Drive timout during spin up

[Ariyala]

Hey,

 

I'm usually getting these log entries during disk spin up:

 

kernel: sd 14:0:3:0: attempting task abort!scmd(0x00000000155105fb), outstanding for 15511 ms & timeout 15000 ms
kernel: sd 14:0:3:0: [sdg] tag#4740 CDB: opcode=0x85 85 06 20 00 00 00 00 00 00 00 00 00 00 40 e5 00
kernel: scsi target14:0:3: handle(0x000b), sas_address(0x4433221104000000), phy(4)
kernel: scsi target14:0:3: enclosure logical id(0x500605b00a071021), slot(7) 
kernel: scsi target14:0:3: enclosure level(0x0000), connector name(     )
kernel: sd 14:0:3:0: task abort: SUCCESS scmd(0x00000000155105fb)

 

If there is an immediate read or write attempt I also get a disk read or write error.

I'm using seagate iron wolf drives. These drives apparently have an average spin up time of 20+ seconds. So I'm assuming the 15 second timeout is the problem here not giving the disks enough time to spin up. Is that timeout coming from the controller or unraid? If it is unraid where can this value be set?

 

So far I have replaced all SAS cables, and the controller card. I also have moved the disks around in the drive cages. Behavior is exactly the same, nothing changed. So increasing the timeout is probably the next logical step.

 

Thanks in advance.

 

 

Edited January 24, 2022 by Ariyala

[JorgeB]

Check this:

https://forums.unraid.net/topic/103938-69x-lsi-controllers-ironwolf-disks-disabling-summary-fix/?do=getNewComment

 

[Ariyala]

Thank you for linking this thread, I will look into the firmware.

However is anyone working on UNRAID looking into this? The features suggested to disable seem like something a NAS would benefit from. Especially low current spin up when you have many drives.

I understand the thread suggests this is a Seagate / LSI problem, but if UNRAID is the only system running into issues with that hardware configuration it seems unlikely.

Is there any way to increase the 15second timeout for spin up? Low current spin up takes longer, so just giving the drives more time might solve it without having to disable beneficial drive features.

Edited January 27, 2022 by Ariyala

[JorgeB]

4 hours ago, Ariyala said:

but if UNRAID is the only system running into issues with that hardware configuration it seems unlikely.

It will affect any Linux based system, not just Unraid.

[Ariyala]

I have disabled EPC and low current spin up in the drive firmware and so far the timeout has not appeared in the log again.

Unfortunate that these two drive features are not usable, but better than rebuilding the array every few days.

 

If anyone has any idea where to increase the drive timeout I would still be happy to get that information and try out if at least low current spin up is usable.

 

Thank you @JorgeB for linking the other thread, as far as I can tell it is a good workaround for my issue.

Edited February 2, 2022 by Ariyala