I upgraded yesterday to 25.04 and since that time have experienced multiple freezes while the system is idle. There are no crash logs or errors, the system just stops responding and when I check the console the curse is no longer blinking and the console is unresponsive. When I first deployed 24.10 to this hardware (hardware summary below) I experience the same behavior but after changing the power supply control setting in the bios on the system to “typical Idle” the system was stable.
I have disabled C-states in the bios as of this morning to try and solve the problem as well, it has not frozen since changing that but I have been active on the system reviewing logs etc so it had not really had a chance to sit idle. CPUPower frequency-info shows an amd pstate driver (which I don’t remember being the case in 24.10) so I am wondering if the old fixes of setting the power supply control and disabling c-states are no longer enough?
cpupower frequency-info
analyzing CPU 0:
driver: amd-pstate-epp
Hardware summary:
Asrock Rack X570D4U-2L2T
5900x CPU
128GB non-ecc RAM - FYI memtest was run multiple times while troubleshooting the freezing on 24.10 and never had a failure and this system was stable as a windows server for over 2 years prior to moving it to Truenas
Just wondering if anyone else running a Ryzen CPU that has upgraded that had previous experienced idle freezing and are now seeing issues.
I’m running a similiar setup and I haven’t had any issues of this variety. I ran that cpupower command and I’m showing the same driver. Are you running any overclocks or voltage curve adjustments? I’ve gone back and forth with an undervolt of -30 in the curve optimizer in the BIOS, but never had stability issues either way. Is your BIOS current? There have been a lot of AGESA updates for Zen 3, if you have an old BIOS maybe the old AGESA code isn’t playing nice with the newer linux kernel? I’d start with a BIOS reset if you haven’t already and then try an update if you’re more than 6 months old on the version.
ASUS Pro WS X570-ACE
Ryzen 5 5600
128GB of ECC memory (4x32GB DDR4-3200)
If anyone has any commmnds they want you to run, I’m happy to play known good system for comparison.
Totally stock bios settings with the exception of power supply control (which was required for 24.10 to not freeze at idle and now c-states disabled as a test), reset multiple times during the 24.10 freeze troubleshooting but I can do it again.
The bios is 1 year old but that is the most current bios from the manufacturer so not much I can do there.
The beta BIOS from a year ago? Their stable BIOS is 2+ years old. The beta BIOS isn’t that old, it’s AGESA 1.2.0.C. I think my currrent BIOS is AGESA 1.2.0.Cc.
This looks like a pretty slick board. VGA through the Aspeed chip, so you don’t need a G variant CPU for basic video out. I’d love to have that.
Maybe reset the BIOS to be sure sure, but I don’t know why you’d get such instability from this update.
I am running the beta as I have reached out the Asrock and they basically said run the beta nothing else planned at this time (at least that was the response a few months ago when dealing with the 24.10 freezes)
Its been a great board until Truenas, ran windows server 2022 with lots of VMs flawlessly.
My only guess is that this has something to do with the amd-pstate-epp driver as I don’t believe that was the driver before 25.04. I could also try changing the power governor from powersave to performance? Not sure if that would help. So far it has been stable with C-states disabled, just less than ideal power consumption running this way.
I just noticed this post so wanted to comment on another potential AMD issue. Some of us on Discord have determined it must be AMD related that we can’t upload large ISOs to new 25.04 Instance volumes. 2GB uploads okay but not 5GB. Can you guys please confirm this?
I was using my laptop (5800U with Fedora 42) to my server (5600 with 25.04.0).
Since you’re having issue on a memory intensive action (uploading a large file), it might be worth doing some memory tests again to verify a stick hasn’t turned on you since you last checked them.