SOLVED How to investigate random reboots ? => Bad PSU

[Marcet]

I'm getting closer to the suspect...
Keep you informed in a while.

 

[Marcet]

I've been doing some CPU Stress test with "System Stability Tester" :

1) 1M Pi digits 1 iteration 1 Thread all disks onboard => Ok
2) 1M Pi digits 1 iteration 8 Thread all disks onboard => Ok
3) 1M Pi digits 1 iteration 16 Thread all disks onboard => FAIL (Instant reboot)
4) 1M Pi digits 1 iteration 16 Thread NO disk onboard => Ok

So I switched the PSU for an old and trusty Antec 500w :

1) 1M Pi digits 1 iteration 1 Thread all disks onboard => Ok
2) 1M Pi digits 1 iteration 8 Thread all disks onboard => Ok
3) 1M Pi digits 1 iteration 16 Thread all disks onboard => Ok

Then I made 40 iterations of the same test, just to feel more satisfied :D

I owe you all apologies, not to have done that in the first place.

As I had bought this system fully assembled, I thought it has been thru QC check.
I was wrong. I'm sure they have tested the RAM, but they have not stressed the system.

FYI, the faulty PSU is a : Seasonic SS-600H2U 2u 80+ Gold

 

[depasseg]

FYI, the faulty PSU is a : Seasonic SS-600H2U 2u 80+ Gold

@Bidule0hm - want to do an autopsy on it?

 

[Marcet]

@Bidule0hm - want to do an autopsy on it?

:D

Would be interesting to know exactly what's wrong. But I will RMA it as it's only 1,5 month old.

I've so bad luck with gear nowadays, remember my thread about the faulty AsRock C2570D4I.

That's weird, I've been building PCs for more than 2 decades with carefully selected consumer grade components.
The moment I decide to change league and build server grade PCs, things gone wild :(

 

[Bidule0hm]

I would be happy to do it but as he said he will RMA it... :)

But yeah, shit happens, even with SeaSonic.

 

[Marcet]

I'm building a pfSense box for my next project : crossing fingers :D

 

[thetallest]

I'm pretty sure I saw this issue related to IPMI watchdog. There was something in the actual IPMI web interface settings, I believe that needed to be adjusted or disabled.

How do you disable the IPMI watchdog because if I try to empty out the fields so it does not go looking for it, it will not accept empty boxes ?

 

[depasseg]

How do you disable the IPMI watchdog because if I try to empty out the fields so it does not go looking for it, it will not accept empty boxes ?

I don't know. I only remember reading about it in another post around here.

 

[thetallest]

I don't know. I only remember reading about it in another post around here.

I found my IPMI setting my BIOS and I have turned it off I am hoping this takes of this issue having my system reboot at random times gets highly annoying ...

 

[Ericloewe]

Would be interesting to know exactly what's wrong. But I will RMA it as it's only 1,5 month old.

Probably some mundane QC issue.

One thing that stuck with me was a large Seasonic being reviewed (might have been an X-1050) with a defective primary side (some major component had released its magic smoke) - the thing handled something like 80% load before gracefully shutting down, all while still staying well within spec on the output.

 

[Marcet]

Probably some mundane QC issue.

One thing that stuck with me was a large Seasonic being reviewed (might have been an X-1050) with a defective primary side (some major component had released its magic smoke) - the thing handled something like 80% load before gracefully shutting down, all while still staying well within spec on the output.

So I believe 80 Plus Gold is just specs and has nothing to do with testing.
That is my first 80+ Gold. Not sure if this label should be trusted.

 

[Ericloewe]

80 Plus Gold

Power consumption only, nothing else.

These days, you can tell that corners were cut if a PSU is not Gold or better, but it's far from an exact science.

 

[Marcet]

My new PSU setup.



I know... You're jalous :D

 

[Ericloewe]

My new PSU setup.



I know... You're jalous :D

Meh, could be worse. At least those things were popular and decent back in the day. Pity it's group regulated, though, judging by the minimum load specs.

 

[Marcet]

Not sure I can sustain 10 disks with that one. It has only 2 molex lines.

 

[Ericloewe]

Not sure I can sustain 10 disks with that one. It has only 2 molex lines.

That's getting tight, since it's an old, group regulated unit. Voltages might not be within spec during disk spinup, leading to all sorts of fun things.

 

[Marcet]

That's getting tight, since it's an old, group regulated unit. Voltages might not be within spec during disk spinup, leading to all sorts of fun things.

Yeah. I'm getting a Cooler Master 750w from my local retailer today, this should make it.

 

[joeschmuck]

Let us know how things go after the new PSU is installed and you test out your system.

 

[Marcet]

Let us know how things go after the new PSU is installed and you test out your system.

So far, I'm running the server under heavy load with a http://www.coolermaster.com/powersupply/modular-gm-series/g750m/.
No problems other than the fact it doesn't fit in the 3U case ;)

The Seasonic will be shipped on monday for RMA.

In a few days, I'm planning to re-plug UPS and re-activate BIOS/CM whatchdog (software watchdog is already running).

 

[joeschmuck]

Well you can always put that Seasonic back in once it comes back from the RMA process. I prefer Seasonic to CoolerMaster but I know you were just dealing with a local distributor to get your system up and running again. But you know, if you can live with it and it works....