AMD based system regularly freezes after a few days

[SweetAndLow]

This conversation makes me think.

I respect the huge competence gathered inside the Freenas community. You guys really know how to make a highly reliable, safe and effective professional server. But I also see a clear opportunity for Freenas to fill the personal file server market with a simple solution running on old hardware. I would love to see this opportunity taken seriously. Boring it may be, though.

For me it is not enough just to say that for a server you need server-grade hardware. Or to avoid possible corruption in the file system you have to have ECC memory. I need real evidence of the real problems caused by consumer hardware and normal memory. Like how often and what kind of corruption is in real life caused by non-ECC memory. Instability is an understandable problem and probably difficult to predict. And again, would be nice to hear some facts. Like how big portion of AMD based installations have stability problems.

There must be a big need for a simple open source file server system built on old useless hardware. Pretty much everybody has pieces of hardware laying around. My short personal experience with Freenas strongly indicates that crap hardware is useful for file server purposes. Performance is excellent and I have seen no other problems than stability.

Enormous amount of everyday personal and business applications is running on crap hardware. I cannot see data integrity as a big problem, except when caused by hard disk failures. For me this means that the same hardware should be useful for personal file servers. Business critical applications like company ERP systems are a completely different story.

Linux based systems I have seen running on different hardware platforms have all been rock solid with absolutely no problems. What makes Freenas so special?

Again you are thinking about this wrong. It's not about ecc memory. That's a nice to have and usually comes with hardware that is actually good. The problem is the cheap motherboard using low end parts. For example realtek nic, realtek Sata chip sets and random wifi and audio stuff. All of these things cause instabilities and performance issues. In top of that the driver support is terrible. The second you move away from the cheap gaming builds you will not see any of these parts. This info is plastered all over the place.

My short personal experience with Freenas strongly indicates that crap hardware is useful for file server purposes. Performance is excellent and I have seen no other problems than stability.

This statement makes me sad, if it's crap hardware don't buy it in the first place or throw it way. Your personal experience is wrong and crap hardware is not useful for a file server.

Sent from my Nexus 5X using Tapatalk

 

[joeschmuck]

Getting back to your original problem of stability.

You need to run some stability tests such as MEMTEST86, run this for 5 days consecutive (I say this long because you have some issue and you need to rule the hardware out and it can take time). If you have no errors then run a CPU stress test like Prime95 for at least 2 hours, but in your case, maybe 6 hours. If you have no issues then you likely do not have failing hardware but a compatibility issue with running FreeNAS. I suspect you have a hardware failure.

Now the RealTek NIC can cause issues as well, I have first hand experience with this. Disabling this in the BIOS and installing an Add-On card would fix this issue.


None if my suggestions are to sidetrack you from the fact that you may be using inadequate hardware for a FreeNAS machine.

 

[DrKK]

I need real evidence of the real problems caused by consumer hardware and normal memory. Like how often and what kind of corruption is in real life caused by non-ECC memory. Instability is an understandable problem and probably difficult to predict. And again, would be nice to hear some facts. Like how big portion of AMD based installations have stability problems.

Sir, that's just not the business we are typically in as a community. We are not here to compile statistics, to make available the precise data and research someone would require in order to replicate and validate our results and knowledge. Etc. You can do that yourself. It is not a refereed academic journal, nor university class. We provide expertise---for free. One is of course at complete liberty to either select, or not select, using the available expertise. There are several people, for example, on their own personal blogs or on reddit, that provide FreeNAS support; most of these we have drummed out of the forum for not knowing what they're talking about---but they are available too, as well, if you are insufficiently convinced of the bonafides of the people in the forum, or don't trust us to know what we're talking about. If someone wants to collect "statistics" or otherwise satisfy people that "must have facts" before they will trust someone with 5 years of experience and 4000 posts in the forum, well that fine, but you certainly cannot transfer the burden for *that* onto us. You'll have to do that research yourself, for example, by noting how few posts from long-term FreeNAS users are in the forum on AMD systems, and by noticing the proportion of AMD users that appear to be having problems. You'll have to assess for yourself whether or not people that listen to danb35, or me, or erieloewe, or stux, or anyone else, are making a mistake.

As for the last question, "What makes FreeNAS so special?" Maybe everything. Maybe nothing. It depends on your needs, technical understanding, requirements, and personality. Only you can answer that. Some people, once they join us, know that they are finally home. Others, do not. Only you can decide.

 

[CookiesLikeWhoa]

Possibly off topic, maybe a bit on topic; if buget is a concern, go used. A system can be put together using some pretty awesome hardware for very little money that have proven track records if you're willing to dig around the used market.

As far as AMD goes; it looks like AMD might finally have a winner with their new chips. If SM and other manufactures jump on the train and start producing quality server motherboards, you'll likely see a lot more AMD posts.

 

[Jussi]

Sir, that's just not the business we are typically in as a community. We are not here to compile statistics, to make available the precise data and research someone would require in order to replicate and validate our results and knowledge. Etc. You can do that yourself. It is not a refereed academic journal, nor university class. We provide expertise---for free. One is of course at complete liberty to either select, or not select, using the available expertise. There are several people, for example, on their own personal blogs or on reddit, that provide FreeNAS support; most of these we have drummed out of the forum for not knowing what they're talking about---but they are available too, as well, if you are insufficiently convinced of the bonafides of the people in the forum, or don't trust us to know what we're talking about. If someone wants to collect "statistics" or otherwise satisfy people that "must have facts" before they will trust someone with 5 years of experience and 4000 posts in the forum, well that fine, but you certainly cannot transfer the burden for *that* onto us. You'll have to do that research yourself, for example, by noting how few posts from long-term FreeNAS users are in the forum on AMD systems, and by noticing the proportion of AMD users that appear to be having problems. You'll have to assess for yourself whether or not people that listen to danb35, or me, or erieloewe, or stux, or anyone else, are making a mistake.

As for the last question, "What makes FreeNAS so special?" Maybe everything. Maybe nothing. It depends on your needs, technical understanding, requirements, and personality. Only you can answer that. Some people, once they join us, know that they are finally home. Others, do not. Only you can decide.

Certainly you are right. It would be nice to see statistics, though. And actually you gave some above.