4 hours ago, MPC561 said:

Don't know. Me did not open one.

May that is one of the bigger Part of those Problems.

1 hour ago, prostuff1 said:

Going to ask a stupid question but what is the actual power difference between 6.12.4 and 6.12.6 where this kernel bug is?

Depends on your mainboard and the used extension cards, as I told nearly all MB´s with RTL network chipset are affected. With the N100DC-ITX it´s not a lot (6.12.4 I had 7W mean consumption, with 6.12.5 I had 9W) so 2W. Other systems may have 5W+ difference. With the energy prices we have in Europe in some countries we discuss here over 5 years of additional energy cost of approx 100 euro in addition. With my system here in Germany and my energy price with 2W more over 5 years it´s only 29 Euro.

 

1 hour ago, prostuff1 said:

A better analogy would be your web browser gets an update and stops working on your laptop so you call Dell to fix it cause they sold you the machine.

Surely not. The correct example would be: Dell sells me a laptop with an OS with hard integrated Browser from another company which stops to work after an update. In this OS I have no admin rights and also BIOS is locked and so it´s not possible to  replace the browser/install a new one. Dell sends me to the browser supplier instead of solving the problem with the browser supplier itself.

 

Limetech selled me UNRAID OS and I expect that it works. They are also my first level support. If they have not a solution they should run to the kernel development team and not me. 

 

49 minutes ago, sonic6 said:

May that is one of the bigger Part of those Problems.

It´s of course much more evil than the problem itself.

 

 

 

Edited December 14, 20232 yr by MPC561

14 minutes ago, MPC561 said:

Depends on your mainboard and the used extension cards, as I told nearly all MB´s with RTL network chipset are affected. With the N100DC-ITX it´s not a lot (6.12.4 I had 7W mean consumption, with 6.12.5 I had 9W) so 2W. Other systems may have 5W+ difference. With the energy prices we have in Europe in some countries we discuss here over 5 years of additional energy cost of approx 100 euro in addition. With my system here in Germany and my energy price with 2W more over 5 years it´s only 29 Euro.

 

Surely not. The correct example would be: Dell sells me a laptop with an OS with hard integrated Browser from another company which stops to work after an update. In this OS I have no admin rights and also BIOS is locked and so it´s not possible to  replace the browser/install a new one. Dell sends me to the browser supplier instead of solving the problem with the browser supplier itself.

 

Limetech selled me UNRAID OS and I expect that it works. They are also my first level support. If they have not a solution they should run to the kernel development team and not me. 

 

It´s of course much more evil than the problem itself.

 

 

 

For sure a few watts more arent that bad at all... The main problem with Realtek is Realtek itself. If they dont provide working drivers for Linux it wont work as expected. So nothing to change.

Since a decade of time its always suggested to avoid Realtek with Linux...

21 minutes ago, RiDDiX said:

Since a decade of time its always suggested to avoid Realtek with Linux..

Don´t missunderstand me. I avoided Realtek Network Chipsets like hell, as I had some negative experiences with other boards (many years ago) which I intended to use with Unraid. I have now 4 Unraid Systems, 2 of them have an Intel Network chipset, the 2 other ones Realtek. And Realtek worked correctly the approx. last 3 years.

 

As Realtek is used in the majority of the smaller boards you can not really avoid them anymore...

Edited December 14, 20232 yr by MPC561

1 hour ago, MPC561 said:

Depends on your mainboard and the used extension cards, as I told nearly all MB´s with RTL network chipset are affected. With the N100DC-ITX it´s not a lot (6.12.4 I had 7W mean consumption, with 6.12.5 I had 9W) so 2W. Other systems may have 5W+ difference. With the energy prices we have in Europe in some countries we discuss here over 5 years of additional energy cost of approx 100 euro in addition. With my system here in Germany and my energy price with 2W more over 5 years it´s only 29 Euro.

 

Surely not. The correct example would be: Dell sells me a laptop with an OS with hard integrated Browser from another company which stops to work after an update. In this OS I have no admin rights and also BIOS is locked and so it´s not possible to  replace the browser/install a new one. Dell sends me to the browser supplier instead of solving the problem with the browser supplier itself.

 

Limetech selled me UNRAID OS and I expect that it works. They are also my first level support. If they have not a solution they should run to the kernel development team and not me. 

 

It´s of course much more evil than the problem itself.

 

 

 

Yeah, I get that the power consumption will be different based off hardware and that could effect things quite a bit.  I still have a hard time getting fussed over that amount of extra power draw.  That amount of money over that amount of time is hard for me to care about.  I have spent far more money in a far shorter amount of time on far stupider things...

 

As for the analogy... Limetech does not hard integrate the Kernel, they have to include it for the machine to run but they do not control it, code it, etc.  If Dell installs Windows OS on your machine and a Windows Update breaks something do you go to MS, Dell, the webcam driver creators website?

 

I am sure Limetech, and a lot of other individual people, have reached out to/created tickets for this issue.

 

 

1 hour ago, RiDDiX said:

For sure a few watts more arent that bad at all... The main problem with Realtek is Realtek itself. If they dont provide working drivers for Linux it wont work as expected. So nothing to change.

Since a decade of time its always suggested to avoid Realtek with Linux...

Truth

4 hours ago, prostuff1 said:

I am sure that this bug has hit more than just the unRAID community and the Kernel devs will be looking at it

We don't know how or where to report this, but I guess unRAID programmers know where this kind of issue is being talked about if it exists, or report if it's not present. How could we find it about?

 

4 hours ago, prostuff1 said:

Going to ask a stupid question but what is the actual power difference between 6.12.4 and 6.12.6 where this kernel bug is?

few watts compared to high consumer nas. much for what alternatives could be: cloud private nas. much in percentage terms, above all if we read the conclusions of the COP28. middle terms of wife approval factor depending on the system XD.

Jokes beside,  I don't see it as a number, but as a quality and personal preferences. For example I prefered an fanless hdd-less nas, to reduce power and noise.

I mean, I see this and I thought, yeah UNRAID is worried about environment and costs:

 

I've read someone saying how minimum 2.5 watts are  (and much more in other kind of motherboards) - even being 25% for n100dc-itx , but at the same time, I can read:

If you decide to spin down, your hard drive power consumption can go down from 3-8 watts idle on average to 0.4 watts (per hard drive) so depending on how large your Array is, you might see large power efficiency gains when spinning down.

I've seen this video as UNRAID advice,

 and is amazing isn't it? How much worried he is about C-States.

And UNRAID itself recommends using C-States and reach C8 states.
"Generally, if your server can reach C8 idle, it will run the most efficiently."

 

 

Huge banner here too.

 

UNRAID has teached me a lot. Now is a bit bitter this bug. Giving it 0 worries after doing all this ... power consumption talking... and c-states talking, is something I did not expect. You got me as a client after reading it, and see some videos of "Spaceinvader One "

 

 

Edited December 14, 20232 yr by vmasip
bad preview

Has anyone else got a system hang up when disabling turbo now in 6.12.6?

 

There is also the power off issue where it hangs at the diagnostics screen. It has successfully unmounted all drives etc but you still have to manually hold the power button for the actual power off. No parity check on reboot.

 

If more people have issues maybe downgrade to 6.12.4?

Edited December 14, 20232 yr by dopeytree

1 hour ago, vmasip said:

We don't know how or where to report this, but I guess unRAID programmers know where this kind of issue is being talked about if it exists, or report if it's not present. How could we find it about?

The place to report / discuss this in the unRAID forums is here, at least i am pretty sure that is the same issue. As for reporting the issue upstream i havent a clue where to go. 

 

Same issue here with another user (in German, but the pictures say enough):

 

https://forums.unraid.net/topic/148272-butcha89s-erster-unraid-server/?do=findComment&comment=1338109

 

The user also did not understand why he only reached C3. I asked him to do:

sudo lspci -vv | awk '/ASPM/{print $0}' RS= | grep --color -P '(^[a-z0-9:.]+|ASPM |Disabled;|Enabled;)'

 

And the result was also that ASPM was disabled. In this case with an RTL8125. (Which supports my fear that now nearly all Boards with Realtek Chipsets are affected)

 

39 minutes ago, primeval_god said:

As for reporting the issue upstream i havent a clue where to go. 

 

Me also not. A German moderator said me I should go to the kernel GitHub something. Made me a bit pissed... Sorry.

 

 

Greets,

Joerg

Edited December 14, 20232 yr by MPC561

1 hour ago, dopeytree said:

Has anyone else got a system hang up when disabling turbo now in 6.12.6?

 

There is also the power off issue where it hangs at the diagnostics screen. It has successfully unmounted all drives etc but you still have to manually hold the power button for the actual power off. No parity check on reboot.

I have three servers running 6.12.6.  My main server has turbo boost disabled via the tips and tweaks plugin.  I have seen none of the problems with system hangs or power off issues that you describe.

1 hour ago, Hoopster said:

I have three servers running 6.12.6.  My main server has turbo boost disabled via the tips and tweaks plugin.  I have seen none of the problems with system hangs or power off issues that you describe.

 

Are you toggling turbo on / off?

Or do you mean you have had turbo turned off for a while like say since 6.12.0 etc and just updated to 6.12.6 and are having no issues. The two statements are different.

Edited December 14, 20232 yr by dopeytree

2 minutes ago, dopeytree said:

Are you toggling turbo on / off?

It is permanently off via Tips and Tweaks.

 

 

No on/off schedule.  With previous versions of Tips and Tweaks it was also off.  I had server crashing issues with certain processes with Turbo Boost on.  Has been off through several versions of Unraid.

 

14 hours ago, MPC561 said:

A German moderator said me I should go to the kernel GitHub something. Made me a bit pissed... Sorry.

that is another huge part of the problem... more that you (and other user) wouldn't report it to the right, but get pissed...

next time switch it up and try the following:

report it to the right place and stop getting pissed. could help everyone.

and please stop repeating your opinion... my eyes are bleeding and you aren't right.

 

Edited December 15, 20232 yr by sonic6

3 minutes ago, sonic6 said:

that another hughe part of the problem... more that you (and other user) wouldn't report it to the right, but get pissed...

No, simply surely not.

 

I say it again that's not my job to get Unraid running correctly. I paid for a product and should run to Kernel Development teams to report bugs and wasting my time? I´m to polite to write what I think about this nonsense you wrote...

 

PS: Please stop repeating your opinion, you aren´t right.

Edited December 15, 20232 yr by MPC561

okay, just a stupid question... you buy a volkswagen and one of the tires a broken...

do you contact volkswagen? i think so.

will volkswagen replace the tire? maybe.

do they have so... not really. they sell the tire as a part of the cars, but when the tire itself broken by developement volkswagen will say: GTFO and tell Michelin!

 

and now i am out.

Edited December 15, 20232 yr by sonic6

35 minutes ago, sonic6 said:

okay, just a stupid question... you buy a volkswagen and one of the tires a broken...

do you contact volkswagen? i think so.

will volkswagen replace the tire? maybe.

do they have so... not really. they sell the tire as a part of the cars, but when the tire itself broken by developement volkswagen will say: GTFO and tell Michelin!

I´m polite and answer your question. If I bought my Skoda SUV (VAG group) and a tire which is on the car is broken (Air goes out) and I detect it several hours later (so the defect was already present when I got my car) I go back to Skoda to replace it as they selled the tires with the car to me. Then they can contact Continental or Michelin and they surely have to replace it. 

 

And your example again is bullshit, the Kernel is more comparable to the engine or a part connected to the engine. The correct example: I buy a Skoda with Diesel/Nafta engine and an Injector/Rail/VCV/AGR/turbo/pump (which was supplied to Skoda from Bosch/Delphi/Conti) has a defect, then I run of course to Skoda and not to Conti.

 

Btw.to clarify something. In my, very polite, first post here I showed the problem and politely said: "Hope for a fix". This does not mean I expect immediately a fix. I´m also working in SW development (btw. Automotive industry) and know this may take a time, especially when it´s a kernel topic.

 

My problem are people which tell me, that me should report and discuss with a kernel team...

 

Edited December 15, 20232 yr by MPC561

1 minute ago, MPC561 said:

My problems are people which tell me that me should report and discuss with a kernel team...

the main reason why people are telling this to you the following:

the more people are reporting the, the higher is the chance that this issue will be solved.

you are in the same boat with other users. if you all report it at the right place, then the chance that this will be fixed, it muuuuuch higher then it will only reported by a few unraid devs.

3 minutes ago, sonic6 said:

the more people are reporting the, the higher is the chance that this issue will be solved.

Mensch ich weiss das auch. Ich Supporte selber hier in dem Forum und in anderen zur genüge, und das weisst Du auch ganz genau. Ich hab nicht mehr die Nerven jetzt auch noch zu einem Kernelteam zu rennen und da Druck zu machen. Und ich WILL Limetech da auch nicht raus lassen. Wenn support für Produkte immer schlechter wird muss ich vielleicht wirklich in 5 Jahren zu Bosch rennen wenn mein AGR im Skoda ein Serienproblem hat und nicht zu Skoda. 

 

Das mit dem Limetch nicht rauslassen ist genauso eine Ideologie wie die deine mehr Druck aufs Kernelteam durch Reports zu machen.

 

Aber jetzt besser wirklich (halbwegs friedlich Schluss).  Manchmal muss man einfach einsehen das man einfach unterschiedliche Meinungen hat und sich nicht einigen kann.

From what I can see ASPM for some Realtek NICs was disabled on purpose by the kernel maintainers, because being enabled was causing issues with the NICs stopping to work for some users, if that's true unlikely they are going to re-enable it soon, a little more power usage is better than losing the network, more info below:

 

https://lists.ubuntu.com/archives/kernel-team/2023-September/142666.html

 

Bug report that resulted in the change:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2031537

 

 

Thanks @JorgeB. As usual a very good answer from you.  /hug

I had to drop back to v6.12.4 today. Things had been relatively stable at first on v6.12.6 however the last two day's I've had two freezes that had to do with my array. 

Testing the two drives that had the issues they had no errors etc after extended tests. So I'm downgrading and I'll report back if it sorts my issue. 

Looks like what ever happened fully dropped my valid parity. I had to unassign it, start, turn off, and then re-assign it now. It is building the parity again. 

jlw-unraid-diagnostics-20231218-1813.zip

Edited December 18, 20232 yr by jlw_4049

On 12/15/2023 at 1:26 PM, JorgeB said:

From what I can see ASPM for some Realtek NICs was disabled on purpose by the kernel maintainers, because being enabled was causing issues with the NICs stopping to work for some users, if that's true unlikely they are going to re-enable it soon, a little more power usage is better than losing the network, more info below:

 

https://lists.ubuntu.com/archives/kernel-team/2023-September/142666.html

 

Bug report that resulted in the change:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2031537

 

 

Great digging work. I think is fixable. I have read more than once that:
Modification are isolated to the r8169 driver

 

So maybe some of us could be lucky at the end.

 

I've also seen this code at this patch:

if (rtl_aspm_is_safe(tp))

substituting one if(enabled) , which give me hopes. But I don't even know if this patch arrived until release.

So, at the end of the second link, someone comments about the issue, but is redirected to create a new bug. He creates this:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2017158

 

would we have to report here? maybe other threads where patches were being discussed.

Please, advices here.

 

Edited December 19, 20232 yr by vmasip

12 hours ago, jlw_4049 said:

I had to drop back to v6.12.4 today. Things had been relatively stable at first on v6.12.6 however the last two day's I've had two freezes that had to do with my array. 

The regular syslog starts over after every boot, if you can try again enable the mirror to flash drive option in the syslog server and post new diags after the problem.

2 hours ago, JorgeB said:

The regular syslog starts over after every boot, if you can try again enable the mirror to flash drive option in the syslog server and post new diags after the problem.

Thanks for the response. If it has that issue again I'll enable that and post proper diags. However, since down grading it's been as stable as it was before without 100% CPU use for no reason. 

I am pretty sure I have a card that is unsupported by the linux kernel in the latest update, although I didn't go into great details to check it out. I just down graded the OS, so far smooth as silk. 

1 hour ago, jlw_4049 said:

I am pretty sure I have a card that is unsupported by the linux kernel in the latest update,

Not from what I can see in the diags, only see an AMD onboard controller and an LSI HBA, both should work fine.