(SOLVED) Random crashes on new 6.4 build and 6.4.1 update

[MacGyver33]

I'm getting random lock ups on a freshly installed Unraid 6.4 build. The system remains stable for anywhere between a few hours to 20-30 hours but then locks up and has to be hard rebooted. I tried the update to 6.4.1 and the problem still appears to be happening. Yesterday I ran Memtest for a few hours and it went through 5 passes on all tests and all CPU cores without errors.

 

I currently have the following dockers running but the lock up issue was happening before these were installed:

• CrashPlan
• PlexMediaServer
• plexpy
• Netdata
• GitLab-CE (disabled)

 

FCP only reports a few errors with the port mapping on GitLab-CE but I have currently disabled that docker to work out an issue with it gobbling RAM.

 

I had FCP in Troubleshooting mode but in looking at the FCPsyslog_tail file I see the last entry was at 4:38am this morning but the last diagnostic dump was at 11:09am and then the lockup happened sometime between 11:09-11:30am. I turned Troubleshooting mode on again in FCP and also added the Tips & Tweaks plugin to enable log syslog Archiving and will post an update the next time it happens.

 

See my attached logs from FCP.

FCPsyslog_tail.txt

fieldsuhs3-diagnostics-20180211-1109.zip

[MacGyver33]

Had another lockup today around 9:20pm. This time I pulled the Unraid USB drive to grab the syslog_tail before reboot. Nothing glaring jumps out at me in looking at the last few entries in the log. Also here is my hardware setup:

• ASRock AB350 PRO4 ATX Motherboard (Updated BIOS to 4.60)
• AMD Ryzen 5 1500X CPU
• Patriot VIPER 4 Series 2400MHz 8GB Dual Channel RAM (2x4GB)
• SanDisk Ultra Fit 32GB USB 3.0 (Unraid drive)
• 2x2TB Seagate Desktop HDD 64MB Cache SATA3 (1-Parity, 1-Disk Pool)

FCPsyslog_tail.txt

fieldsuhs3-diagnostics-20180211-2100.zip

[Transient]

When performing a memory test, I believe the general recommendation is to run it for at least 24 hours. Personally, if I was seeing lockups every 20-30 hours, I would run a memtest at least that long, probably even double just to be sure.

 

You might also want to try Passmark's Memtest instead of the one bundled with unRAID as it has more features (and is still free for personal use). There's some info on that here:

 

[JorgeB]

6 hours ago, MacGyver33 said:

AMD Ryzen 5 1500X CPU

There are known issues with Ryzen, are you disabling C-states?

 

 

[MacGyver33]

18 hours ago, johnnie.black said:

There are known issues with Ryzen, are you disabling C-states?

 

 

 

I hadn't seen this in the release notes and after checking my BIOS settings I found Global C-states set to Auto which likely means it was enabled. I'll try this and see how it goes. Tried another round of Memtest all day today without errors. I'll report back in a few days if this resolved the issue or not. Thanks!

[MacGyver33]

On 2/12/2018 at 4:40 AM, johnnie.black said:

There are known issues with Ryzen, are you disabling C-states?

 

 

I've had nearly 4 days of continuous up-time after making this change so I am going to mark this as solved. Thanks for your help!