## Disabling upowerd on a Desktop Linux computer that never actually had a battery.

One of the facts which I’d like to make sure the reader knows, is what, exactly, the process ‘upowerd’ is responsible for on any Linux computer, lest the reader disable something that he or she may really need.

Many computers, including laptops, have batteries. ‘upowerd’ is the Linux process that monitors the battery, or the batteries, in case there is more than one to monitor. If the computer has batteries to monitor, then this process is essential and should not be disabled. But, in certain specific other cases, this process can become a problem, as it recently was on my desktop computer named ‘Phosphene’, which is running Debian 9 / Stretch, which has Plasma 5.8 as its desktop manager, and which will only ever run on A/C power, due to the very way in which its hardware is designed.

Several months ago, I noticed that the ‘North Bridge’ of this computer – an essential chip on its motherboard – was becoming rather hot. And the motherboard in question, the ‘Sabertooth X58′, manufactured in the year 2011, was already famous for the design flaw that causes its North Bridge to become 76.5⁰C hot when idle, which is actually hotter than the GPU on that computer will normally become, when pressed to work hard. Several Months ago, the North Bridge temperature when idle was 80⁰C or even 81⁰C! Whether it’s actually safe for a chip to be that hot is a subject open to opinion. This temperature will not cause an immediate failure, but if the chip is so hot continuously, then its lifespan will become shorter, than what the lifespan of chips is supposed to be, in general.

Therefore, months ago, when I first noticed this, I opened up the tower / desktop computer and examined it, to look for failed cooling fans, etc. There was no such cause for the elevated NB temperature. Not being able to remedy the problem, I just left it that way.

But only yesterday, I noticed that the process ‘upowerd’ was consuming an inordinate amount of CPU time. On the octa-core machine, that process was consuming maybe 1% of available CPU time, which was quite a lot, considering that there should have been nothing for it to do. And, never having noticed this before, it seemed possible to me that ‘upowerd’ may have been consuming 1% of available CPU time (unnoticed), since months ago.

When ‘upowerd’ misbehaves in this way, sometimes this happens, because of hardware signals, such as perhaps, a battery which continuously disconnects and reconnects, etc… Therefore, before a software kludge is attempted, all possibilities need to be followed, to find hardware causes for this behaviour, to which ‘upowerd’ would simply be responding in a normal fashion. Yet, even given hardware reasons to be active, ‘upowerd’ should not be consuming much more than 1% of available CPU time, in case the reader has some situation where this process is consuming, say, 10% of the CPU.

## I have reinstalled the O/S on an existing computer.

According to an earlier posting of mine, the computer which I named ‘Plato’ was experiencing technical issues. Its power-supply was dead.

Technically, I succeeded in replacing the power supply. But unfortunately it also turned out, that its main hard-drive was dead. Therefore, it is now using a different hard-drive as well, which means that I needed to reinstall the O/S. I installed the latest build of Kanotix, and all is well again with that computer.

That computer has now been renamed ‘Phosphene’.

I suppose that one question which now remains unanswered, is whether I should switch ‘Phosphene’ to the proprietary NVIDIA drivers, as I had done with ‘Plato’, or whether I should keep it with the open-source ‘Mesa’ drivers, that include the ‘Nouveau’ drivers.

## My computer Plato is having a technical issue.

One of the main computers which I’ve been using, that is named ‘Plato’, that was running Debian / Stretch, has experienced a major technical problem. When I got home this afternoon, I found it was not running. And, when I pushed the power button, it did not turn on.

A basic, automatic idea which would pop into people’s heads is, ‘The power-supply burned out.’ If the only task which lies ahead really was, to replace the power supply, I’d have it easy. This is a tower-computer from the year 2011, with a Sabertooth X58 motherboard.

• The correct power-supplies for this old MB may have become hard to find,
• Even if I had a replacement power-supply, it would be very cumbersome to replace because the harnesses of the present one loops behind too many recessed compartments, within the case.

The only thing I’ve done so far, is to perform a diagnostic test. I disconnected all the jacks between the power-supply and the MB, and retried the power button. My purpose behind that was, the idea that modern power supplies will refuse to turn on, if they sense a short-circuit between their load, and ground. Thus, if the power supply had been able to resume, with the MB disconnected, I’d know it was the MB, and I’d also know there’s no point in replacing the power-supply. But thankfully, the power-supply also did not power up like that. So I reconnected the power-supply to the MB.

So as it stands, I don’t know the best way to proceed, but am without the use of that trusty computer for now.

(Update 2/7/2019, 14h15 : )

One reason this apparent loss is unfortunate is the fact that, being my only Debian / Stretch computer, that computer was also the only one, which had “SageMath” installed and working on it. So my available Computer Algebra Systems are reduced to “Maxima” and “Yacas” for now.

(Update 2/9/2019, 18h50 : )

Actually, I’ve learned that my so-called diagnostic test was pointless. The power button these days, does not have a direct connection to the power-supply, to signal that the power-supply should turn on. The power button has its connection to the M.B., which tells the power-supply to turn on. Therefore, with the M.B. disconnected from the power-supply, there was no way for the power-supply even to get the signal, to turn on.

A personal friend of mine has lent me a power-supply tester, so that I’ll next be able to test that more properly. And, hoping that it is just the power-supply which is faulty, I’ll look into replacing it.

(As of 2/7/2019, 14h15 … )

## Weak Power-Supply

The current computer ‘Phoenix‘ (I own several computers) has suffered from a string of malfunctions in the past, which I had trouble diagnosing the cause of.

I think I’ve found the problem: Its power-supply is weak in some way, which can also lead to low-voltage conditions that it subjects the high-speed electronics to. When high-speed logic circuits are fed low supply-voltages, the computer can spontaneously crash.

The reader may wonder how I know this.

When I’ve left the computer idling and come back to it, thus entering my password to dismiss the screen-saver, the case-fan speed seems to be stable around 3,500 RPM. But as soon as I fire up my Web-browser, the CPU usage goes from low-usage to nearly-100% usage for less than a minute, and as soon as that happens, the case-fan speed becomes unstable, sometimes resulting in a reading of ~40 RPM, which means that ‘The fan has stopped spinning.’ Then, as soon as I allow CPU usage to go below 5% again, the case-fan speed sometimes stabilizes again, within the same sitting.

Well there is no valid logic by which the motherboard would signal for the fan to stop spinning, or to slow down, at the moment the CPU usage is high. And so the only other explanation I can think of, is that the CPU – and possibly other circuits in the box – are drawing more current, and that this is causing a temporary dip in supply-voltage, just enough for the recently-installed fan to stop spinning.

But then, such a weakness also makes this computer more susceptible to such phenomena as brown-outs. Even though my eyes can see power-fluctuations that take place within a fraction of a second, I cannot see a low-voltage condition in the A/C power we are fed, if that low-voltage condition has set in over a period of minutes.

I might start looking for a new power-supply for this old box, rather than a new case-fan.