Periph destroyed after some time, again and again...

[chrWal]

Coud any FreeNAS specialist explain to me what is going wrong with my install since some weeks ? Disk ada0 goes destroyed repeatedly. I already changed the disk few weeks ago, then the SATA cable... Sometimes ada5 is destroyed as well (14:42:06). Then the disk is no more listed, and the pool is unreacheable.
If I switch the system off it stays waiting for stopping the ZFS PID, and eventually I have to switch it off manually.
When I restart the pool is back for some hours, and the same problem comes back again...
Hardware : last FreeNAS release on HP N54L, 6 disks 4TB in RAID1, 16GB RAM ECC.

Below are a screen with warnings and logs file. Things start to mess at 14:15:09.


Thanks, I don't know what to think and to do.

 

[Heracles]

Hi @chrWal,

The error message being about ATA error, the problem may be as deep as in the motherboard itself... The message is not about SMART error, so is probably not related to the disk. You also mentioned that you changed the disk anyway. Also, because it is mostly the same drive that is failing again and again, it really points to a hardware problem.

Do you have backups ? If you do not, do some ASAP.

FreeNAS does not care about physical position of its drive. As such, you can move drives from a place to another. If you do that and move that drive to a different port, does the symptoms change ?

 

[Alecmascot]

Just off the top of my head....
The N54L has a 150W power supply, with 6 off 7200 drives you are using half of that just for the drives. You also have twice the memory that it was designed for.
How are the extra two drives attached ?
Are you running the patched bios to allow full speed on all the sata connections ?
I would hang a meter on the 12v and see if anything untoward is happening.

 

[jgreco]

The 200W power supply in the G7 MicroServer series is generally too small even for four high power drives;

https://www.ixsystems.com/community/threads/proper-power-supply-sizing-guidance.38811/

You are very likely getting drive brownouts, and I don't think it supports hotswap, so once a drive drops, it isn't redetected. The N{36,40,54}L Microservers are well known for having a weedy power supply that often blows out even on four 5400RPM drives, and there are a few sellers on eBay that will provide a replacement 350W PSU. That might solve your immediate problem, but you will still be cooking drives unless you've been careful about the cooling.

 

[chrWal]

Just off the top of my head....
The N54L has a 150W power supply, with 6 off 7200 drives you are using half of that just for the drives. You also have twice the memory that it was designed for.
How are the extra two drives attached ?
Are you running the patched bios to allow full speed on all the sata connections ?
I would hang a meter on the 12v and see if anything untoward is happening.

BIOS is the patched one, extra drives are attached on a SATA card on the PCI slot included on the motherboard.

 

[chrWal]

Hi @chrWal,

The error message being about ATA error, the problem may be as deep as in the motherboard itself... The message is not about SMART error, so is probably not related to the disk. You also mentioned that you changed the disk anyway. Also, because it is mostly the same drive that is failing again and again, it really points to a hardware problem.

Do you have backups ? If you do not, do some ASAP.

FreeNAS does not care about physical position of its drive. As such, you can move drives from a place to another. If you do that and move that drive to a different port, does the symptoms change ?

I didn't know position doesn't matter for the drives. So I could switch 2 with no problem ? What I am worried about the most is the 4 in checksum for the ada5 drive. Yet this is the ada0 which is being destroyed repeatedly...

The 200W power supply in the G7 MicroServer series is generally too small even for four high power drives;

https://www.ixsystems.com/community/threads/proper-power-supply-sizing-guidance.38811/

You are very likely getting drive brownouts, and I don't think it supports hotswap, so once a drive drops, it isn't redetected. The N{36,40,54}L Microservers are well known for having a weedy power supply that often blows out even on four 5400RPM drives, and there are a few sellers on eBay that will provide a replacement 350W PSU. That might solve your immediate problem, but you will still be cooking drives unless you've been careful about the cooling.

I have 6 drives, 2 more on a SATA extension PCI card... I think I'll start with a power supply upgrade