Hello, I am hoping to get some help with my unstable UNRAID server.  every few days my server seems to crash, sometimes it is completely inaccessable (no ping), sometimes I can ssh in but no gui.  Even if the sever is running normally if I restart a docker, try to stop the array, or reboot the serve it can cause it to lock up and require me to do a hard reset using the front panel button to get it working.

 

 I have run memtest with no errors, I have slowed the speed down to 2133mhz with the same results.  I recently replaced the motherboard, processor, and power supply and the results are the same.  I am running out of ideas to get a stable system, I have run an Unraid server since the 4.x.x days and its only in the last few months that I have had nothing but trouble.  I have attached my diagnostics.

 

Thank you

volknas-diagnostics-20230618-1042.zip

Start here:

https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=819173

 

After the server was idle for several hours today, an attempt was made to watch a show using Plex.  As soon as the show loaded the server crashed, no ping and display indicated a fatal kernel panic.  This maybe points to the server having an issue with c-states as you pointed out for me me to read, after long period where it was idle it crashed as soon it had to come out of some sort of power saving state?  I thought that the AMD c-states issue was only present on the first generation ryzen's and not what I am currently running (5700x on x570).

 

I have tried disabling c-states to see if that helps.

After some more reading I was wondering my server crashes could be related to a docker macvlan issue?  I have a number of dockers that are set to bridge mode and a number of other dockers that are set to custom with a dedicated static IP address.  This issue has been reported by other users that when using macvlan and using dedicated ip for dockers when using one LAN interface for the server it can result in crashes.  I have followed some guides and have dedicated one NIC to dockers only.  I am hoping I am on the right track and can start to seem my stability improving.

5 hours ago, volkr said:

After some more reading I was wondering my server crashes could be related to a docker macvlan issue?

Could be, enable the syslog server and post that after a crash, macvlan issues usually leave call traces before the crash.