Questionable SMART behaviour

[leenux_tux]

Got an odd issue with a mirrored pair of SSD's in my FreeNAS box.

Both the SSD's are used as a LOG and a CACHE for a RAID10 POOL used as an NFS share for PROXMOX VM's. I have had three occasions where FreeNAS is reporting a READ error on one of the SSD's. I have issued SMART tests (short and long) against the drive each time the issue is reported and every time there are no errors coming back ? Please see below.

Has anyone else seen issues like this with their installation ? Does FreeNAS use a different SMARTCTL mechanism for its disk testing to the one I have used from the command line ?

 

[joeschmuck]

What are the models of the SSDs and which one is giving you the errors? What is the error message? How is this drive connected to your hardware? "smartctl" is the program you use for testing and reading the drive SMART data. "smartd" is what monitors the drives for smart errors during normal operations.

If this is always one specific drive then I'd replace the SATA cable first. If the problem persists, replace the SSD. If it still persists then maybe it's the SATA controller.

 

[leenux_tux]

Both SSD's are 40GB. model numbers are (camcontrol devlist)

INTEL SSDSA2CT040G3 4PC10362
INTEL SSDSA2M040G2GC 2CV102HB

Error message from FreeNAS GUI is

CRITICAL: 15 Nov 2016, 4:03 p.m. - Device: /dev/ada1, Read SMART Self-Test Log Failed

As far as I was aware smartd is a daemon (AKA subsystem?) that runs in the background, am I wrong in thinking that smartd runs smartctl to get stats on the state of the drives or does smartd have another internal command set that it uses ?

Like the idea of replacing the cable. Both of these drives are directly connected to the motherboard, not to the two internal M1015 that I have installed. I might swap the cables around on the motherbard first just as a test.

 

[joeschmuck]

smartctl is a separate program from smartd, meaning smartctl does not need smartd.

 

[leenux_tux]

Does smartd need smartctl though ? If I did something to remove smartctl would smartd not be able to monitor the disks ?

 

[joeschmuck]

You can't remove these items, they will come back when you reboot the system. Okay, well if you were knowledgeable enough you might be able to remove these but I doubt that will happen.

 

[leenux_tux]

Looks like I got a bad connector on the motherboard. I swapped the cables around on the mobo yesterday. Had another SMARTD error today.

CRITICAL: 1 Dec 2016, 3:59 p.m. - Device: /dev/ada0, Read SMART Self-Test Log Failed

The first time I had the error it was against ada1. Will have to try on of the other 4 free connectors

 

[nojohnny101]

The first time I had the error it was against ada1.

You should be referencing serial numbers to determine which drives are dropping, not the arbitrary "adaX" designators. They are randomly assigned and mean nothing.