305 only as of now
As of its release notes it has a more sofiscicated NCQ fix then outright disabling it:

Code:

ID: DCSG00398894
Headline: SATA only: WRITE SAME NCQ encapsulation assumes NonData NCQ is supported if Zero EXT is Supported
Description Of Change: Disable NCQ encapsulation if Zero EXT is supported but Non Data NCQ is not supported
Issue Description: In case if Zero EXT is supported but Non Data NCQ is not supported by drive, WRITE SAME
NCQ encapsulation would send Non Data NCQ command to the drive.
Drive would fail the command as Non Data NCQ is not supported by drive.
This will cause command failure to host.
Steps To Reproduce: IO errors are observed when mkfs.ext4 operation is done on drives that support Zero EXT but
do not support Non Data NCQ.

It is Different from 16.00.12.00 we have for 9300 here. Could someone who has 9305 or compatible OEM and affected drives check if that bugfix works?