I posted this problem to the fedora users list first:
=============== original post
fedora 28 up-to-date.
I have a new array, not yet commissioned, and one disk now complains
Device: /dev/sde [SAT], FAILED SMART self-check. BACK UP DATA NOW!
and the reason, I understand, is this
ID# ATTRIBUTE_NAME FLAGS VALUE WORST THRESH FAIL RAW_VALUE
200 Multi_Zone_Error_Rate PO---K 001 001 001 NOW 0
My issue is that the disk works fine and never had any errors reported.
Also, smartmontools db does not include this disk
ST12000NM0007-2A1101
and
update-smart-drivedb
is failing with
/usr/share/smartmontools/drivedb.h.error.raw: *** BAD signature ***
I know that some disks report funny numbers and smartmontools needs to know which
one to trust. I see a relevant ticket from 2018-06-15....
https://www.smartmontools.org/attachment/ticket/1042/smartctl-SEAGATE-ST12000NM0007.txt
And I see that smartmontools has other issues with these disks:
Write SCT (Get) Feature Control Command failed: scsi error badly formed scsi parameters
Wt Cache Reorder: Unknown (SCT Feature Control command failed)
I ran some short and long smart tests but all failed immediately.
SMART Extended Self-test Log Version: 1 (1 sectors)
Num Test_Description Status Remaining LifeTime(hours) LBA_of_first_error
# 1 Extended offline Completed: unknown failure 90% 250 0
# 2 Short offline Completed: unknown failure 90% 247 0
# 3 Short offline Completed: unknown failure 90% 247 0
# 4 Extended offline Completed without error 00% 17 -
The other (identical) disks in the array report
ID# ATTRIBUTE_NAME FLAGS VALUE WORST THRESH FAIL RAW_VALUE
200 Multi_Zone_Error_Rate PO---K 100 100 001 - 0
I can request a replacement, but want to ensure that this is not just a smart software issue which
may pop up again on another disk.
Should we have smartmontools v7.0 (released end of last year) by now? It includes this disk.
As I mentioned, the array is not yet commissioned and I can afford to fiddle with it for a while.
TIA
=============== end original post
I tested with a current smartmontools-7.0.5 with same results.
I expect people on this list are now using these disks, and I wonder if anyone saw this kind of failure,
where the Multi_Zone_Error_Rate drops from 100 to 001 without any I/O error recorded and the disk still
apparently in working order.
I just ran fsck on the array (7 disk RAID6 which is 30% full - 18TB out of 60TB).
Completed without a problem. iostat shows all members with same number of transactions.
--
Eyal at Home (eyal@xxxxxxxxxxxxxx)
