vmasip

Of course I respect opinions. And I respect someone is not interested. I respect silence. I respect opinions so much, that I've kept the good manners spite of that flagrant violation of common sense and socially acceptable human conduct. So you are welcome. I just pointed out the misconceptions about me you've used ( I said "you" but it's been 95% @MPC561 ) with bad manners when I was trying to help. Defend myself against bad words about me like "headline readers", that I don't read or don't pay attention, or, insults directly calling out word like stupid, imply more than once that I am problematic, or not skilled enough to copy paste (against detail oriented engineers). I am f**** defending myself all the time. I hope moderators are reading this,because it's completely exhausting. But above all I've been replying because @MPC561 has liked a lot putting words in my mouth which I haven't said, or conclusions I haven't bring out: "Think about or google it" -> I had already done (Can be checked) Bring out twice, even after clarification that I am confusing my motherboard. -> I am not. I didn't fingerpoint anything (maybe you translated wrong) ->that word has a bad underlying meaning and I just tried to help. Say that I hadn't read all the thread -> I did. Moderators, should I memorize or should you instead recognizing past complementary info should be quoted? I say that setpci doesn't work in most cases. Or I don't believe it -> I just really believe it, just that the bad case, is a possibility. Like I proved. He has tried to make the forum believes that I try to steal the authority of the reddit commad -> I just pointed this out, to defend myself that I am not that kind of person. I hate when people moves my words, or move the object of debate continuously. And, I am really glad that has been stopped with his self-silence. It's s not acknowledgement, and is painful with sentences like this. (.. because all, all,....... all? I would say that violate norms as you say with insults is not friendly) But at least, I have to stop to defend myself which is a good point, a relief, and easier to keep on using good manners.

I can keep the manners and you don't. You are proving being rude and disgusting , and I hope moderators take note. I can read, you don't. I know for sure it worked for you, but not me: once again here goes this capture:) This is the 4th time I say this. The conclusion I said earlier for everyone: be aware , because maybe at some point anyone gets into the same problem than me(for an obscure reason -> setpci not working ) and that they can test the alternative I mentioned previously. I am not going to make a video you to believe it. Just go beyond self ego/self awareness , and try to think it's possible that your setpcie doesn't work because of causes you don't know. Or you know everything? XD Yes. And this side effect with the network can appear when the system is working normally, but mainly, when it has changes which affect to its energy. So with a so dockerized busy system, his ethernet driver is working at the best scenario - I mean it is opossite conditions to bugzilla inestability kernel bug report - . The normal case scenario to test this stability would be: disconnecting ethernet, sleeping the system, having the computer not working at all completely idle, and having the system in deeper c-states (I've proved all of them with no issues). Not working the NIC creates change in power, which causes THE BUG was reported. You can read it following the bugzilla thread. I guess is not the case of the containers of @alturismo. So, from my point of view, is not a very "detailed" or useful test. I just was trying to make this point from the beggining. Is completely reasoned here, and I've been able to do it without being rude. Do you get both? This proves you read diagonally. You don't have to tell me, I pasted it with a huge reddit preview. LOL

I've told you at the last post: I am talking about this: I replied to this at the same posts you are inquiring about (I quote myself): Ok, I just assume all the test , of course the ones I have done, are with a fresh installation (unraid and ubuntu) with no docker at all,no processes. And obviously no more hardware. @MPC561 BIOS settings. Just running ubuntu doing nothing, I mean, just running ubuntu default daemons. So I think that my test were taking into account what you've trying to explain. 2 OS and 0 load more than default daemons. Don't you think? You don't have to give any proof because I believe you. Is you who don't believe me XD . But as I said I said - and you said - , I am happy with my command so , I don't stuck around. I just pointed out for others who may encounter same problem. I don't know why you keep pushing this against me. You can keep me saying that I can't copy and paste. I will keep for advising others that it can be possible that this doesn't work for you, but don't make me record to write it here for being believed. So, fact: I have copy and pasted it, and tried like 10 times with and without sudo. At the go file or just at the terminal. In ubuntu and unraid. I know I didn't have to. How is he going to cross check anything without deactivating it? I mean, please. If can't deactivate he can't cross check. Above all, because I have read from you in the past (90% sure it was you, but maybe another one), that when you encountered this issue when new unraid kernel fucked realtek, you tried to overcome it, ASPM said it was enabled, but in fact you were still being stuck in same c states. So my advice is to deactivate it, then other typical things. My thought was, first this basic steps from that point, I could try to recreate and approximation to his main apps - if that is possible- and check if I've got issues too. If it's not an option, there is no space to anything. Do I have to dig very much in many pages to find out that is not an option? That's all my point.

I had brain-keyboards issues . Sorry. Just, I meant that I pointed to the other n100 box and after that I said: You can see it at this link. So obviously, china box is not mine, but here there was speculation about depending on the cpu and core kinds, time at different C-States can be different. I just wanted to point that other n100 users had better c-states performance. And, you are at that thread, yes XD. I did it right. Maybe I should do a video of me using your commands to believe me, but I won't get angry. Neither because of you for calling me idiot for doing bad a copy paste. I just tried to help @alturismo , and he called me idiot because his huge last post was - according him- a part of many. So, I won't go through everypage finding every piece of information. If he doesn't even mind to quote those previous pieces of information. I just made my reasoning with the huge post. If he doesn't like the help of a rookie much less intelligent than you two - my lords- , just don't answer. Is my advice for him. @MPC561 , If you read again, you'll notice that I defend from attacks, I don't attack, I don't call anyone - not even implicit way- idiot, annoying, or similar. You can't deny that. I haven't fought at all, word which implies attacks, and is not the case. I defend myself against inquisitorial replies. I made a recommendation offering my help. If you don't like me trying to help, just say who can try to help or who does not. But that doesn't fit in any logic with this translated sentence yours : "Sorry, but it's the last time that everyone here wants all the answers from me for the world's problems." That was the sentence made me try to help So you can say , "I have done all that." Is more polite. More gentleman way. And we can start talking about daemons, OS, docker configurations. I wouldn't pretending everyone remembers all the post written here - without proper quoting to them - at the last piece of info you have given. This is an eleven pages thread. If someone need this info, doesn't read it or similar, the normal way is answering quoting to it. I don't say it heartless or angry. I just say it because is the normal thing. The accusation way, adjectives like headliners ... that way you don't win a "friend" to debug., who,ups, maybe know more than you about docker or linux kernel. maybe not. And yes, I know you are trying to confirm instability issues about the famous commands. But the case and my points is the same

I have read al this thread. Any meany others. I don't memorize them, and above all , i don't re-read them again if I am not looking something I remember. I know your name, so I know you are not new here. I think your questions and great. You made a huge post with a lot of info, so I guessed is all there. I don't want to have to investigate pages and pages to find out every detail you have posted in this forum. Sorry but no, You already know it is a docker issue (when you said "that's exactly the issue" ) , as you said possibly a unique case (because when you say docker you mean docker containers running, not the docker container service running, isn't it?). And it's not at all specific of this motherboard, so why in this thread where the talks go about this motherboard? I mean, is not related at all with the motherboard. It's specific about the container, it's OS, it's daemons, it's apps, configuration used to deploy it. It's guess is not related at all with the motherboard, because this motherboard and this OS you say let you go to deep CStates. If you agree with this, I recommend use a new thread,specific for the users who have same docker apps. Because the issue is for more users than the owners of this MB. If that is the case, you can provide more specifics about the aspect of your docker run,compose,build, so on. We can help to reproduce or others who run your apps succesfully with powertop.

After saying After the capture I said: There you'll notice that I use the capture from there. In addition I say "other machines" like saying not mine. And the address thing , yes, is hugely confusing. I've tested a lot. I made with and without sudo. I've made a reset and retry several times. None of it worked. The echo command at the first attempt. Curious, because echo didn't work for you. But as you say, very happy

Lack of info: Which ram do you use? which nvme do you use? Not everyone of them allow deeper c states. Which sata disks do you have? Something connected at pcie? I guess you don't by your picture. Haven't you stopped docker to try this? Why? Disable it and stop the array to do this test? Advices: ASPM enabled messages can be not very useful. You only can trust completely when is disabled. Disconnect usb , mouse, hdmi, and anything you can have connected. How in order to get the results of powertop? Disconnect just after: `sleep 10 && powertop --auto-tune && sleep 10 && powertop -C report.csv -t 100` My USB with unraid doesn't affect. I have installed ubuntu at different nvme and booted with it and without any usb with same results(I connect it to the phishycal usb connectors of the motherboard). Follow @MPC561 BIOS settings completely as it is. If you go beyond this, is uncharted territories. - I have only tried different sata settings by myself with no changes at all- Maybe the new unraid version has sightly different kernel, which encounters problems with your nvme or ram as from unraid 6.12.5 version linux kernel created the realtek issue with our asrock NIC. You can use new and old ubuntu to find this out. (Or new old unraid) But first of everything disconnect the docker, is simple and an clear root of problems

Thanks for the pcie addresses. Useful info I could not find googling it. Anyway, somehow, my echo command worked. And for those who wonder, I have same motherboard as everyone here and addresses were not fine for my motherboard because setpcie command didn't work ,so everyone have to find their own addresses.

Ok, they are different ones. I didn't found the address beyond the part of : 00:1c.0. I mean, this part is easy with: But I haven't found how you have found out 0x50 and 0x80 part. I gues 0x42 is the "yes". Somehow, the method I pointed out worked echo 1 | sudo tee /sys/bus/pci/drivers/r8169/0000\:02\:00.0/link/l1_aspm You say it wouldn't. Was that because something I am gonna find out from now? Because by now...: And amazingly, Now I go beyond 60%, something I think you don't have, and I didn't have disconnecting ethernet. Curious. I think without Unraid UI and/or USB this test could go higher. But now, we know this is the way and everything is correct.

Ok, I just assume all the test , of course the ones I have done, are with a fresh installation (unraid and ubuntu) with no docker at all,no processes. And obviously no more hardware. @MPC561 BIOS settings. Just running ubuntu doing nothing, I mean, just running ubuntu default daemons. And I have seen other machines with n100 having better performances. This is a n100 ChinaBox. You can see it here: https://www-hardwareluxx-de.translate.goog/community/threads/die-sparsamsten-systeme-30w-idle.1007101/page-102?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=en#post-30269188 I just wanted to dig into this, but... google is not offering me a lot of info about what is awakening the processors.

I did, I googled and chatgpted that. But every content is about getting into higher states and it's problems, but no content about the % of time. I've made the same assumption, but I've seen captures with higher results (80-90%), and I thought, hmmm, interesting... Is something not right if you get only half?

Great. I've got same board. So you haven't seen stability issues by now. I'll test this myself too. My question is: If the machine goes into C8 states, and keep in that state, shouldn't the % increase over time? Why it doesn't go higher than 40~50 %?

Hello From a reddit forum: Which one is correct? Or are both the same? By the way, I tested this with Ubuntu, with ubuntu installed at nvme. Tested with other nvme brand too. ASPM output was the same for me: disables for one pci and realtek. (But as i said in my posts, disconnecting ethernet reveals c8 states are possible with ASPM disabled.) So, with Ubuntu, I had no usb connected, not keyboard and no mouse. But still: So, no idea why is this. I explained this here:

Hello, I have been able to reach higher states than c3 disconnecting ethernet. So I guess, there is no a problem of ASPM beyond kernel NIC Realtek bug. I guess, because I get higuer than c6, but ASPM lscpi results are the same, and higuer states doesn't look like stable. So, the results comes with a few caveats I cannot solve and I am being completely ignored at "Reduce power with powertop" thread.

Does this still being a restriction?