Things were running so well... until I saw a red X by one of my drives and it saying 643 errors.  Oh man!!

 

System has only been up and running a couple of weeks, and all the drives were brand new.

 

I guess I'm looking for advice on what next steps I should take.  It says the drive is being emulated.  

 

I'd like to attach an external USB 3.0 drive and back things up... I haven't really done any backups yet or figured out the backup process yet.

 

Any advice would be appreciated.  thanks.

 

attached is my diagnostic logs.

 

brownbear-diagnostics-20170623-2310.zip

You're using the Marvell SATA ports on the board, these are known to frequently drop disks, you still have a lot of free Intel ports, use those, if you really want to use the Marvell ones go to Asrock site and update their firmware, it appears to fix or at least mitigate the problem.

Ya, I was thinking about that... I already have done the firmware upgrade a couple weeks ago and things appeared to have be working great.  I'll have to give the other ports a try as you suggest.

 

I have trouble believing it is a problem directly with the drives themselves.... fingers crossed. 

 

Thanks.

Edited June 24, 20179 yr by DarkHorse

Didn't remember you had already updated firmware, this confirms my suspicions that the new firmware helps but doesn't completely fix the problem and the Marvell ports should not be used on those boards.

ya, I think you might be right.  So, I should probably move all of my drives off of the Marvel ports.  

 

I now have all the data I'm concerned about pretty much backed up.  What is the best way to move the drives over to the non-Marvell ports, while minimizing rebuild time and preserving as much of my current dockers / VMs / and data?

 

Can I move all three drives (2 8TB data drives and 1 8TB parity) all at once?  or, to preserve the array as currently setup, I need to move just 1 drive at a time?

 

thanks.

 

It's very easy, shutdown the server, move all disks off the Marvell, power up and rebuild the disable disk.

 

PS: before rebuilding check SMART for the disable disk since it's not on the diags because it dropped offline, but it should be fine.

Forgot, to rebuild using the old disk:

 

https://wiki.lime-technology.com/Troubleshooting#Re-enable_the_drive

Thank you for the links... the rebuild completed successfully.

 

My EP2C602 board has 6 non-Marvell ports, 2 are SATAIII 6 Gbps and 4 are SATAII 3Gbps.

 

I moved all of the array drives over to the non-Marvell 3Gbps SATAII ports.  My cache drives were already on the non-Marvell 6Gbps ports.

 

I guess I might see a little bit of a performance hit, especially with the Parity drive?  Maybe I'll look for a LSI 9207-8i board and just place all my array devices there.

 

Thanks again for the excellent advice.  Much appreciated.

There's no performance penalty for current HDDs on SATA2 ports, only for SSDs.

Ah... good to know.  thanks.  I ran a quick Disk Speed Test and I'm happy with how things are... was running a couple VMs at the same time the test was running... it does look like the Seagate Ironwolf drive (used for my parity drive) does indeed perform better than the Seagate Archive drives, which I anticipated.

 

Edited June 26, 20179 yr by DarkHorse