PDA

View Full Version : Does 10.9 suffer from same slow shutdown issue as 10.8?


Krazy Bill
Jun 12, 2013, 09:28 AM
Exact same 20 second timeout bug. (Check your console logs). LOL! I give up. :)

P.S. Queuing the apologists who are about to lecture on how stupid it is to shut down/reboot.

But hey... multi monitor support is kind of awesome. It's like I have 2 different and independent machines sitting side-by-side. (sort of) :eek:

justperry
Jun 12, 2013, 09:31 AM
We probably have to wait until 10.10 for it to get fixed Krazy Bill, all we can do now is killing these processes ourselves, sadly.:mad:

chrfr
Jun 12, 2013, 09:31 AM
Exact same 20 second timeout bug. (Check your console logs). LOL! I give up. :)
It's safe to assume that Apple does not consider this a bug.

RedRaven571
Jun 12, 2013, 09:46 AM
*grumble*, *grumble*

I may just have to bite the bullet this time and live with the slow(er) shut down, Apple is adding too many cool things for me to stay on SL.

Krazy Bill
Jun 12, 2013, 09:49 AM
We probably have to wait until 10.10 for it to get fixed Krazy Bill, all we can do now is killing these processes ourselves, sadly.:mad:

It's safe to assume that Apple does not consider this a bug.

Accurate assessments on all counts. I just wish they'd make that little gray spinning wheel more entertaining since it's there for so long. Maybe put arrows and numbers on it. Pick one as we reboot and see if it lands on it just before the black screen.

Sometimes I bootcamp into Windows and forget why. :eek:

justperry
Jun 12, 2013, 09:52 AM
Accurate assessments on all counts. I just wish they'd make that little gray spinning wheel more entertaining since it's there for so long. Maybe put arrows and numbers on it. Pick one as we reboot and see if it lands on it just before the black screen.

Sometimes I bootcamp into Windows and forget why. :eek:

Or switch on verbose mode, at least you can then see when it restarts.:)

w0lf
Jun 12, 2013, 11:14 AM
For me my shutdown has been even slower than Mountain Lion. Not really a problem for me but I'm just throwing that out there.

Krazy Bill
Jun 12, 2013, 11:26 AM
For me my shutdown has been even slower than Mountain Lion.Well, 20 seconds or 27... it's all the same to me when you're being hypnotized by the spinning gear wheel. :D

I haven't tried the terminal hacks yet. Will do that eventually. I can certainly live with it after that. Just can't believe nobody at Apple shuts down their machines.

chrfr
Jun 12, 2013, 12:45 PM
Just can't believe nobody at Apple shuts down their machines.
Or they have a valid reason for increasing the timeout.

RedRaven571
Jun 12, 2013, 01:04 PM
Or they have a valid reason for increasing the timeout.

It is during the time you are watching the spinning gray wheel that your mac is sending subliminal messages to your brain....

BUY MORE APPLE PRODUCTS
BUY MORE APPLE PRODUCTS
BUY MORE APPLE PRODUCTS
BUY MORE APPLE PRODUCTS
BUY MORE APPLE PRODUCTS
BUY MORE APPLE PRODUCTS

:eek::eek::eek:

Krazy Bill
Jun 12, 2013, 01:12 PM
Or they have a valid reason for increasing the timeout.What pray tell could that be? (I'll wait here). A process "times out" because something failed to do what it was supposed to do. :) Try shutting down from the log in screen... when nothing of consequence has loaded like your sky drive/drop box accounts, messages, mail... you'll get the same results.

Don't drink the apology kool-aide. There are just times when Apple screws up. :D

chrfr
Jun 12, 2013, 01:37 PM
Don't drink the apology kool-aide. There are just times when Apple screws up. :D

I'm not apologizing for Apple. The timeout change seems too simple to be accidental.

JohnDoe98
Jun 13, 2013, 07:17 AM
I'm not apologizing for Apple. The timeout change seems too simple to be accidental.

The fact that there is a timeout, however long, isn't an issue. The issue is why aren't certain processes responding within that interval, over and over? Something is wrong with the way those processes shutdown, or I should say, fail to shutdown consistently. That's what needs to be fixed, not changing the timeout to kill them earlier. We change the interval to deal with the issue but this isn't a permanent solution, more of a band-aid remedy in the meantime.

definitive
Jun 19, 2013, 11:25 AM
I've seen countless posts (especially over at Apple's support forums) mention that Mountain Lion has a slow shut down issue, even after 10.8.4 has been released. I'm experiencing this on two Macs that I own, and haven't found any way to fix this (repaired permissions, cleared cache, fiddled with Terminal and preferences files, and even reinstalled the OS from scratch). It takes roughly 15-20 seconds for the system to shutdown, compared to 2-5 seconds in Lion and early release builds of Mountain Lion.

Has this been addressed in 10.9?

Galaxas0
Jun 19, 2013, 11:31 AM
I think this is still there, but literally five minutes before I saw this post, I was optimizing away at faster shutdown and boot up speeds. I know I get 5 second boot up into my completely active desktop (Mac Mini + 512 SSD + 16GB RAM), but I recall seeing ~20 second shutdown times, and console logs may have indicated a runaway process waiting for a timeout.

Eithanius
Jun 19, 2013, 11:39 AM
Yes, no doubt about it...

RedRaven571
Jun 19, 2013, 11:45 AM
Yes, my experience is that shutdown with 10.9 is as slow as 10.8 was on my machine. Granted, we're talking about 20-30 seconds, but definitely significantly slower than SL.

Drew017
Jun 19, 2013, 02:30 PM
Mine will sometimes shutdown in less than 5 seconds, and then other times it takes 20 or so... it's like it can't decide.

w0lf
Jun 19, 2013, 03:17 PM
Nope not fixed.

Still takes me 20-30+ seconds to shut down with an SSD.

smithrh
Jun 19, 2013, 03:22 PM
Really wondering why anyone thinks this is an issue in the first place.

Sleeping your Mac is energy-efficient, especially on the most recent models, and you get nearly instant-on as well.

Why the compulsion with a few seconds in performing an infrequent activity?

JohnDoe98
Jun 19, 2013, 03:39 PM
Really wondering why anyone thinks this is an issue in the first place.

Sleeping your Mac is energy-efficient, especially on the most recent models, and you get nearly instant-on as well.

Why the compulsion with a few seconds in performing an infrequent activity?

Well some people need to dual boot into Windows, so for them shutdown down might be a more frequent activity. Others might find that shutting down and rebooting actually preserves more battery than letting the computer sleep, so for them the extra battery life is more of a priority than instant on. In those cases it's hard to be happy with a system taking longer to shutdown then it ought to.

In any case, since the systems are not shutting down properly, as indicated by the Console logs, what I don't understand is why this isn't being fixed in the first place. Yes it's priority might not be very high, but it does seem like it ought to be addressed. Our systems should be running optimally after all.

Quercus Alba
Jun 19, 2013, 06:51 PM
Really wondering why anyone thinks this is an issue in the first place.

Sleeping your Mac is energy-efficient, especially on the most recent models, and you get nearly instant-on as well.

Why the compulsion with a few seconds in performing an infrequent activity?

I shut my iMac down every night and I think it would be weird to leave it sleeping overnight. It'd waste power too. It is a pain to have to stand there waiting for 30 seconds for the silly thing to switch off before I can turn the mains off, especially when it used to shut down in about 5 seconds with Leopard > Lion

Tubamajuba
Jun 22, 2013, 06:49 PM
I shut my iMac down every night and I think it would be weird to leave it sleeping overnight. It'd waste power too. It is a pain to have to stand there waiting for 30 seconds for the silly thing to switch off before I can turn the mains off, especially when it used to shut down in about 5 seconds with Leopard > Lion

Not gonna be ridiculous and scold you for an absolutely harmless personal decision, but I do want to point out that the amount of power your iMac uses while sleeping is pretty inconsequential- certainly not enough to cancel out the benefit of it being ready to go almost right away in the morning.

Still though, you shouldn't have to deal with long shutdown times if you choose to power off your iMac every night. I can't believe this has been an issue for as long as it has been- either the OS X team doesn't think it's a high priority issue, or the root of the problem is deeper than we realize. It doesn't take away from how amazing Mavericks is so far, but it's a strange quirk for sure.

ScottishCaptain
Jun 23, 2013, 12:18 AM
Not gonna be ridiculous and scold you for an absolutely harmless personal decision, but I do want to point out that the amount of power your iMac uses while sleeping is pretty inconsequential- certainly not enough to cancel out the benefit of it being ready to go almost right away in the morning.

Still though, you shouldn't have to deal with long shutdown times if you choose to power off your iMac every night. I can't believe this has been an issue for as long as it has been- either the OS X team doesn't think it's a high priority issue, or the root of the problem is deeper than we realize. It doesn't take away from how amazing Mavericks is so far, but it's a strange quirk for sure.

How do you know that it is even a problem to begin with? How do you know that the quick shutdowns weren't the actual bug that got "fixed"?

OS X is a very, very complicated beast. There are hundreds of processes running in the background that combine to form the environment you, as a user, interact with. When a shutdown event is triggered, a process called launchd (which all other processes run under) attempts to shut down as many processes as possible in parallel.

A lot of the times, this doesn't go as planned. Some processes need to be shut down before others. Some processes need to wait for a system resource to quiesce before they can terminate cleanly. Others will try to dump data to disk before they're killed, and if there's multiple processes all doing this at the same time it can block the disk I/O, causing the shutdown to slow down or momentarily stall.

I honestly don't think the OS X team cares about this, because in the end- the system always shuts down, and that is all they care about. Trying to track down the cause of something that stalls the shutdown procedure for 5 minutes would be easy- that's a bug that should be fixed. Trying to figure out why the process is taking 20 seconds instead of 5 could be impossible. That extra 15 seconds could be divided among a hundred different processes, all of which are each taking slightly longer to quit for whatever reason. Are you going to pull a thousand different people off fixing critical bugs and writing new code just to go hunting around to shave off a few milliseconds off how some process is supposed to terminate? I doubt it.

The only thing any of you can do, if it really matters to you- is to send feedback to Apple:

http://www.apple.com/feedback/

If you are a developer, you can send them a bug report through Radar:

http://radar.apple.com

-SC

w0lf
Jun 23, 2013, 12:33 AM
How do you know that it is even a problem to begin with? How do you know that the quick shutdowns weren't the actual bug that got "fixed"?

OS X is a very, very complicated beast. There are hundreds of processes running in the background that combine to form the environment you, as a user, interact with. When a shutdown event is triggered, a process called launchd (which all other processes run under) attempts to shut down as many processes as possible in parallel.

A lot of the times, this doesn't go as planned. Some processes need to be shut down before others. Some processes need to wait for a system resource to quiesce before they can terminate cleanly. Others will try to dump data to disk before they're killed, and if there's multiple processes all doing this at the same time it can block the disk I/O, causing the shutdown to slow down or momentarily stall.

I honestly don't think the OS X team cares about this, because in the end- the system always shuts down, and that is all they care about. Trying to track down the cause of something that stalls the shutdown procedure for 5 minutes would be easy- that's a bug that should be fixed. Trying to figure out why the process is taking 20 seconds instead of 5 could be impossible. That extra 15 seconds could be divided among a hundred different processes, all of which are each taking slightly longer to quit for whatever reason. Are you going to pull a thousand different people off fixing critical bugs and writing new code just to go hunting around to shave off a few milliseconds off how some process is supposed to terminate? I doubt it.


Services crashing and being force quit by the system is not a feature.

Since the bug appeared in Mountain Lion, it has always been about 4-5 services (which vary from person to person) that crash almost every time the machine is shut down and have to be killed by a built in 20 second timer.

It's not as if anyone is actually complaining about the time it takes for their machine to shut down because it takes about 5 seconds to enter the terminal commands to 'fix' the issue. The real question is why has Apple not addressed the problem.

Walter White
Jun 23, 2013, 12:50 AM
How do you know that it is even a problem to begin with? How do you know that the quick shutdowns weren't the actual bug that got "fixed"?


Are you ****ing serious?

DarkRyoushii
Jun 23, 2013, 02:49 AM
Are you ****ing serious?

Hate to break it to you but it's certainly a possibility. Just because something is faster doesn't mean it's always the best way to do it. If that were the case you would just take the battery out/switch off the power at the wall whenever you wanted a 'faster shutdown' but since we all know this may lead to data corruption on the drives no one does this.

Maybe those faster shutdowns were causing more issues than what they fixed so they were slowed down to give processes more time to finish properly or a plethora of other reasons.

ScottishCaptain
Jun 23, 2013, 03:13 AM
Services crashing and being force quit by the system is not feature.

It has always been about 4-5 services that crash almost every time and have to be killed by a built in 20 second timer.

It's not as if anyone is actually complaining about the time because it takes about 5 seconds to enter the terminal commands to 'fix' the issue. The real question is why has Apple not addressed the problem.

Interesting.

Then I apologize, you are quite correct. I did not know this was an actual technical problem. I blindly assumed you were all crazy when, in fact, there appears to be technical merit behind these concerns.

I checked and it appears as though securityd and appleevents consistently fail to shut down in a sane manner over here on my primary workstation. I have filed the relevant bug reports on radar, if you are a developer I would advise you do the same.

It is possible to check which processes are hanging up the shutdown by running the following command in Terminal.app:

cat /var/log/com.apple.launchd/launchd-shutdown.system.log | grep "Exit timeout"

This will print out a list of processes that fail to shutdown within 20 seconds of being issued SIGTERM. I'm only seeing listings for com.apple.securityd and com.apple.coreservices.appleevents over here, this is highly dependant on your system software so these two may or may not be the source of your problems.

To "fix" your slow shutdown (this isn't really a fix), issuing these commands into Terminal.app should do the trick:

sudo defaults write /System/Library/LaunchDaemons/com.apple.coreservices.appleevents ExitTimeOut -int 2
sudo defaults write /System/Library/LaunchAgents/com.apple.coreservices.appleid.authentication ExitTimeOut -int 1

Again, this is not really a fix. You are setting the exit timeout to 2 and 1 second(s), and terminating the process after that time elapses (instead of waiting 20 seconds). These processes should exit cleanly within 20 seconds, and they are not- so this fix is kind of forcing things and you shouldn't need to do that.

Oddly enough, I just installed 10.8.4 inside a virtual machine to check out this issue, and it doesn't exist on a fresh install. Both securityd and appleevents terminate almost instantaneously, so there is something else hanging them up. The VM shuts down and restarts almost instantly. securityd is actually part of the Security project, the source code for which is available on opensource.apple.com. It is a relatively simple utility, but from what I can see there should be no reason why it would hang up and refuse to quit when signalled with SIGTERM- so I'm guessing there is some sort of deeper issue within CoreFoundation that needs to be researched and fixed.

In any case, the only thing I can recommend at this point is to let Apple know about it. If this stuff really bugs you, you can always run the two above commands (the ones with `defaults` in them). This should fix the issue, but it's not an ideal solution at all.

-SC

Walter White
Jun 23, 2013, 04:57 AM
Whatever the issue it didn't exists in the ML GM and later 10.8.1. Something changed in 10.8.2 and from then on it only got worse. Also the fact that same issue exists in the 10.9 dev preview … don't get your hopes high.

Krazy Bill
Jun 23, 2013, 10:15 AM
Interesting.

In any case, the only thing I can recommend at this point is to let Apple know about it.
-SC

Many of us have developer accounts and have reported the problem since September of 2012 (and earlier). I've submitted console logs. Apple does indeed know about this. Suffice it to say, I've given up.

Chasing an issue that's hard to replicate is one thing (been there), but choosing not to fix one that can be reproduced on practically every reboot is another - it's actually mind boggling.

Is the slow shutdown a showstopper? No. But this bug merely gives one insight into the cranial workings of the OSX team and their methodology (or lack thereof) which in turn makes one scratch his/her head about all the other nuances in OSX they are charged with. To me, THIS is the interesting part and very telling.

RedRaven571
Jun 23, 2013, 11:47 AM
Hate to break it to you but it's certainly a possibility. Just because something is faster doesn't mean it's always the best way to do it. If that were the case you would just take the battery out/switch off the power at the wall whenever you wanted a 'faster shutdown' but since we all know this may lead to data corruption on the drives no one does this.

Maybe those faster shutdowns were causing more issues than what they fixed so they were slowed down to give processes more time to finish properly or a plethora of other reasons.

What about all of us folks using SL (and its quick shut downs) for years without a problem, don't you think we would have noticed?

Transeau
Jun 23, 2013, 12:49 PM
The "wasting money and electricity" doesn't work for me. The math is pretty simple and I wont bore anyone with it. But the basic facts are this....

iMac On: 80W
iMac Sleeping: 1W
iMac Off: .2W
Cost of Electricity (Southern California): $0.14/kWh

The basic math shows that the difference between OFF and SLEEPING is 2.4W over 12 Hours, OR $0.00112. In other words, a sleeping iMac (27", Core i7 with 16GB ram) takes 24 hours to use the same amount of power as a child's night-light uses in 1 hour. Or, 12 hours of sleep is the same as 8 minutes of running with all 4 cores and GPU maxed out.

I don't power down my Macs at night. I don't see the need. Besides, 75% of what you would save is simply wasted in the amount of time it takes to power off at night, and back on in the morning.

Krazy Bill
Jun 23, 2013, 01:54 PM
I don't power down my Macs at night. I don't see the need.Neither do I. When I'm at home. But try traveling 300-400 miles in a day with a dozen stops - each requiring either:

A.) Sleep, or...
B.) Shut Down

I'd prefer "A" but over the course of a long day I lose at least 10-15% battery (at least). This often makes all the difference between a necessary adapter plug-in. So... I shutdown and reboot between stops.

Recently though, when on the road I've just set my startup drive to a Windows 8 partition. 2-3 seconds shutdown max. My battery life is about 30 minutes longer as well.

All this is moot though. OSX should shutdown quickly like it did in the past. (If only because it's a bug that should piss off any rational-thinking coder). Those of us who have habits that rely on this, good for us - those that don't care can just continue to ignore it.

xgman
Jun 24, 2013, 12:22 PM
Really wondering why anyone thinks this is an issue in the first place.



Clearly Apple does not....

SeenJeen
Jun 24, 2013, 04:25 PM
Installed Mavericks today. It's just as bad as Mountain Lion.

The thing is, I bought this Mac Mini (late 2012) for my office, and sometimes I like to bring the Mac Mini home to watch Netflix or whatever on my big screen.

So at the end of the day, sometimes I'm standing around for a minute feeling like an idiot waiting for it to shut down.

My 2009 Snow Leopard machine was off in 5 seconds flat.

I'm grateful for Mavericks (its much snappier than Mountain Lion... which made this new Mac run slower than my SL Mac... :mad: )

ultraspiracle
Jul 6, 2013, 06:36 PM
Yes, 10.9 lags with shutdown just like 10.8.4. It appears to be Launchd processes again, and the same ones!

Note that the shutdown issue is fixed in 10.8.5 12f17. So far, I haven't seen it do a slow shutdown yet, no matter what apps are running in the background on my friend's beta machine; it shuts down completely cold in less than 5 seconds. Unless something changes with the next preview, I think 10.8.5 is set.

Following the same trend, I expect 10.9 will be fixed at some point.

w0lf
Jul 6, 2013, 07:08 PM
Note that the shutdown issue is fixed in 10.8.5 12f17. So far, I haven't seen it do a slow shutdown yet, no matter what apps are running in the background on my friend's beta machine; it shuts down completely cold in less than 5 seconds. Unless something changes with the next preview, I think 10.8.5 is set.

Following the same trend, I expect 10.9 will be fixed at some point.

My fingers are crossed.

kensic
Jul 7, 2013, 01:11 AM
my shut down issue was fixed with 10.8.3, well 95% of the time. so im very happy with that.

ultraspiracle
Jul 7, 2013, 05:21 AM
Have you guys tried beta 2 of 10.8.5? On my ssd machine the problem has gone away permanently. I confirm that mavericks dp2 is still slow, however.

Not even the terminal commands to kill the stuck processes worked for me in 10.8.4...slow as molasses to shut down .

I Drink Your
Jul 7, 2013, 05:25 AM
Have you guys tried beta 2 of 10.8.5? On my ssd machine the problem has gone away permanently. I confirm that mavericks dp2 is still slow, however.

Not even the terminal commands to kill the stuck processes worked for me in 10.8.4...slow as molasses to shut down .

Just as slow as before and it has nothing to do with SSD.

ultraspiracle
Jul 7, 2013, 03:56 PM
Just as slow as before and it has nothing to do with SSD.

Do you have an SSD, Retina or Macbook Air?

The main point is that SSD and Flash mem users notice the shutdown anomaly more since the machines boot and shut down so fast already....or at least they're supposed to.

Not sure if Retina or Air users have slow shutdown problems with 10.8.5. How about it guys? Anyone confirm?

ultraspiracle
Jul 11, 2013, 07:44 PM
Yes, 10.9 lags with shutdown just like 10.8.4. It appears to be Launchd processes again, and the same ones!

Note that the shutdown issue is fixed in 10.8.5 12f17. So far, I haven't seen it do a slow shutdown yet, no matter what apps are running in the background on my friend's beta machine; it shuts down completely cold in less than 5 seconds. Unless something changes with the next preview, I think 10.8.5 is set.

Following the same trend, I expect 10.9 will be fixed at some point.

I guess I'm about ready to give up on this issue too. Slow shut down is back in 10.8.5 12f20; clearly the lack of it in 12f17 was something unique to my machine.

They closed a recent bug report filed by someone I know on this issue as a duplicate, saying "there was no more information available"

I'm going to keep digging at it, I guess.