Badly need a stable Linux system with ecc?

[k666k]

What would you say is a good setup that’s common and bulletproof like workstation or server level with good ecc documentation for novice?

[Zedicus]

@Zedicus

option 1. uber low cost tank build.
an opteron H8SGL supermicro board with a 6380, or 6366HE CPU. this uses real ECC REG DDR3 RAM, can be viewd by OS utilities, has lots of cores, makes for a good VM host and has just enough PCIE to get some HBAs and a 10gb nic. under 500$. this stuff is old, but i am still running this exact setup as a NAS at a location and it is bulletproof. its uptime is measured in YEARS.

option 2. all it takes is money
EPYC stuff. used or new if you want. same hardware benefits as the opteron rig, you get working ECC DDR4 RAM, more and faster PCIE, newer CPU instruction sets, but more expensive and equal or more power draw. an H11ssl and a EPYC 7252 are about as low cost as you will get and you will easily eat 1500$ trying to complete this build.

this seems to be going in circles.
Right, Reliable, Cheap. pick any 2.
if you go ‘out of normal’ by bolting ECC onto AM4 or something, you are skewing the data.

[k666k]

H11ssl is an amazing product, I can get it for new for around £600 I think v2(need to double check that), and it’s popular as you say.
The scary thing is I never worked on server hardware.
I think its the one, but I like tabletop things way more than mounted things, specially if it’s something I would 100% move to keep safe.
It’s predecessor Had To Be Moved recently as I was doing high frequency welding in the garage next door lol

[sami_m]

I did some testing on my Asus and it seems to give out ECC error information. It was bit tricky to get machine to state where it would boot but be unstable enough to cause ECC errors. I finally managed to get it setting DRAM voltage to minimum 1.2V and then through trial and error lowering timings and when I found spot where the system wouldn’t boot. I backed of a bit and raised memory clock until I either started to see ECC errors in Memtest or normal errors (if the reporting would not be working).

So I got ECC error which means that the system has working ECC error reporting (and correction as there are no actual errors reported):

After that I checked if errors are reported also in linux. On dmesg following was shown:
[ 939.354873] mce: [Hardware Error]: Machine check events logged
But edac-util was not showing anything. But still success as the errors are logged at least some level.

I have been doing some reading on ECC and from what I was able to find that on some Asus boards there have been good reports on working ECC. On Asrock not so much, they even have some AM4 rack boards that are mentioned to support ECC. But don’t actually report errors.

In general we as in customers should request that manufacturers allow consumer platform to have possibility to good reliable computing (working ECC). Why my or your home computer should be unreliable.

[k666k]

You are an absolute legend, thank you so much for checking this, I appreciate the help I have had on this thread so much.

So this means you can trust its 100% doing its thing periodically but I won’t be able to set up a programable response from an server error trigger right?
But still amazing for a super stable work or gaming station it seems

I think I want to follow the direction of @Zedicus and go epyc as you say, the manufacturers really don’t have the incentives or customer demands to sacrifice server sales lol.

I’ve recently seen our leader lol, wendel build gamers nexus a really stable and stress free looking threadripper zen3 16core beast, but the Gigabyte MC62-G40 motherboard(£1,100), and it is available in shops, so now I have a very expensive backup plan lol.

As a guy who dose this as needed(every few years) and am not into buying new things that often I don’t know what pressure I can put on these companies, we can only complain and educate as we are already all doing at this point, people will find a way in their expert fields and share with the none experts lol, the kindness of people will always triumphant.

One day we will have a 10+ year phone maybe I hope.

[k666k]

Yes Samsung is great, I have no issue getting cheap used ddr4 and ddr3 Samsung.
I mean if their working with me, letting me know when their skipping bits regularly and need replacing I guess.

Would you say that’s a reliable long term strategy?