Rebooting: Power cycle vs. switch

[royarcher]

I have read often in this and another forum about unplugging your router from the wall socket for a few minutes then booting it up again.What are the benefits in doing a reboot this way compared to just through the switch either on the back of the router or by logging into it and hitting the reboot switch?

 

[dosborne]

If something goes screwy in memory or the firmware, then a scheduled reboot or a reboot through the GUI may not actually work, as it is all just software.

In most cases, there is no difference between using the physical switch (not a soft switch), unplugging the power connector from the router or unplugging the wall-wart from the outlet. However, in rare cases, the hardware can become "locked" in a bad state and a power-down using one of these options is required. Physically removing the power to the hardware allows things like capacitors and other components to fully reset and forces hung processes to reset.

 

[royarcher]

Ok that makes sense thanks quriosity fed

 

[FarmerBob]

I just put my ONT and Router a dual outlet digital timer. I was under the impression that it was a dual control unit that I could have had each unit operated individually. The description was written poorly.

So every Sunday morning at 6am both go off, then at 6:15 both go back on.

I know that you are supposed to do each individually, Router after ONT, and then visa versa. But I trust that this "should" work since I have done it a billion times. A pair of these were under 10 bucks and can be programmed to the moon and back. So.....

Sent from my SM-T380 using Tapatalk

 

[umarmung]

Unless you need to power cycle electronic equipment, this is not something you should intend to do. Power cycling results in thermal stress which reduces the longevity and reliability of electrical devices over time.

On the other hand, it is entirely possible you do not care or are satisfied about the trade offs within the expected lifecycle of that device ...

 

[dosborne]

Unless you need to power cycle electronic equipment, this is not something you should intend to do. Power cycling results in thermal stress which reduces the longevity and reliability of electrical devices over time.

For as long as I can remember running electronic equipment, at least since the '80s, I've had my modem (or current equivalent) connected to a digital timer (one from Radio Shack when they still existed in Canada) that would power the device off for 1 minute at the preset time. It is only recently (within the last year) that I took the timer out of operation as I use the scheduler in the router to reboot daily. "Thermal stress" from powercycling anything made in this century is a joke.

 

[ColinTaylor]

"Thermal stress" from powercycling anything made in this century is a joke.

I'm sure all those people that suffered from the PS3's yellow light of death will take great comfort in knowing they were imagining it.

https://en.wikipedia.org/wiki/Solder_fatigue

 

[dosborne]

There are of course some exceptions due to poor design and / or quality control, but I stand by my statement as a general rule. If a device cannot withstand being turned on / off once a day, is it worth having?

(In my opinion the answer is "no")

Don't get me wrong, I'm not saying you should *have* to reboot something every day, just that it should not matter in a modern electronic sense. I take pride in how long I have various systems running without being rebooted. Some have literally been in operation for over a year.

 

[Ken Atkinson]

There are of course some exceptions due to poor design and / or quality control, but I stand by my statement as a general rule. If a device cannot withstand being turned on / off once a day, is it worth having?

(In my opinion the answer is "no")

Don't get me wrong, I'm not saying you should *have* to reboot something every day, just that it should not matter in a modern electronic sense. I take pride in how long I have various systems running without being rebooted. Some have literally been in operation for over a year.

Providing no firmware updates available, no issue with device not been rebooted for that time as long as its running fine, if updates haven't been applied that as bad as booting every day for the sake of it IMHO


Sent from my ONEPLUS A6013 using Tapatalk

 

[RMerlin]

"Thermal stress" from powercycling anything made in this century is a joke.

Depends on the device. The actual issue isn't thermal stress, but rush current. Ever noticed how a traditional light bulb almost always fails right as you turn it on? That's because of the rush current at power on.

The issue is less prevalent with modern electronic, but it's still there for anything that can draw a significant amount of current at power on. Hard disks come to mind - in addition to the strain of their motors having to spin from at rest up to 7200-15000 rpms. Large arrays use scattered spin-up to at least reduce the amount of current drawn from the power supply.

 

[dosborne]

Yes hard drives, compressors and a bunch of other stuff.

Generally speaking though, I stand by my statement. A router for instance (not powering stuff from the USB), a radio, a cell phone, a tablet, a remote control.

A non-mechanical, non-motorized basic electronic device. You guys happy now? LOL

I've been turning devices off every day via timer for 30+ years and have yet to have an issue. YMMV

 

[System Error Message]

Sorry guys but most of you got it wrong.
Power cycling for a few minutes allows the capacitors to drain. Not only the capacitors that are actual capacitors, but one of the problems in board design is capacitance between conductors/circuits, which can cause signals to be all wrong and cause it to freeze. The 2 things that cause it to freeze, 1) wrong voltage that the circuit is unable to modulate to the right voltage, 2) capacitance that messes with signal at high rates.

!st is simple, it can be any component relating to power which is switched such as the resistors overheating that it changes the resistor value or at some point you get capacitance or a bad input voltage that the system cannot account for and so outputs the wrong voltage. In a high rate switching circuit like the CPU, it is very sensitive to small changes in power.

2nd is the capacitance which buffers the signal causing the circuit to be unable to receive signals at a high rate, causing it to freeze as the data/instructions are new jumbled. Having an unearthed metal case makes this problem even worse, and dlink is notorious for this. Measuring dlink with an EM mater i measure 400+ V around the case which is very bad. Such a problem has the chance to fry electronics as well. Switching power circuits to change voltage dont have to operate at high volts, and even if you do, it is necessary to observe the spacing and insulation required so that it does not mess with wifi and in general cause capacitance and with the potential to fry components not rated for the voltage.

This is why some routers have to be power cycled for a few minutes. On a PC the same thing can happen, and this is where leaving it plugged in (but mains turned off with ground connected) can help by pressing the button for a minute to drain the capacitors, which devoids it of power completely allowing things like the bios or any controller RAM to clear when you next boot it.

Some well designed devices, especially those targeted towards enterprise dont need to be rebooted and will run fine for years.

Any device with leaky caps dont work well for power cycling if thats what you're talking about. Power cycling does no harm to devices that do not use leaky caps or anything that wears out from big change. Hard drives that start and stop and make your table shake when it does dont last long from power cycling because of their poor design to dampen any big change, this includes how fast they accelerate the disk to the right speed. I have an old WD green still working fine, because i modified the firmware to disable parking, but my friends same WD green stopped working because he never modified the firmware.

Edit: to be particularly clear about answering the thread, im talking about using the mains which is better than the power switch on the device. No one ever talks about using the mains so usually they'll say unplug because not every mains has switches.

 

[sfx2000]

I have read often in this and another forum about unplugging your router from the wall socket for a few minutes then booting it up again.What are the benefits in doing a reboot this way compared to just through the switch either on the back of the router or by logging into it and hitting the reboot switch?

Doing a reboot from SW - either via WebUI or on the shell is always preferred to pulling power - this allows open files to be properly closed out, caches flushed, file system synced, etc.

One can corrupt the file system by doing the hard reboot by either the power switch or unplugging.

 

[royarcher]

Doing a reboot from SW - either via WebUI or on the shell is always preferred to pulling power - this allows open files to be properly closed out, caches flushed, file system synced, etc.

One can corrupt the file system by doing the hard reboot by either the power switch or unplugging.

Thanks good to know

 

[sfx2000]

Depends on the device. The actual issue isn't thermal stress, but rush current. Ever noticed how a traditional light bulb almost always fails right as you turn it on? That's because of the rush current at power on.

The issue is less prevalent with modern electronic, but it's still there for anything that can draw a significant amount of current at power on. Hard disks come to mind - in addition to the strain of their motors having to spin from at rest up to 7200-15000 rpms. Large arrays use scattered spin-up to at least reduce the amount of current drawn from the power supply.

Agree - stuff left on will generally run for a very long time...

Fun story - Back a couple of jobs ago - we had a lab/dev rack full of servers and drive arrays - some of the machines had 7 years uptime (long story, it was a 'forgotten' rack from a developer that left years ago) - anyways, the data center lost power, the UPS in the rack had dead batteries, so everything went down hard. When power came back up, half that rack was dead, and the servers had to be decommissioned.

I had to remind my fellow devs to take inventory, and more importantly, do a walk-thru to ID machines that should be removed...

 

[royarcher]

Depends on the device. The actual issue isn't thermal stress, but rush current. Ever noticed how a traditional light bulb almost always fails right as you turn it on? That's because of the rush current at power on.

The issue is less prevalent with modern electronic, but it's still there for anything that can draw a significant amount of current at power on. Hard disks come to mind - in addition to the strain of their motors having to spin from at rest up to 7200-15000 rpms. Large arrays use scattered spin-up to at least reduce the amount of current drawn from the power supply.

Just rereading this post has me wondering if this is the cause of a particular problem I've noticed lately.After swapping from using a VPN to a DNS proxy server to stream US content it will behave just like you mentioned here.The connection will stutter,going from red led to no led connect again for a moment then do it all over again for about 5to 10 minutes until it settle's down Ps I always do a reset after swapping to change the IP address so I can register it,it's just a bit faster than waiting for it to change from the VPN IP but I don't reset by pulling the plug just by hitting reset on the router's homepage

 

[royarcher]

Sorry guys but most of you got it wrong.
Power cycling for a few minutes allows the capacitors to drain. Not only the capacitors that are actual capacitors, but one of the problems in board design is capacitance between conductors/circuits, which can cause signals to be all wrong and cause it to freeze. The 2 things that cause it to freeze, 1) wrong voltage that the circuit is unable to modulate to the right voltage, 2) capacitance that messes with signal at high rates.

!st is simple, it can be any component relating to power which is switched such as the resistors overheating that it changes the resistor value or at some point you get capacitance or a bad input voltage that the system cannot account for and so outputs the wrong voltage. In a high rate switching circuit like the CPU, it is very sensitive to small changes in power.

2nd is the capacitance which buffers the signal causing the circuit to be unable to receive signals at a high rate, causing it to freeze as the data/instructions are new jumbled. Having an unearthed metal case makes this problem even worse, and dlink is notorious for this. Measuring dlink with an EM mater i measure 400+ V around the case which is very bad. Such a problem has the chance to fry electronics as well. Switching power circuits to change voltage dont have to operate at high volts, and even if you do, it is necessary to observe the spacing and insulation required so that it does not mess with wifi and in general cause capacitance and with the potential to fry components not rated for the voltage.

This is why some routers have to be power cycled for a few minutes. On a PC the same thing can happen, and this is where leaving it plugged in (but mains turned off with ground connected) can help by pressing the button for a minute to drain the capacitors, which devoids it of power completely allowing things like the bios or any controller RAM to clear when you next boot it.

Some well designed devices, especially those targeted towards enterprise dont need to be rebooted and will run fine for years.

Any device with leaky caps dont work well for power cycling if thats what you're talking about. Power cycling does no harm to devices that do not use leaky caps or anything that wears out from big change. Hard drives that start and stop and make your table shake when it does dont last long from power cycling because of their poor design to dampen any big change, this includes how fast they accelerate the disk to the right speed. I have an old WD green still working fine, because i modified the firmware to disable parking, but my friends same WD green stopped working because he never modified the firmware.

Edit: to be particularly clear about answering the thread, im talking about using the mains which is better than the power switch on the device. No one ever talks about using the mains so usually they'll say unplug because not every mains has switches.

So I understand what you are saying, A.power cycling does no harm and in fact can be a good way to drain everything and restart afresh.B.the best way to do it is by pulling the plug out of the wall socket?

 

[RMerlin]

On a PC the same thing can happen, and this is where leaving it plugged in (but mains turned off with ground connected) can help by pressing the button for a minute to drain the capacitors, which devoids it of power completely allowing things like the bios or any controller RAM to clear when you next boot it.

I've fixed a couple of "dead" USB ports on PCs that way - just having the customer press the power button while the power cord is unplugged.

It's also why I often recommend doing what I call an "electrical reset" when people post about a dead switch port or radio on their router.

 

[RMerlin]

Fun story - Back a couple of jobs ago - we had a lab/dev rack full of servers and drive arrays - some of the machines had 7 years uptime (long story, it was a 'forgotten' rack from a developer that left years ago) - anyways, the data center lost power, the UPS in the rack had dead batteries, so everything went down hard. When power came back up, half that rack was dead, and the servers had to be decommissioned.

Also had a friend who moved a bunch of PCs to a different floor at his office, a long time ago, many of them had been running for years. More than a few failed to power back on after the move.

 

[RMerlin]

So I understand what you are saying, A.power cycling does no harm and in fact can be a good way to drain everything and restart afresh.B.the best way to do it is by pulling the plug out of the wall socket?

If it's a router, I would suggest you eject any plugged USB disk first before pulling on the power plug