SMART Service Fails after 11.3 Upgrade

jlw52761

Explorer
Joined
Jan 6, 2020
Messages
86
So far the only issue I'm running into is the SMART service is not starting. I am still a little new to FreeNAS and FreeBSD in general (Linux and Windows guy here) so not sure where all the different logs would be at to see why the service is failing. Otherwise, my jails and VMs are working as expected and the VMware iSCSI integration is all good to go.
 
Joined
Jan 4, 2014
Messages
1,644
Are you using USB sticks for boot devices? If you are, I seem to recall reading somewhere that the SMART service will not start if SMART is enabled for these devices. To check what SMART is enabled on, under Storage > Disks, expose the Enabled SMART column. You can disable SMART for specific devices.
 

jlw52761

Explorer
Joined
Jan 6, 2020
Messages
86
So I do have a USB thumbdrive as the boot device, and it did have SMART enabled. I disabled it, but still couldn't start the service.
 
Joined
Jan 4, 2014
Messages
1,644
As a first step to troubleshooting this, try disabling SMART for all devices under Storage > Disks and then see if you can start the service. If it does, progressively enable SMART for each device, restarting the service in between, until you find the errant device.
 

jlw52761

Explorer
Joined
Jan 6, 2020
Messages
86
So just an update, it seems that my drives no longer report any SMART information, there's some error in some table that is being reported. So, possibly it's going to be a bust for me on this.
 

jlw52761

Explorer
Joined
Jan 6, 2020
Messages
86
/dev/da3 is one of my 12 data drives.

Code:
smartctl -a /dev/da3
smartctl 7.0 2018-12-30 r4883 [FreeBSD 11.3-RELEASE-p6 amd64] (local build)
Copyright (C) 2002-18, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Vendor:               Intel
Product:              RMS25CB080
Revision:             3.46
Compliance:           SPC-3
User Capacity:        1,998,998,994,944 bytes [1.99 TB]
Logical block size:   512 bytes
>> Terminate command early due to bad response to IEC mode page
A mandatory SMART command failed: exiting. To continue, add one or more '-T permissive' options.


/dev/da0 and /dev/da1 are my two SSDs being used for ZIL and L2ARC
Code:
smartctl -a /dev/da0
smartctl 7.0 2018-12-30 r4883 [FreeBSD 11.3-RELEASE-p6 amd64] (local build)
Copyright (C) 2002-18, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Vendor:               Intel
Product:              RMS25CB080
Revision:             3.46
Compliance:           SPC-3
User Capacity:        98,999,205,888 bytes [98.9 GB]
Logical block size:   512 bytes
>> Terminate command early due to bad response to IEC mode page
A mandatory SMART command failed: exiting. To continue, add one or more '-T permissive' options.


Now the curiosity I have is that the Vendor and Product are the SAS controller, not the actual drives. In this server, the controller is a RAID controller without JBOD, so I have each drive set as it's own logical drive. In earlier versions of FreeNAS I'm sure this worked just fine, just 11.3 seems to have started having problems. Maybe I goofed something in the BIOS, it's an older device so anythings possible I guess.
 

seanm

Guru
Joined
Jun 11, 2018
Messages
570
So smartctl was updated in FreeNAS 11.3 to version 7, I suppose there could be a regression in smartctl, you might poke around their bug database and/or release notes. I notice there's a 7.1 now too: https://www.smartmontools.org/browser/tags/RELEASE_7_1/smartmontools/NEWS Looks like it's been in FreeBSD for just a couple of days now https://svnweb.freebsd.org/ports/head/sysutils/smartmontools/ hopefully the next FreeNAS will get it. You could build a bootable FreeBSD USB stick and test with 7.1 to see if that works...

In this server, the controller is a RAID controller without JBOD

Are you aware that that's strongly discouraged? ZFS needs to see raw disks to work best/properly.
 

tfran1990

Patron
Joined
Oct 18, 2017
Messages
293
As seanm pointed out, it could be your problem.
hit up ebay and get a recommended HBA, they come flashed for 60-75$ or get a HP H220 and flash it yourself for 30$.( theres a very detailed guide on this forum)
the controller is a RAID controller without JBOD,
even if the raid card is not the problem, it would be a good idea to get one.
 

jlw52761

Explorer
Joined
Jan 6, 2020
Messages
86
So while I understand that having ZFS on anything but raw disks isn't the best, in this particular configuration it's all I have. The drive backplane and the card are a matched pair, I would pretty much have to scrap the entire thing, of which I paid exactly $0 for, so not really wanting to do that at this time.

Now I can understand the SMART having issues, which it worked before so could jsut be some goofy thing,, but why would ZFS care since I am presenting single disks to it, not any RAID disks? That is really what I am most curious about because from a technical standpoint, ZFS needs to utilize each disk individually, and if I'm presenting each disk individually, then how does that differ from a pure JBOD configuration?

I have disabled all caching on the controller, and each of the 12 disks are set as individual logical disks, so really, from the standpoint of ZFS it's 12 disks near as I can tell. Here's the specs of what I have hardware wise:

Dual Xeon(R) CPU E5-2603 0 @ 1.80GHz 4-core (No Hyperthreading)
128GB DDR-3 RAM
Dual Intel I350 Gigabit Network Connection Quad Port
Intel RMS25CB080 RAID Module
(x2) 100GB Micron P400m100-MTFDDAK SATA 6Gbps SSD [d0,d1]
(x12) 2TB Seagate Constellation ES ST2000NM0011 7.2k SATA 6Gbps HDD [d2 ~ d13]

This is a re-purposed server that was "thrown out" and ended up on my shelf, and with the horsepower I figured would be awesome for FreeNAS. Now if that really is such a show stopper, having 12 RAID 0 drives instead of 12 JBOD drives, then that's interesting and deserves a much deeper dive and discussion, but I'm also going to have to figure out what HBA I can use with the backplane I have to power the 12 drives, plus the two SSD's that I'm using for ZIL and L2ARC.

Otherwise, I will have an ESXi box with an absolute crap ton of local storage to use, which would only be great if I had a couple of these things, 10Gbe networking, and a full vSphere license with vSAN, which none of those exist.
 

seanm

Guru
Joined
Jun 11, 2018
Messages
570
I personally don't know much about hardware, as my usage is for business, so I just bought from iX. Hopefully someone else will follow up.

But even if inadvisable, it used to work, as you say, so you've probably found a smartctl bug.
 

jlw52761

Explorer
Joined
Jan 6, 2020
Messages
86
The more I'm thinking about it, I'm wondering if it didn't work and what I was seeing and thought was working was some weird artifact....
 
Top