Continual unRAID freezes

[Daken00]

Hi all!

I'm running unRAID 6.8.3 on a server with a Gigabyte Tech B450 I AORUS PRO motherboard with an AMD Ryzen 5 2600 processor, 16 GB DDR4, RAID 4 - 1 8TB Parity drive, 1 8TB storage drive, 256GB cache drive. Plug-ins are CA Auto Update Applications, CA Backup / Restore Appdata, Community Applications, Dynamix Cache Directories, Dynamix SSD TRIM, Dynamix System Temperature, Fix Common Problems, Nerd Tools, Tips and Tweaks, Unassigned Devices, Unassigned Devices Plus and User Scripts. Docker containers are Plex, Tautulli, Sonarr, Radarr, Jackett, DelugeVPN and rTorrentVPN. All are bin-hex or linusserver.

 

I have been running unRAID for about a year. In December, I started getting server freezes (no Web UI, console unresponsive) that required a hard reboot. And this seemed to be happening every 3-5 days. No consistent timing. I've worked with the guy who built my server, and we've tried a few tweaks, but nothing lasts more than a week. I had even gotten to the point of daily reboots, which I wasn't happy about but did seem to keep the freezes at bay. I've had suspicions that DelugeVPN might be causing issues during downloads, since most of the server freezes were overnight, but I've had a couple during the day, including this morning, when nothing was downloading. Today all that was happening on the server was a local Plex stream. I mirror my syslog to the flash drive, so I have full logs back to the beginning of the month (attached). That includes four separate freezes (2/9,2/13,2/16,2/19). Any help you guys can provide would be greatly appreciated.

 

syslog.zip fett-diagnostics-20210219-1534.zip

[ChatNoir]

Hello,

Did you check this section of the FAQ ?

 

 

[JorgeB]

Also run memtest, multiple programs are segfaulting.

[Daken00]

@ChatNoir I had not seen that question in the FAQ, Working down through it now.

 

@JorgeB I had run memtest a few months ago in trying to figure this out, but only ran 1 or 2 passes and had no errors. How may passes should I let it run?

[JorgeB]

16 hours ago, Daken00 said:

How may passes should I let it run?

The usual recommendation is 24H, though in most cases if there's a problem a few hours are usually enough to find it.

[Daken00]

9 hours ago, JorgeB said:

The usual recommendation is 24H, though in most cases if there's a problem a few hours are usually enough to find it.

 Okay, it ran over 24 hours and a full 8 passes with no errors. Hopefully we are good there. Now that that test is done, I'll start looking at the FAQ stuff.

[Daken00]

Server locked up again overnight. I took a pic of the console before I restarted it (attached).

@ChatNoir I've gone through the article on the lockups, but I can't find the Power Supply Idle Control setting in the BIOS (version F50) and I don't believe the RAM is overclocked. I am going to try upgrading the BIOS to the newest version (F60h) and see if that helps. Let me know if you see anything in what was spit out before the lockup.

[ChatNoir]

From the Manual : https://download.gigabyte.com/FileList/Manual/mb_manual_b450-aorus-pro-wifi_1002_e_190528.pdf

I see mention of :

• Global C-state Control
• Power Supply Idle Control

It looks like it is in the Advanced CPU Core Settings section. But this could change depending on BIOS version.

 

I agree that with only 2 sticks of RAM at 2133MT/s you are good for a 2600.

[Vr2Io]

It look like relate idle problem, if everytime call-trace have " RIP: ... cpuidle_enter_state  ...... ", then it should relate this. My suggestion was try without start array or mount any storage, let system in most idle status, then check does trace will happen within 1hrs.

[Daken00]

I found the Power Supply Idle Control and Global C-States setting this morning. Power Supply setting was set to Typical current idle ( I think I had changed this previously), but Global C-States was still at Auto. I disabled that, so now I'll wait and see if the server stays up.

[Daken00]

After almost 8 days of uptime, I'm cautiously optimistic that the global c-states setting being disabled has fixed my issue. Thanks for all the assistance!