Just recently started after being stable for a long long time. Happened while I was sitting at the desk once, everything in the tower just goes dark. Corsair led ram stays lit, which is odd. Won’t power back on unless I toggle the psu  switch first.

joeserver-diagnostics-20210824-0939.zip

Edited August 26, 20214 yr by joecool169
Solved

Replace PSU

I think you're right. Just curious what made you say that so quickly?

 

1000W EVGA not even a year old - doesn't mean it is good though I know.

Jason had problems with his EVGA 1KW PSU too - warranty it if you can

 

 

1 hour ago, joecool169 said:

I think you're right. Just curious what made you say that so quickly?

Probably this.

 

5 hours ago, joecool169 said:

everything in the tower just goes dark. [...] Won’t power back on unless I toggle the psu  switch first.

 

Thanks for the replies,

 

 Bought a new psu yesterday. 1300W EVGA. So far no shut downs.

 

Gave it a little stress test last night. Found that with 100% gpu load and around 30% cpu load my system will pull over 800 watts. Perhaps the previous 1000 watt psu was not quite enough.

 

I'll mark this as solved.

 

Just did it again 😡

Edited August 26, 20214 yr by joecool169
not fixed

I'm thinking motherboard at this point.

Since you are using a Ryzen CPU, did you check this FAQ section :

 

You seem to be over the CPU specs for RAM settings for your configuration.

Also check your Power Supply / C-State status.

I could see that setting causing a lockup as the post suggests. But I've been stable for months. Now suddenly the system shuts down. This is not a freeze or lockup, this is the system goes completely dark with no warning. Temps are all good. The only thing I have changed lately was moving governor off performance mode so the a/c in that room doesn't work quite as hard.

I'm thinking motherboard at this point.

If it's not the PSU the board is the next logical culprit, but also make sure CPU is not overheating.

Before replacing more components, I'd start trying to narrow it down further - do you have any stats tracking set up that you could share (grafana etc)? Are you tracking your power usage at all so you might see potential for over current protection to trip?

 

As far as narrowing it down, pull out everything but the ram and run memtest for a bit, if that all clears, pull down to the minimum ram (single channel / one dimm) and stress test the GPU, and so on.

Also, I don't suppose you're using one of the new fractal torrent cases are you? Even if not, if you're using a fan hub, it'd be one of the things in the chain I'd remove during testing to narrow things down 👍

It's so random. Going to be hard to find. It's been running since Thursday at 7 with no issues. It might run 4 days and not do it.