Ubuntu 12.04s Great Flaw: No Hibernation Feature

So, you don't think the argument that “it was found to be unreliable, very slow and confusing to have two suspend modes.”

Well, I'm vot too keen on the "too confusing" argument. myself. But I have family members who would agree. The unreliability part of it sounds pretty much in line with what I've been hearing.

The author of the article said clearly that it's been reliable on his hardware. I've had exactly the same experience. Might it be unreliable on some hardware? Sure. I haven't seen it. What has been unreliable on my hardware is Ubuntu in general. I have better results with other distros in terms of hardware support.

Quoting:Might it be unreliable on some hardware?

Yes, absolutely, I've experienced the problem myself with both Ubuntu and [of course] Mint. Turning it off by default is fine with me, making it difficult to locate and turn back on isn't, leaving it as a turned off option right next to Suspend is my preference.

@tracyanne

That makes a lot of sense.

But from a marketing point of view it probably scream "Danger! Danger!". It basically tells (or at least reminds) the user that there is a common, long standard feature, that may not work properly, which the competition claims to deliver reliably.

And let's face it, many people will believe that claim even in face of personal contrary experience, because of years of [strike]consumer indoctrination[/strike] advertizing exposure. Add this to the simple fact that OEMs at least go to the effort of ensuring that their laptops do (at least when new/as shipped) suspend and hibernate properly at least 4 times out of five, under Windows -- regardless of how funky, non-standard or idiosyncratic there 2-cents cheaper parts may be. Finally, even if Linux performs (hibernates and suspends) just as well as Windows, its failures to resume properly will be judged more harshly than for Windows.

I don't like the decision, either. But I strongly suspect that If I worked at Canonical, I would have supported it nonetheless.

HP and Toshiba, in my experience, have done an excellent job with hibernation on their hardware. 4 time out of 5? More like 99 out of 100. Granted, I have only owned a small subset of their hardware, but I keep coming back for a lot of reasons with reliability at the top of the list.

Just out of interest does anyone know whether hibernate as opposed to suspend draws a lot less power? I just tried hibernate on a Thinkpad and it is actually faster in recovery than suspend...

Edit: I was fooled by Mint. Hibernate is longer but not too bad. Looks like it draws virtually no power at all while suspend has a half moon light on suggesting a small power usage.

HIbernate should take a bit longer. I don't know an accurate way to measure power consumption in a suspended or hibernated state but I'm under the impression that power draw during hibernate is nearly nil.

I'm also not surprised that a Thinkpad works well. I haven't used one since IBM sold them off to Lenovo but back in the day they were always very decent machines.

Measuring power shouldn't be hard; fully charge the notebook's battery (or maybe even just take it out altogether), and plug it into a $20 "Kill-A-Watt" power meter. It'll show the instantaneous power draw.

Power draw during Hibernate should be exactly zero, as the computer will be completely powered down. During Suspend it should draw a very small amount of power, just enough to keep RAM active.

It won't be zero as long as you have a power adapter plugged in and you're measuring with that, as the adapter itself consumes power even if the notebook isn't. Further, it's likely there's some circuits there drawing some small amount of power even when "off" or in hibernate. After all, the power button works, even though it's not a mechanical on-off switch; soft power buttons have circuitry that always draws power, and of course that circuitry is powered by power supplies that are not 100% efficient at converting the incoming 20VDC to their low voltage. No modern computer is ever "completely powered down", unless it has the battery removed and is unplugged. However, the $20 power meters might not have enough resolution to detect such a low power draw.

What makes you think the power switch isn't mechanical? I think it is on all my systems. It's a slide switch on the HP and a push button on the Toshiba and on my desktop.

The switch is "instantaneous", in that you press it and it closes, you release it and it opens. But opening the switch doesn't cut power to the entire computer, like in the old PC, XT, and AT days. Cutting power is a function of software now, as well. Which means that starting power can also be a function of software.

How else does the BIOS timed power-up work? How would the ability to turn on the system via a password on the USB keyboard work? What about the wake-up-via-LAN? These can all be monitored via BIOS ACPI, the same as the power button. And these require power, from somewhere, to run the CPU and necessary devices, as well as their controllers, and power the ROM with the necessary BIOS code.

The switch itself is mechanical, but that mechanical nature is only incidental to controlling the power on a modern PC.

To get really technical, the switch is an SPST momentary off-(on) switch. You can't control power with that. That's why pressing the power button doesn't turn off the computer immediately, you have to hold the switch for a few seconds to force a power-off, and why on some systems pressing it once will put it into a suspend state.

The ATX standard back in the mid-90s made on-off power switches impossible on all PCs, as they had to have soft power buttons as required by the standard, so that the features gus3 mentions could be possble (wake-on-LAN, etc.). Notebooks are derived from the ATX standard (obviously, they jettison the physical form factor parts, but they mostly keep the power supply parts).

Quoting:on some systems pressing it once will put it into a suspend state.

That's a function of software. The running system merely registers the power-button press as another input event, dispatched to an ACPI handler. It could result in a suspend, a shutdown, a hibernate, some other crazy action, or nothing. There's nothing inherent in the switch that causes a suspension.

And if I'm having trouble with Xorg, and my keyboard and mouse become non-responsive, I'm mighty glad I've configured the power button to "telinit 3". It's a good way to kill the GUI quickly, with the only functional input I have, without shutting down other services.

@gus3: telinit changes the runlevel. You really don't have to do that. Just hit CTRL-ALT-F<#> to switch to a virtual terminal and kill your X session. If you are at runlevel for a GUI switching back to your original desktop should bring you to a display manager screen and you can then login to a new X session

Ctrl-Alt-Fn?

I said, "keyboard is non-responsive in Xorg." Keypresses do nothing. Ctrl-Alt-Cuss does nothing. Not even Caps Lock and Num Lock. The power button is the ONLY event-generating user input I have.

OK, it's late and I completely misread what you wrote. Yes, your solution makes sense for that instance.

My own experience is that when X locks up hard I can often still get a response from a key sequence. You are describing not a lock up but rather a failure to recognize the keyboard. I do understand now. Disregard what I wrote.

> And if I'm having trouble with Xorg, and my keyboard and mouse become non-responsive,

Are you using the NVidia driver? I had similar problems with it on my hardware.

no, it was the keyboard and mouse drivers not being updated with the rest of the system.

Thanks, gus. I was just curious if it was the same problem I had experienced or not, since I'd never heard anyone else reporting it. Obviously not.