I upgraded my cache drive today to a Samsung 860 EVO 500G.  Before it was some other Samsung 120G.  Now I'm getting "CRC error count" errors.  This new drive replaced the old one.... same cable, same power connector, etc.

 

Do I have a bad drive?  The error count was 0 with the old drive.  This thing has been up for just 2 hours and has 8 errors.

CRC errors typically are a problem with the connection

Just now, trurl said:

CRC errors typically are a problem with the connection

Yeah, I've read that.  That's why I mentioned that it's using the same cable previously used with the drive it replaced.  I have also heard that if it clicks, it's a good connection.

 

I don't have a ton a cables lying around, and honestly it was working just fine with the old drive.  I'll humor everyone and power it down tomorrow morning, and unplug/replug it back in.

 

Are these numbers stored on the drive itself or in Unraid?... is there a way to reset it?

The CRC error counts are stored internally on the drive and never reset.  You basically want them to stop increasing.

Okay, just rebooted.  Before bringing it down I had 12 errors, immediately upon bringing it back up it was 12, but now it has increased to 13 after less than 5 minutes.

 

EDIT: by reboot I mean, powered down disconnected the drive, re-connected the drive, and powered back on.

Edited May 7, 20206 yr by eric.frederich

Just came across this:  https://superuser.com/questions/1294158/what-may-cause-very-high-crc-errors-on-ssd-apart-from-bad-sata-cables-if-any
Someone said could be issue with AMD SATA controller and the drive.  He recommended disabling NCQ but had a Windows registry solution.  Is it possible to disable NCQ on Unraid?... is it advisable?

It's a known issue with Samsung SSDs and some AMD controllers.

 

1 hour ago, eric.frederich said:

Is it possible to disable NCQ on Unraid?

It's already disable by default (Setting -> Disk Settings)

3 hours ago, johnnie.black said:

It's a known issue with Samsung SSDs and some AMD controllers.

 

It's already disable by default (Setting -> Disk Settings)

Thanks.  Can you point me to any documentation regarding this?  Like, is there a list of SSDs or chipsets / controllers?

The drive I upgraded from was a Samsung 840 120G, this is a Samsung 860 500G.

 

Are the errors being caught and handled?... or should I replace the drive?

 

This morning I see the count is now 21 after being idle last night.  It seems to have a new error every 31 minutes or so.

12 minutes ago, eric.frederich said:

Can you point me to any documentation regarding this? 

There no documentation, AFAIK, but if you google it you'll find a lot of similar issues.

Thanks... I found some things on both Samsung and AMD website... none of them seem to acknowledge the issue.

 

Is this only a problem on older AMD chipsets or even new ones?  This is an old AMD 970 board.

Would upgrading to a new Ryzen based system fix the issue?  I can't justify spending money right now on a new system, but I'm curious for down the road.

 

I just received the drive yesterday, so I should be able to return it to Amazon.

From my research, it seemed Samsung EVO was the way to go for SSD with DRAM cache.

What is the best 2nd choice?... what should I use instead of Samsung EVO 860?

 

EDIT: I know it's more of an opinion, but I see many options, Western Digital, Kingston, SanDick, Crucial, PNY, SK Hynix, ADATA, etc...  I can read reviews on Amazon but I'd like to know from an Unraid perspective if there are ones to use or ones to stay away from.

 

Edited May 7, 20206 yr by eric.frederich

13 minutes ago, eric.frederich said:

Is this only a problem on older AMD chipsets or even new ones? 

Pre-Ryzen.

 

13 minutes ago, eric.frederich said:

what should I use instead of Samsung EVO 860?

Crucial MX500 and WD Blue 3D are similar in performance, especially the Crucial though there's also a known issue with that one.

9 minutes ago, johnnie.black said:

Pre-Ryzen.

 

Crucial MX500 and WD Blue 3D are similar in performance, especially the Crucial though there's also a known issue with that one.

I was just looking at the MX500.  It's almost impossible to see on the specs whether these things have DRAM cache or not.  It seems the Crucial has something called Momentum Cache but that uses your system RAM as a buffer and is only for Microsoft Windows.