Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I guess these are the kinds of things I think about on the weekend...

When I was growing up (not that long ago) my parents always taught us to wait 30 seconds after shutting down the computer before turning it back on again.

Fast forward to today in professional IT, and I know a good number of people that still do the same.

Where did the "30 second" rule come from? Has anyone out there actually caused damage to a machine by powering it off and on within a few seconds?

share|improve this question
3  
30 seems a bit excessive - I've only heard it as 10 seconds. –  Dentrasi Jun 28 '09 at 15:53

12 Answers 12

up vote 41 down vote accepted

You want all capacitors to discharge. A poorly designed/constructed device could be damaged. But the more likely issue is that since you're power-cycling it to reset an unexpected/unhandled failure, a capacitor not being discharged could leave the system/circuit/device not fully reset.

On computers I tell people to wait for all fans to stop spinning. It's a fair compromise. This 30-second advice is much more relevant to a non-computer (simpler, bigger capacitors) device. We know that the complex parts of a computer will be reset upon power cycling them, regardless of any random capacitors.

I've certainly power-cycled things quickly, had it not work, then waited a significant time with it off, and had it work. No evidence if this mattered of course.

share|improve this answer
11  
"Be kind to your power supplies. You'll miss them when they're gone." - bofhcam.org/pfy/leatherman.txt –  pgs Jun 28 '09 at 3:40
3  
As an additional suggestion, if you want to be really sure, pull out the power plug, then push the power button on the front of the computer. –  Alex Jurkiewicz Jun 29 '09 at 8:14
3  
That reminds me of a great phone support idea. Never say "re-seat the network cable". A lot of people will see the obviously well-seated cable and refuse to. Say to switch the ends of it. This -requires- it be re-seated, and few people will question why. –  carlito Jun 29 '09 at 17:49

I am not sure about current hardware but older generations of DRAM relied on capacitors that took up to 5 to 10 seconds to completely discharge. also some early switching power supplies could be damaged by inrush current if they had not completely discharged. Depending on the power supply again up to 15 seconds was needed. SO why tell everyone 30 seconds? Because all of you are impatient and don't follow instructions.

share|improve this answer
2  
Tell them 30 seconds so that they will wait at least the required 10 seconds. :-) –  Les Jun 29 '09 at 18:02

When I have phoned support for a broadband connection I've been asked to reboot my router and wait 10 or 30 seconds (I forget how many) to make sure that the modem(s) I was connecting to in the telephone exchange had properly disconnected my router.

That's the only time I've believed there might be something in waiting. I mean Microsoft doesn't wait any length of time when I hit the Restart option...

share|improve this answer

This is sometimes in the manual for computers. I remember reading it in either a Sun hardware guide or SGI hardware guide. I think it was 10 seconds though.

So it's not just made up by computer illiterates.

share|improve this answer

The IBM PC and PC/XT had this annoying feature. If you turned the power off and then on too quickly, nothing would happen. This was even more frustrating for people since there wasn't a power or on/off light to look at... you know you did it wrong because after a minute or so the computer hadn't done anything. Hence the common 30 second rule for rebooting.

share|improve this answer

I was a tech support rep for Gateway 2000 back when that name still seemed futuristic. They had a two week training course for tech support reps back then, which was pretty comprehensive. I remember the instructor saying that among the reasons above, another reason to let a computer sit for 30 seconds had something to do with viruses that could push themselves in to VRAM and then be activated after a quick reboot where a full power off and 5 second wait would make sure the VRAM was clear.

share|improve this answer

Probably good to have time to let the wetware interface between keyboard and chair cool down as well, as a swift restart is often a result of some irritating failure :) (See also: guru meditation)

share|improve this answer

Other thing to consider: Back in the old days, most microprocessors didn't have any way to tell if they had the right voltages on them as they ran. If the +5V rail went from 0 to +4.50, that was often good enough for the transistors in the chip to start conducting.

Often, the CPU would be "half-on" with the +5V rail OK, but the +12V rail at, say, 10V. This had unpredictable results.

Many CPUs today have supervisory circuits that won't let them run unless everything's good on the power rails. The Power Good signal that PC motherboards put out is one example of such a mechanism.

One still sees portable devices that flake out and only half-function on weak batteries. There you really do have to pop the battery out for a few minutes to get it going again. I have a radio with a CPU that is famous for locking up and sometimes won't play again unless the batteries are removed and the reset button hole is depressed.

share|improve this answer

I think it comes from older hardware that didn't handle rapid power cycling well. I had two old commadore 64 computers that died because I reset them by flicking the power off then on again too quickly.

share|improve this answer

As far as I remember when I was told when I started to work on computer it was so the Hard Drive would stop spinning after the shut down. Then you would restart and the hard drive would spin normally as opposed to start up when it was still spinning from the shut off.

Who knows if this makes a difference or not, but I still wait 30 seconds.

share|improve this answer

Yes as the other posts explained this comes from the time that the state of a capacitor could influence the outcome of the expected result. However these days it is either broken or not and only in small amount of times something is fixed by waiting.

The most logical explanation is that heat dissipates which allows a previous loose connection to reconnect again until there another external source (like movement which forces it loose again) occurs.

The other way around works too, sometimes by vibration it connects again and by heat (due to electrical resistance) disconnects.

That is why waiting 30 seconds/till the machine is cold/till all light are off work in certain cases. This also the reason why sometimes it actually works if you kick or hit a PC/TV or turn it upside down/certain angle.

share|improve this answer

Shut the computer down, take the side cover off, and look inside to see if there are LED's on the motherboard that stay on when there's standby power. Now pull the plug and watch how long it takes for those lights to go out.

This is due to those capacitors carlito is talking about in another answer.

share|improve this answer
1  
The onboard NICS in most newer motherboards will always be lit when the board is plugged in, due to wake-on-lan support. An unlit NIC is a good indication that there's no power at all. Good to know when you swap parts. –  dmoisan Jun 28 '09 at 15:39
1  
You could also just check the light that is on some power supplies. –  Brad Gilbert Jun 28 '09 at 19:18

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.