SOLVED Read SMART Error Log Failed after every test

[Murac]

So this is my first post to this fine forum. I've generally always been able to find answers in existing threads but this one is curious. I get errors such as these:

Code:

 CRITICAL: Device: /dev/ada5, Read SMART Error Log Failed
CRITICAL: Device: /dev/ada5, failed to read SMART Attribute Data
CRITICAL: Device: /dev/ada3, Read SMART Error Log Failed
CRITICAL: Device: /dev/ada3, Read SMART Self-Test Log Failed

After every single smart test, short and long. It cycles through all the different drives over time, without fail. My first question, is where is this self-test log located in the filesystem?
A few months back when smartd was having issues during certain updates (when the reporting was being debuted?) I believe I may have caused a permission issue with the log file. At that time, i was getting some sort of error that showed a path to the log file and something else along the lines of permission denied or does not exist. So, I thought I'd try and create the file myself, however, I most likely did not set the permissions properly, and once the bugs were fixed and rolled out in an update, the faulty file i created remained and created an obstacle for the reporting. Does this make any sense?
I know the drives are all testing without errors, it's just the reading of the test logs.

This has been bugging me for months and I've finally gotten the wherewithall to ask for help. I do appreciate you taking the time to tackle this problem! Thank you!

I haven't included my system specs because I don't see how it would be relevant, but I did build according to one of those guides using the intel c2750 quad-core boards so i'm pretty certain all components are kosher. also using 6 4-tb red drives in z2. I will quickly post any new information requested, if necessary.

final note is that I see my drives get up to 53C every so often during tests, should I be incredibly concerned with these numbers?

THANK YOU!!!!

-Murac

 

[DrKK]

The "failed logs" and "unable to read attributes" errors you are receiving has nothing to do with any files on the filesystem. These are errors you are receiving DIRECTLY FROM THE DRIVE FIRMWARE, which is saying that it cannot read SMART data of various types.

Could you please tell us precisely what your hardware is, with particular attention to any HBA/disk controllers that you may be using?

 

[Murac]

Ps: Seasonic SS-350SFE 350W Active PFC Power Supply
Memory: Crucial 16GB Kit (8GBx2) DDR3L 1600MT/s (PC3-12800) DR x8 ECC UDIMM
Cpu/mobo: ASRock C2550D4I Intel Avoton C2550
Hdds: 6 x WD Red 4TB NAS Hard Drive: 1 to 8-bay RAID Hard Drive: 3.5-inch SATA in RaidZ2
Case: Silverstone Tek Premium Mini-ITX DTX Small Form Factor NAS Computer

2 SATA3 6.0Gbps, 4 SATA2 3.0Gbps by C25504 x SATA3 6.0 Gb/s by Marvell SE9230, 2 x SATA3 6.0 Gb/s by Marvell SE9172


Have I provided sufficient information?

 

[DrKK]

Yes, you certainly have.

Something is definitely wrong. Can you tell us if you have the hard drives plugged into the Marvell, or the Intel ports? A lot of people plug them into the Marvell ports because they think it's "faster". That would be the wrong decision, if that's what you did.

 

[Murac]

I can't be entirely sure without opening the box but upon looking at the diagram I see how I may have mixed and matched. I've also noticed ahcich timeout errors from time to time..would switch my drives to strictly intel ports be your first suggestion?

 

[DrKK]

Well, the problem you're experiencing seems unusual. I doubt that will fix it. but:

There is no doubt whatsoever that the Intel ports are more reliable, and that if you have six or fewer drives, that 100% of the FreeNAS people would counsel you to use strictly the Intel ports.

 

[russnas]

Yes use native Intel ports over 3rd party, I recalled my asrock motherboard used intel and had asmedia, full bandwidth vs shared,

Your raid should use Intel ports ,then another raid for the Marvell ports
That mb has 2 marvell controllers I believe.

See the avaton family, supports a total of 6 sata ports
ark.intel.com/products/77982/Intel-Atom-Processor-C2550-2M-Cache-2_40-GHz

Another thread about the issue
https://forums.freenas.org/index.php?threads/asrock-c2750d4i-vs-x10sl7-f-sata-ports.30847/

See info below
www.tomshardware.com/forum/308172-30-sata-controller-question#4425211

 

[Ericloewe]

final note is that I see my drives get up to 53C every so often during tests, should I be incredibly concerned with these numbers?

Yes. 40 degrees Celsius is the ceiling you should aim for.

 

[fta]

There is no doubt whatsoever that the Intel ports are more reliable, and that if you have six or fewer drives, that 100% of the FreeNAS people would counsel you to use strictly the Intel ports.

I have doubts, and apparently I'm among the 0% who say he's fine on the Marvell controller. He has the same Marvell controller as the FreeNAS mini. My mini came with everything plugged into the Marvell controller. I've had zero problems.

 

[Ericloewe]

I have doubts, and apparently I'm among the 0% who say he's fine on the Marvell controller. He has the same Marvell controller as the FreeNAS mini. My mini came with everything plugged into the Marvell controller. I've had zero problems.

Your experience is the best case. Many people have had problems caused by the Marvell SATA controllers.
In fact, newer FreeNAS minis use the Intel ports, from what I've heard.

 

[cyberjock]

I have doubts, and apparently I'm among the 0% who say he's fine on the Marvell controller. He has the same Marvell controller as the FreeNAS mini. My mini came with everything plugged into the Marvell controller. I've had zero problems.

I can tell you that the original design for the Mini called for 2 ports on the Intel and 2 ports on the Marvell. Currently all ports go to Intel because of problems with the Marvell.

So if you are going to say that everything is plugged into the Marvell then it was a mistake during manufacturing. Plugging everything into the Marvell controller has never been a proper configuration per the design the Mini was tested to.

 

[fta]

I can tell you that the original design for the Mini called for 2 ports on the Intel and 2 ports on the Marvell. Currently all ports go to Intel because of problems with the Marvell.

I looked again and you are mostly correct, 3 ports of the Marvell are used, 2 of the intel are used.

per the design the Mini was tested to.

So clearly the mini was tested using the Marvell controller, and it passed. Or are you saying they didn't actually test until problems were reported?

 

[DrKK]

I looked again and you are mostly correct, 3 ports of the Marvell are used, 2 of the intel are used.



So clearly the mini was tested using the Marvell controller, and it passed. Or are you saying they didn't actually test until problems were reported?

I'll leave this to @cyberjock to answer, or someone official. It's complicated.

 

[Murac]

Hey everyone, so it's only been one whole year, but I finally garnered the courage and motivation to wrangle this little issue. I imagine more than a few use the ASRock C2550 board so I thought it may benefit others to mention my solution. It was quite simple really.

UPDATE THE MARVELL FIRMWARE. lol.

Since I first posted this issue I've repeatedly gotten READ SMART ERRORS every time a scan was performed. I probably should have just turned them off all together until I figured out the problem, but nonetheless, instead of switching out the wires, or changing the ports, I went to the asrock website and found fairly recent BIOS and Firmware updates. First I updated the IPMI, just because. Then I updated the board BIOS (recommended according to instructions). And finally went for the Marvell controllers after that. I used bootable USB keys for the board and controllers. It's been a week now, no errors, and SMART scans have come and gone without any issues. Hurrah!

Here are the links to ASRock if anyone finds themselves on this thread seeking answers:

BIOS/BMC: http://www.asrockrack.com/general/productdetail.asp?Model=C2550D4I#Download
Marvell 9230 FW Update Procedure: http://www.asrockrack.com/support/faq.asp

Edit: looks like my temps have gone down by 10C as well...unless it's because I moved the box. Either way, more good news

 

[Ericloewe]

Is there a somewhat recent update? When the board became popular, they released an update that did help a lot. However, some people still had issues with that controller...

 

[Murac]

03/2016 feels pretty recent to me....that's the BIOS. not sure when the Marvell update was released but my problems seem to have gone away.

 

[DrKK]

This is very good information. I am sure this thread will be helpful to several people.

 

[DaveY]

This is very good information. I am sure this thread will be helpful to several people.

Indeed it was! Thanks for posting this. I had the exact same problem. The SMART errors went away as soon as I updated the Marvell driver. Just FYI, I have a C2750D4i and not C2550D4i so looks like the fix works on both.

 

[Murac]

Indeed it was! Thanks for posting this. I had the exact same problem. The SMART errors went away as soon as I updated the Marvell driver. Just FYI, I have a C2750D4i and not C2550D4i so looks like the fix works on both.

Gosh, I've been kicking myself for not spending the extra hundred for that board.....

 

[DaveY]

I think you're the smart one for not spending the extra $$ considering my board died couple of days after this post. I've had the board for just a little over a year. I think reliability of these boards are still questionable.:(