If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Rate Thread | Display Modes |
#1
|
|||
|
|||
Slow Boot Time
I just updated to version 1709 (OS Build 16299.64). The process went
slow but luckily uneventful. However the start-up time now has slowed down considerably (3-4-mins). The graphic driver is up-to-date and 'sfc /scannow' revealed nothing. I disabled the fast-boot option rebooted and then re-enabled fast-boot but the start-up time remains very slow. I then disabled my two 'high impact' start-up services (Dropbox and OneDrive but this didn't make any difference either; In fact I can not detect any notable time difference between fast boot and 'normal' boot options. Any suggestions to improve the start-up time are most welcome. |
Ads |
#2
|
|||
|
|||
Slow Boot Time
|
#3
|
|||
|
|||
Slow Boot Time
WayFarer wrote:
I just updated to version 1709 (OS Build 16299.64). The process went slow but luckily uneventful. However the start-up time now has slowed down considerably (3-4-mins). The graphic driver is up-to-date and 'sfc /scannow' revealed nothing. I disabled the fast-boot option rebooted and then re-enabled fast-boot but the start-up time remains very slow. I then disabled my two 'high impact' start-up services (Dropbox and OneDrive but this didn't make any difference either; In fact I can not detect any notable time difference between fast boot and 'normal' boot options. Any suggestions to improve the start-up time are most welcome. In my travels, I've noticed the system collecting this file. It is bootckcl.etl First of all, ETW is a tracing system that Windows has now. It's been around for a while, probably from WinXP days. The sysinternals.com "Process Monitor" uses ETW, to collect real-time events from all the running programs. However, Process Monitor does not have any data mining in it. Process Monitor even has duplicate capabilities to the bootckcl file, but the Process Monitor implementation has a few bugs. It's kinda hit and miss, to get any log at all. Process Monitor injects Procmon23.dll (hidden file) into the system folder, when it wants to do a boot trace. The 23 is a version number. And in a shocking lack of hygiene, that DLL is left there after the session is completed. When the OS does tracing, it can use more than one file extension. In this case, it's an Event Tracing Log. So the OS has inherited the feature from Process Monitor. I noticed bootckcl.etl, because it kept showing up in a defragmenter log. But I didn't do a search on the file name until just now. https://blogs.technet.microsoft.com/...at-every-boot/ Using my disk inventory, I can see that both Windows 8 and Windows 10 still have versions of xperf (it's a separate download, but I noticed that both Win8 and Win10 kits exist). The xbootmgr also allows collecting a boot trace, if you want to learn how to use it. The xperf analyses the output. I:\Program Files (x86)\Windows Kits\8.1\Windows Performance Toolkit\xbootmgr.exe I:\Program Files (x86)\Windows Kits\8.1\Windows Performance Toolkit\xperf.exe N:\Program Files (x86)\Windows Kits\10\Windows Performance Toolkit\xbootmgr.exe N:\Program Files (x86)\Windows Kits\10\Windows Performance Toolkit\xperf.exe The Technet article says to look for WADK. This is the stub loader for WADK. You tick the tick box for WPT, and *only* WPT is downloaded (maybe 50MB or so). If you download the entire WADK, it's gigabytes in size. All you want, is a tiny fraction of that. WPT contains WPA and WPR, but all you'd want is xperf.exe and maybe xbootmgr.exe (if they're not in there, then you'd have to go back to the Win8 kit, but the disk information above suggests there is a Win10 version). WPR and WPA take a couple hours to run their default trace, which is a special kind of lunacy. http://download.microsoft.com/downlo...k/adksetup.exe ( https://developer.microsoft.com/en-u...deployment-kit ) This bootckcl.etl is hard to see. It won't show up in a search. I tried Agent Ransack just now, and it ignored this folder. However, because I keep a collection of file lists for all 22 disks here, I could easily consult my index and get a general idea where to look. Once I knew where to look, drilling down with File Explorer was easy. This is on the Win10 machine, running across from me. C:\Windows\System32\WDI\LogFiles\BootCKCL.etl 31,457,280 bytes Now, as traces go, that's pretty small, which means it didn't collect a lot of detail. A larger trace, with a few more switches, could be on the order of 800MB. One of the pet peeves with boot traces, is when do they stop ? Some stop before the "interesting" activity happens. I've even used the command line, to set a duration for the damn trace, and the tracing software ignored me. The trick with all these tools, is getting numbers and graphs that humans can use. I've *never* been satisfied with the above mentioned tools. Only Bootvis from WinXP days, came even close. And Bootvis isn't going to be a good candidate for any newer systems (orphan software). AFAIK, these tools continue to have the "SVCHOST problem". When services hide in a SVCHOST, we can't get their name. We don't know who is doing it. All we know is on this boot cycle, SVCHOST PID 916 is busy, then on the next boot, we know SVCHOST PID 1023 is busy. The process ID is random, and a function of race conditions in the OS (no two boots are exactly alike). So we know some SVCHOST of the umpteen SVCHOSTS did it, but we don't know which one. The tracing system doesn't allow correlating this. I haven't found a way yet, but I'm sure Mark Russinovich knows. If a Service is holding up your boot, then you'd be hard pressed to find the necessary info to get an ID string. But if I had to figure out a slow boot, the above is how I'd try to do it. You can use your intuition of course. If you have a file share that you automatically mount each time the system boots, and you happen to be plugged into a different network, that mount will fail. And that'll add five minutes to your boot time. Mark Russinovich wrote an article about using his tool kit to figure out it was the network mount - when we know he would have had a suspicion about that, with no tools at all. Still, it made for an interesting read. ******* Just for fun, somebody in Japan appears to have used WPA to analyze a bootckcl.etl. http://n.mynv.jp/column/windows/287/images/005l.jpg ( http://news.mynavi.jp/column/windows/287/ ) So that tells me, you're not limited to xperf.exe, you can also try WPA and load that file. You'll be exposed to all sorts of nonsense, such as DLLhost, BackgroundTaskHost, a bunch of launchers with different names that "hide" the executable. Again, reprehensible practices when you want to get your job done and get out of Dodge. Why the **** can't they design an OS which is *traceable* ? Surely their own staff must run into problems with this. ******* Good luck with your search, Paul |
#4
|
|||
|
|||
Slow Boot Time
On 19-Nov-17 1:29 PM, Paul wrote:
WayFarer wrote: I just updated to version 1709 (OS Build 16299.64). The process went slow but luckily uneventful. However the start-up time now has slowed down considerably (3-4-mins). The graphic driver is up-to-date and 'sfc /scannow' revealed nothing. I disabled the fast-boot option rebooted and then re-enabled fast-boot but the start-up time remains very slow. I then disabled my two 'high impact' start-up services (Dropbox and OneDrive but this didn't make any difference either; In fact I can not detect any notable time difference between fast boot and 'normal' boot options. Any suggestions to improve the start-up time are most welcome. In my travels, I've noticed the system collecting this file. It is bootckcl.etl First of all, ETW is a tracing system that Windows has now. It's been around for a while, probably from WinXP days. The sysinternals.com "Process Monitor" uses ETW, to collect real-time events from all the running programs. However, Process Monitor does not have any data mining in it. Process Monitor even has duplicate capabilities to the bootckcl file, but the Process Monitor implementation has a few bugs. It's kinda hit and miss, to get any log at all. Process Monitor injects Procmon23.dll (hidden file) into the system folder, when it wants to do a boot trace. The 23 is a version number. And in a shocking lack of hygiene, that DLL is left there after the session is completed. When the OS does tracing, it can use more than one file extension. In this case, it's an Event Tracing Log. So the OS has inherited the feature from Process Monitor. I noticed bootckcl.etl, because it kept showing up in a defragmenter log. But I didn't do a search on the file name until just now. https://blogs.technet.microsoft.com/...at-every-boot/ Using my disk inventory, I can see that both Windows 8 and Windows 10 still have versions of xperf (it's a separate download, but I noticed that both Win8 and Win10 kits exist). The xbootmgr also allows collecting a boot trace, if you want to learn how to use it. The xperf analyses the output. I:\Program Files (x86)\Windows Kits\8.1\Windows Performance Toolkit\xbootmgr.exe I:\Program Files (x86)\Windows Kits\8.1\Windows Performance Toolkit\xperf.exe N:\Program Files (x86)\Windows Kits\10\Windows Performance Toolkit\xbootmgr.exe N:\Program Files (x86)\Windows Kits\10\Windows Performance Toolkit\xperf.exe The Technet article says to look for WADK. This is the stub loader for WADK. You tick the tick box for WPT, and *only* WPT is downloaded (maybe 50MB or so). If you download the entire WADK, it's gigabytes in size. All you want, is a tiny fraction of that. WPT contains WPA and WPR, but all you'd want is xperf.exe and maybe xbootmgr.exe (if they're not in there, then you'd have to go back to the Win8 kit, but the disk information above suggests there is a Win10 version). WPR and WPA take a couple hours to run their default trace, which is a special kind of lunacy. http://download.microsoft.com/downlo...k/adksetup.exe ( https://developer.microsoft.com/en-u...deployment-kit ) This bootckcl.etl is hard to see. It won't show up in a search. I tried Agent Ransack just now, and it ignored this folder. However, because I keep a collection of file lists for all 22 disks here, I could easily consult my index and get a general idea where to look. Once I knew where to look, drilling down with File Explorer was easy. This is on the Win10 machine, running across from me. C:\Windows\System32\WDI\LogFiles\BootCKCL.etlÂ*Â*Â *Â*Â*Â* 31,457,280 bytes Now, as traces go, that's pretty small, which means it didn't collect a lot of detail. A larger trace, with a few more switches, could be on the order of 800MB. One of the pet peeves with boot traces, is when do they stop ? Some stop before the "interesting" activity happens. I've even used the command line, to set a duration for the damn trace, and the tracing software ignored me. The trick with all these tools, is getting numbers and graphs that humans can use. I've *never* been satisfied with the above mentioned tools. Only Bootvis from WinXP days, came even close. And Bootvis isn't going to be a good candidate for any newer systems (orphan software). AFAIK, these tools continue to have the "SVCHOST problem". When services hide in a SVCHOST, we can't get their name. We don't know who is doing it. All we know is on this boot cycle, SVCHOST PID 916 is busy, then on the next boot, we know SVCHOST PID 1023 is busy. The process ID is random, and a function of race conditions in the OS (no two boots are exactly alike). So we know some SVCHOST of the umpteen SVCHOSTS did it, but we don't know which one. The tracing system doesn't allow correlating this. I haven't found a way yet, but I'm sure Mark Russinovich knows. If a Service is holding up your boot, then you'd be hard pressed to find the necessary info to get an ID string. But if I had to figure out a slow boot, the above is how I'd try to do it. You can use your intuition of course. If you have a file share that you automatically mount each time the system boots, and you happen to be plugged into a different network, that mount will fail. And that'll add five minutes to your boot time. Mark Russinovich wrote an article about using his tool kit to figure out it was the network mount - when we know he would have had a suspicion about that, with no tools at all. Still, it made for an interesting read. ******* Just for fun, somebody in Japan appears to have used WPA to analyze a bootckcl.etl. http://n.mynv.jp/column/windows/287/images/005l.jpg ( http://news.mynavi.jp/column/windows/287/ ) So that tells me, you're not limited to xperf.exe, you can also try WPA and load that file. You'll be exposed to all sorts of nonsense, such as DLLhost, BackgroundTaskHost, a bunch of launchers with different names that "hide" the executable. Again, reprehensible practices when you want to get your job done and get out of Dodge. Why the **** can't they design an OS which is *traceable* ? Surely their own staff must run into problems with this. ******* Good luck with your search, Â*Â* Paul Thanks for comprehensive and educational response. This will take time to research... Aside what I have tried already I've customized the the size of the paging file to its recommended value which alas didn't speed up the start-up time either. Maybe one of the future Tuesday patches will fix the boot time issue. Thanks again... |
#5
|
|||
|
|||
Slow Boot Time
|
#6
|
|||
|
|||
Slow Boot Time
On Sun, 19 Nov 2017 10:59:31 +0700, WayFarer
wrote: I just updated to version 1709 (OS Build 16299.64). The process went slow but luckily uneventful. However the start-up time now has slowed down considerably (3-4-mins). Each to his own of course, but 3-4 minutes wouldn't bother me at all. My personal view is that the attention many people pay to how long it takes to boot is unwarranted. Assuming that the computer's speed is otherwise satisfactory, it is not generally worth worrying about. Most people start their computers once a day or even less frequently. In the overall scheme of things, even a few minutes to start up isn't very important. Personally I power on my computer when I get up in the morning, then go get my coffee. When I come back, it's done booting. I don't know how long it took to boot and I don't care. |
#7
|
|||
|
|||
Slow Boot Time
"WayFarer" wrote in message news I just updated to version 1709 (OS Build 16299.64). The process went slow but luckily uneventful. However the start-up time now has slowed down considerably (3-4-mins). The graphic driver is up-to-date and 'sfc /scannow' revealed nothing. I disabled the fast-boot option rebooted and then re-enabled fast-boot but the start-up time remains very slow. I then disabled my two 'high impact' start-up services (Dropbox and OneDrive but this didn't make any difference either; In fact I can not detect any notable time difference between fast boot and 'normal' boot options. Any suggestions to improve the start-up time are most welcome. By any chance did you make a change in the UEFI or BIOS to enable "Legacy Boot"? May want to check that setting. Bob S. |
#8
|
|||
|
|||
Slow Boot Time
On 20-Nov-17 7:55 AM, Ken Blake wrote:
On Sun, 19 Nov 2017 10:59:31 +0700, WayFarer wrote: I just updated to version 1709 (OS Build 16299.64). The process went slow but luckily uneventful. However the start-up time now has slowed down considerably (3-4-mins). Each to his own of course, but 3-4 minutes wouldn't bother me at all. It's not a bother per se rather an 'eyebrow-raising' surprise of dissimilarity since prior version 1709 the boot time was about less than 30 seconds. |
#9
|
|||
|
|||
Slow Boot Time
On 20-Nov-17 8:19 AM, Bob_S wrote:
"WayFarer"Â* wrote in message news I just updated to version 1709 (OS Build 16299.64). The process went slow but luckily uneventful. However the start-up time now has slowed down considerably (3-4-mins). The graphic driver is up-to-date and 'sfc /scannow' revealed nothing. I disabled the fast-boot option rebooted and then re-enabled fast-boot but the start-up time remains very slow. I then disabled my two 'high impact' start-up services (Dropbox and OneDrive but this didn't make any difference either; In fact I can not detect any notable time difference between fast boot and 'normal' boot options. Any suggestions to improve the start-up time are most welcome. By any chance did you make a change in the UEFI or BIOS to enable "Legacy Boot"? May want to check that setting. Bob S. I did not make any changes to the mobo, just a plain, straightforward implementation of Win update but will double-check on the settings; Thanks. |
#10
|
|||
|
|||
Slow Boot Time
On Sun, 19 Nov 2017 20:20:52 -0500, Wolf K
wrote: On 2017-11-19 19:55, Ken Blake wrote: On Sun, 19 Nov 2017 10:59:31 +0700, WayFarer wrote: I just updated to version 1709 (OS Build 16299.64). The process went slow but luckily uneventful. However the start-up time now has slowed down considerably (3-4-mins). Each to his own of course, but 3-4 minutes wouldn't bother me at all. My personal view is that the attention many people pay to how long it takes to boot is unwarranted. Assuming that the computer's speed is otherwise satisfactory, it is not generally worth worrying about. Most people start their computers once a day or even less frequently. In the overall scheme of things, even a few minutes to start up isn't very important. Personally I power on my computer when I get up in the morning, then go get my coffee. When I come back, it's done booting. I don't know how long it took to boot and I don't care. Slow boot time means more "background services" being loaded. Yes. These take up RAM, They take up virtual memory (in the Microsoft sense of the term: RAM + page file), not RAM. and many of them are so rarely in actual use that IMO there's no point loading them. If they are not in use, what they take up is page file, not RAM. There's no penalty for that. They also steal cycles, Only if they are in use. but that almost never matters. However, some of these services are trackers, and those should not be permitted. We agree on that. |
Thread Tools | |
Display Modes | Rate This Thread |
|
|