Short and Long SMART tests clear so far, just that the ERC is set to 0.1s, which seems odd.
I can’t change it’s value however. With smartctl -l scterc,70,70 /dev/sdd it just keeps reverting to 0.1 seconds right away.
(This is happening to all 6 of my 24TB WD HC580 refurbished drives).
This is for a home-use NAS, with the 6 drives in RAIDZ2.
Will setting the kernel SCSI timeout to 180 be a good fix, for me to go ahead and run badblocks and create the pool, for my TrueNAS build?
----
System specs:
6 x WD Ultrastar DC HC580 WUH722424ALE604 0F62798 24TB 7.2K RPM SATA 6Gb/s 512e 3.5in Recertified Hard Drives
Running TrueNAS 25.04.2.6
Case: Cooler Master HAF 922 (old 2011 Case, repurposed)
CPU: AMD Ryzen PRO 4750G
CPU Cooler: Noctua NH-U14S
Motherboard: ASRock B550 Pro4
RAM: 64GB UDIMM ECC (2 x 32GB Kingston KSM26ED8/32HC 2666 CL19 ECC 288 PC4)
Mirrored Boot OS SATA SSDs: 2 x Intel SSD DC S3700 200GB (used enterprise gear)
HBA Card: LSI 9305-16i
PSU: Corsair RM850x
---
Update:
Message I got from ServerPartDeals regarding the issue. Is this sound advice?:
The 0.1s ERC value is not the normal factory default — it should be 7 seconds on enterprise drives like these, and the refurb/recertification process likely reset it. Totally understandable to notice and question it.
The reason it keeps reverting is most likely your HBA silently dropping the SCT command rather than passing it through to the drive. This is a known quirk with SAS HBAs and SATA drives. You can try:
smartctl -d sat,16 -l scterc,70,70 /dev/sdX
If it still won’t stick, setting the kernel SCSI timeout (echo 180 > /sys/block/sdX/device/timeout) is a good fallback that gives you similar protection for ZFS.
You’re safe to go ahead with badblocks either way.