ZFS-8000-8A and then...

[DrBobba]

First a little background info.

I very much like the idea of ZFS and how it handles. Got a HP microserver with 8GB of ECC ram. With 3 disks of 1.5 TB and one of 2 TB making one Z1 pool of 4.5 TB.

After my most recent scrub I got a ZFS-8000-8A message. Saying it has 10 write errors in my first and largest dataset.

Found out that 5 files are affected. The files that are affected are not a major issue, but I want to known what made this error happen. Cause maybe next time other files are effected that I can't have being messed up.

I checked all disks for the smart statuses, everything passed. I got original HP ECC ram for the microserver which was ok when I installed the system. I'm running FreeNAS-9.3-STABLE-201511040813 .

I want to find the cause of the problem and take measures to prevent it in the future. Cause as probably many I use it as a backup system. Which I want to be able to count on.

Any help or thoughts are most appreciated.

 

[zambanini]

use sas drives, if your data is important.

 

[DrBobba]

I would first like to know how this problem actually happened.

 

[BigDave]

I checked all disks for the smart statuses

Meaning they were all listed as:

Code:

=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED

Or are you saying there was a long history of short and long tests listed as:

Code:

SMART Self-test log structure revision number 1
Num  Test_Description  Status  Remaining  LifeTime(hours)  LBA_of_first_error
# 1  Short offline  Completed without error  00%  9781  -
# 2  Extended offline  Completed without error  00%  9644  -
# 3  Short offline  Completed without error  00%  9542  -
# 4  Short offline  Completed without error  00%  9445  -
# 5  Short offline  Completed without error  00%  9390  -
# 6  Short offline  Completed without error  00%  8139  -
# 7  Extended offline  Completed without error  00%  8048  -
# 8  Short offline  Completed without error  00%  7982  -
# 9  Short offline  Completed without error  00%  7796  -
#10  Extended offline  Completed without error  00%  7656  -
#11  Short offline  Completed without error  00%  7553  -
#12  Short offline  Completed without error  00%  7457  -
#13  Extended offline  Completed without error  00%  7368  -
#14  Short offline  Completed without error  00%  7241  -
#15  Short offline  Completed without error  00%  7050  -
#16  Extended offline  Interrupted (host reset)  60%  6910  -
#17  Short offline  Completed without error  00%  6812  -
#18  Short offline  Completed without error  00%  6740  -
#19  Extended offline  Completed without error  00%  6655  -
#20  Short offline  Completed without error  00%  6334  -
#21  Extended offline  Completed without error  00%  6192  -

 

[DrBobba]

Meaning a list of 20 completed checks without errors.

 

[BigDave]

I got original HP ECC ram for the microserver which was ok when I installed the system.

If you are convinced it's not a drive issue, test the RAM.

 

[DrBobba]

I'll do a full ram test, but first I'm copying the data to an other system. Which goes fine actually without any more errors. And without touching the effected files of course.

 

[BigDave]

Good luck and let us know how you're getting along :)

 

[DrBobba]

Thanks! The weird thing is there are only 9 errors on 9 small files.

 

[jgreco]

use sas drives, if your data is important.

This seems like a nonsensical suggestion. SAS and SATA drives differ primarily in the interface technology.

The real question here is why didn't redundancy kick in to allow the problem to be corrected. Are you not using RAIDZ2?

 

[Ericloewe]

This seems like a nonsensical suggestion. SAS and SATA drives differ primarily in the interface technology.

The real question here is why didn't redundancy kick in to allow the problem to be corrected. Are you not using RAIDZ2?

A RAIDZ2 setup with "unreliable" drives is a lot more reliable than a single half as "unreliable" drive.

 

[Robert Trevellyan]

The real question here is why didn't redundancy kick in

Good question.

OP, please post output from zpool status between CODE tags.