Advice - ZFS Clear or Replace?

[TooMuchData]

I am currently running my pool (6 x 6TB WD Reds in raidz2) in an external box via an LSI 9200-8e (P20) in a Lenovo TS140 (latest 9.3.1).

Yesterday I received a Critical Alert email stating "The volume zVol (ZFS) state is ONLINE: One or more devices is currently being resilvered. The pool will continue to function, possibly in a degraded state."

The system log showed:

(Sorry, one line is missing - similar to the first da1 line)
May 26 20:18:37 freenas (da0:mps0:0:8:0): CAM status: SCSI Status Error May 26 20:18:37 freenas (da0:mps0:0:8:0): SCSI status: Check Condition
May 26 20:18:37 freenas (da0:mps0:0:8:0): SCSI sense: ILLEGAL REQUEST asc:21,0 (Logical block address out of range)
May 26 20:18:37 freenas (da0:mps0:0:8:0): Info: 0x28fbb3f58 May 26 20:18:37 freenas (da0:mps0:0:8:0): Error 22, Unretryable error
May 26 22:19:07 freenas (da1:mps0:0:9:0): WRITE(16). CDB: 8a 00 00 00 00 02 8b 6f e3 c8 00 00 00 08 00 00
May 26 22:19:07 freenas (da1:mps0:0:9:0): CAM status: SCSI Status Error May 26 22:19:07 freenas (da1:mps0:0:9:0): SCSI status: Check Condition
May 26 22:19:07 freenas (da1:mps0:0:9:0): SCSI sense: ILLEGAL REQUEST asc:21,0 (Logical block address out of range)
May 26 22:19:07 freenas (da1:mps0:0:9:0): Info: 0x28b6fe3c8 May 26 22:19:07 freenas (da1:mps0:0:9:0): Error 22, Unretryable error​

a zPool report stated:

pool: zVol
state: ONLINE
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://illumos.org/msg/ZFS-8000-9P
scan: resilvered 8.11M in 0h0m with 0 errors on Thu May 26 22:19:15 2016
config:

NAME STATE READ WRITE CKSUM
zVol ONLINE 0 0 0
raidz2-0 ONLINE 0 0 0
gptid/831ef1ef-a343-11e4-adaf-d050995044d1 ONLINE 0 0 0
gptid/83ec38c1-a343-11e4-adaf-d050995044d1 ONLINE 0 0 0
gptid/84b80ccd-a343-11e4-adaf-d050995044d1 ONLINE 0 0 0
gptid/85856a70-a343-11e4-adaf-d050995044d1 ONLINE 0 1 0
gptid/86542f6a-a343-11e4-adaf-d050995044d1 ONLINE 0 0 0
gptid/8726b5b1-a343-11e4-adaf-d050995044d1 ONLINE 0 1 0

errors: No known data errors​

The six WD Reds get weekly short SMART tests and bi-weekly long tests. None of the drives has logged an error, and all tests have completed satisfactorily. The drives are about 20 months old and have been in continuous service. I have one spare drive available.

Given that two disks experienced the same problem moments apart, I'm inclined to suspect another, common component (the 9200-8e or the cabling). So, I'm inclined to do "zpool clear", but would appreciate any comments or suggestions before I do so.

Thanks for your help.

 

[Sakuru]

Please post the output of "smartctl -a /dev/XXX" for all of your drives in code tags.

 

[TooMuchData]

How about as text files?

 

[Sakuru]

Eh, that works, but I'd prefer code tags (hit the button to the left of the save button in the toolbar) or something like pastebin. That way I don't have to download all those files :)

 

[Sakuru]

Hmm, your SMART stats look great.
Are you using any of the Marvell ports on that board?

 

[TooMuchData]

Thank you, Sakuru.
Please see initial post. No Marvel ports are involved in this error.

 

[Sakuru]

Ah, so this isn't the one in your signature.
What is the "external box" that you refer to?

 

[Mirfster]

Check the LSI downloads, they did release another update (04/04/2016 if I recall) that is 20.00.07.00 which while not really required for all does have a fix mainly for those seeing a particular issue. I don't recall what it was off the top of my head but might be worth checking out.

 

[TooMuchData]

Thank you, Mirfster and Sakuru.
The LSI has 20.00.07.00.
The external box is another Node 304 with an SSR-450RM PSU for the disks, and an SFF-8088 external connector that fans out to eight internal SAS/SATA cables six of which are connected to the disks. A CSE-PTJBOD-CB1 ATX board does power on-off control.

Any reason why I should not do zpool clear?

My replacement C2750D4I arrived today, so I plan to put all the disks back into my other Node 304 in a few days.

 

[Mirfster]

Any reason why I should not do zpool clear?

I personally would perform a system reboot first and see if things are cleared up. I am not one to just clear messages ;)

 

[cyberjock]

A reboot will clear the errors since the numbers are reset on an unmount. ;P

 

[Mirfster]

Ahh, it is good to be reassured by a more knowledgeable member. ;)