Samsung P043S7T6 EMC7680 SSD

[The Gecko]

I bought two similar drives (A & B) from a vendor. Drive A got assimilated by the system quickly and is happily being used by ZFS. Drive B showed "issues". I investigated as best I could, was convinced there was something physically wrong with the drive, contacted the vendor, and received a replacement today (Drive C). Drive C is now showing the same behavior as the first, always centered around the failure to create backup GPT. This leads me to believe it is not a physical failure. I need help. Any ideas?

Background:

• The drive is connected to a Supermicro system, using an LSI SAS-2 controller in IT mode.
• Proxmox 6.0-4 (Debian)
• SmartMonTools has been upgraded to v7.2

Preparing the disk with GPT:

root@proxmox03:~/smartmontools# sgdisk /dev/sde -U R
Warning! Error 5 reading partition table for CRC check!
Warning! One or more CRCs don't match. You should repair the disk!

****************************************************************************
Caution: Found protective or hybrid MBR and corrupt GPT. Using GPT, but disk
verification and recovery are STRONGLY recommended.
****************************************************************************
Unable to save backup partition table! Perhaps the 'e' option on the experts'
menu will resolve this problem.
Warning! An error was reported when writing the partition table! This error
MIGHT be harmless, or the disk might be damaged! Checking it is advisable.

Smartmon report:

root@proxmox03:~/smartmontools# ./smartctl -a /dev/sde
smartctl 7.2 2020-02-25 r5033 [x86_64-linux-5.0.15-1-pve] (local build)
Copyright (C) 2002-20, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Vendor:               SAMSUNG
Product:              P043S7T6 EMC7680
Revision:             ESV4
Compliance:           SPC-5
User Capacity:        7,680,475,267,072 bytes [7.68 TB]
Logical block size:   512 bytes
Physical block size:  4096 bytes
LU is resource provisioned, LBPRZ=1
Rotation Rate:        Solid State Device
Form Factor:          2.5 inches
Logical Unit id:      0x5002538b0940f110
Device type:          disk
Transport protocol:   SAS (SPL-3)
Local Time is:        Wed Feb 26 14:18:13 2020 EST
SMART support is:     Available - device has SMART capability.
SMART support is:     Enabled
Temperature Warning:  Enabled

=== START OF READ SMART DATA SECTION ===
SMART Health Status: OK

Percentage used endurance indicator: 0%
Current Drive Temperature:     36 C
Drive Trip Temperature:        73 C

Manufactured in week 14 of year 2019
Accumulated start-stop cycles:  44
Specified load-unload count over device lifetime:  0
Accumulated load-unload cycles:  0
Elements in grown defect list: 0

Error counter log:
           Errors Corrected by           Total   Correction     Gigabytes    Total
               ECC          rereads/    errors   algorithm      processed    uncorrected
           fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
read:          0        0         0         0          0        182.468           0
write:         0        0         0         0          0        525.059           0
verify:        0        0         0         0          0          0.009           0

Non-medium error count:     5696

SMART Self-test log
Num  Test              Status                 segment  LifeTime  LBA_first_err [SK ASC ASQ]
     Description                              number   (hours)
# 1  Background short  Completed                   -      10                 - [-   -    -]
# 2  Reserved(7)       Completed                   -       1                 - [-   -    -]
# 3  Reserved(3)       Completed                   -       1                 - [-   -    -]
# 4  Reserved(7)       Completed                   -       1                 - [-   -    -]
# 5  Reserved(3)       Completed                   -       1                 - [-   -    -]

Long (extended) Self-test duration: 90 seconds [1.5 minutes]

 

[vanfawx]

What happens if you "sgdisk --zap-all /dev/sde" first, before you write a new GPT? You shouldn't need anything other than the --zap-all, as ZFS will write out its own GPT table when you add it into a vdev.

Most likely this happened because the drive had a pre-existing partition table, or the remnants of one.

 

[The Gecko]

Results:

root@proxmox04:~# sgdisk --zap-all /dev/sde
Warning! Error 5 reading partition table for CRC check!
Warning! One or more CRCs don't match. You should repair the disk!

****************************************************************************
Caution: Found protective or hybrid MBR and corrupt GPT. Using GPT, but disk
verification and recovery are STRONGLY recommended.
****************************************************************************
Warning! GPT backup partition table not overwritten! Error is 5
Warning: The kernel is still using the old partition table.
The new table will be used at the next reboot or after you
run partprobe(8) or kpartx(8)
GPT data structures destroyed! You may now partition the disk using fdisk or
other utilities.

root@proxmox04:~# apt install parted
root@proxmox04:~# which partprobe
/usr/sbin/partprobe
root@proxmox04:~# partprobe /dev/sde
<no_output>

root@proxmox04:~# zpool create singledisk /dev/sde
invalid vdev specification
use '-f' to override the following errors:
/dev/sde contains a corrupt primary EFI label.

root@proxmox04:~# zpool create singledisk /dev/sde -f
<no_output>

root@proxmox04:~# zpool status singledisk
  pool: singledisk
 state: DEGRADED
status: One or more devices has experienced an unrecoverable error.  An
        attempt was made to correct the error.  Applications are unaffected.
action: Determine if the device needs to be replaced, and clear the errors
        using 'zpool clear' or replace the device with 'zpool replace'.
   see: http://zfsonlinux.org/msg/ZFS-8000-9P
  scan: none requested
config:

        NAME        STATE     READ WRITE CKSUM
        singledisk  DEGRADED     0     0     0
          sdaa      DEGRADED    14     0     0  too many errors

errors: No known data errors

 

[vanfawx]

That last error seems to be the most helpful. Here's the options you can check on your system that might have adopted the drive:

1) mdadm (check in /proc/mdstat if sde is listed)
2) lvm (check with the output of pvs if sde is listed as a physical volume)
3) might be mounted (grep sde /proc/mounts)