I have been running an Unraid server for a few years now. Never had any sync errors until now. Now with each parity check I am getting hundred of errors. I have tried it a couple times now. I am not getting any SMART errors. What is the best approach to this? The only thing that changed recently is I upgraded to Unraid 6.8.3. No recent hardware changes, power outages, etc. The only other anomaly I noticed is my Emby docker randomly vanished. I haven't noticed any other misbehavior. I have attached diagnostics. Thanks!

 

 

tower-diagnostics-20200406-1853.zip

Edited April 8, 20206 yr by lockdown571

If after a correcting check you get more sync errors (without an unclean shutdown) first thing to do is to run memtest.

Just now, johnnie.black said:

If after a correcting check you get more sync errors (without an unclean shutdown) first thing to do is to run memtest.

Thanks! I will try that out and report back.

I assume memtest just runs indefinitely? I've been running it over 24 hours now (13 passes) without any errors. What should be my next step in troubleshooting?

Do a couple of consecutive correcting parity checks and post the diags.

Here are two. Running another now.

tower-diagnostics-20200406-1853.zip tower-diagnostics-20200409-1635.zip

Possibly I wasn't clear, do a couple of consecutive correcting parity checks without rebooting and post the diags.

On 4/11/2020 at 9:10 AM, johnnie.black said:

Possibly I wasn't clear, do a couple of consecutive correcting parity checks without rebooting and post the diags.

Those were from last week, but from what I remember those were back to back without rebooting the server. I will run it a couple more times consecutively and post the new diags later this week.

9 hours ago, lockdown571 said:

those were back to back without rebooting the server

I checked the newest diags and there was only one check logged.

Did three consecutive parity checks (all had errors). No reboot in between. Grabbed the diags after each check (1536 is the most recent). Really appreciate the help!

tower-diagnostics-20200412-1950.zip tower-diagnostics-20200413-2137.zip tower-diagnostics-20200415-1536.zip

Many of the error sectors are the same, which suggests that on one pass they can be found bad and the next one found good and changed back, my prime suspect would be the Sil3132 controller, since some of these are known to corrupt data when both ports are used simultaneously, so first thing would be to replace it with a Asmedia controller (if you don't plan on needing more than those 2 extra ports).

8 hours ago, johnnie.black said:

Many of the error sectors are the same, which suggests that on one pass they can be found bad and the next one found good and changed back, my prime suspect would be the Sil3132 controller, since some of these are known to corrupt data when both ports are used simultaneously, so first thing would be to replace it with a Asmedia controller (if you don't plan on needing more than those 2 extra ports).

Thanks! Something like this? Also, glancing through some other forum posts, I should be able to shutdown the server, replaced the SATA controller, and reboot? Anything else I need to do to the array before replacing the sata controller?

Edited April 16, 20206 yr by lockdown571

Yes and yes.

Tried the ASM1061 card. Of course that did not work. Would have been way too easy. My mobo won't detect it at all. Any other SATA controllers you recommend?

 

Edited April 19, 20206 yr by lockdown571

Your MB uses the earlier "F1" BIOS dated 11/25/2016

Quote

Apr 11 08:52:42 Tower kernel: DMI: Gigabyte Technology Co., Ltd. H270N-WIFI/H270N-WIFI-CF, BIOS F1 11/25/2016

You should update to the latest one and try the controller again.

14 hours ago, lockdown571 said:

Any other SATA controllers you recommend?

JMB582 based controllers also work well with Unraid.

Updated my mobo BIOS and nothing. The card doesn't even post. I swear every time I update my Unraid server version I end up with a totally bizarre chain of issues. I wish I would learn my lesson. I will try a JMB582 card.