Paging Errors causing long boot times

Discussion in 'Computer Support' started by Lowly Engineer, Aug 10, 2006.

  1. My machine takes an extraordinarily long time to boot. During bootup,
    I notice that the hard drive activity light does not glow steadily.
    It just flashes at about a 2 times/second rate.

    Subsequent examination of the error log shows that there were a long
    series of errors on paging operations. This happens every time I
    reboot.

    Can someone tell me what to look for in order to fix this problem?

    System:

    Win XP Pro - all the latest updates
    Athlon 64 dual-core 4600
    DFI "LanParty" UT4 motherboard
    2 GB RAM
    Seagate 300GB SATA drive

    I tried a Hitachi 500 GB SATA drive - same problem.


    TIA
     
    Lowly Engineer, Aug 10, 2006
    #1
    1. Advertisements

  2. Lowly Engineer

    Au79 Guest

    Try running chkdsk /F /R from the command prompt.

    You may also try to set the swapfile to its own private partition.
     
    Au79, Aug 10, 2006
    #2
    1. Advertisements

  3. Lowly Engineer

    why? Guest

    It wouldn't help to not remember to include the error message, event id
    and source information from the event log? I guess not.
    <snip>

    Me
     
    why?, Aug 10, 2006
    #3
  4. Lowly Engineer

    Ron Martell Guest

    The /F parameter includes /R so all that is needed is CHKDKS /F
    Not a good idea for Windows under any circumstances. The Windows
    swap file is used entirely differently than it is in Linux and
    rules/procedures that apply to one operating system do not necessarily
    apply to the other.

    Ron Martell Duncan B.C. Canada
    --
    Microsoft MVP (1997 - 2006)
    On-Line Help Computer Service
    http://onlinehelp.bc.ca
    Syberfix Remote Computer Repair

    "Anyone who thinks that they are too small to make a difference
    has never been in bed with a mosquito."
     
    Ron Martell, Aug 10, 2006
    #4
  5. Lowly Engineer

    Ron Martell Guest

    2 gb of RAM and you are getting paging errors during startup?

    Can you provide the detailed particulars of some of these errors?

    Ron Martell Duncan B.C. Canada
    --
    Microsoft MVP (1997 - 2006)
    On-Line Help Computer Service
    http://onlinehelp.bc.ca
    Syberfix Remote Computer Repair

    "Anyone who thinks that they are too small to make a difference
    has never been in bed with a mosquito."
     
    Ron Martell, Aug 10, 2006
    #5
  6. All are ERROR TYPE 51 and I get 10 or 20 of them each time I start
    up.


    I attached a screenshot of EVENT VIEWER.
     
    LowlyEngineer, Aug 10, 2006
    #6
  7. Lowly Engineer

    Ron Martell Guest

    Got that one backwards. The /R parameter is the higher level
    function, and implies /F so the correct command is CHKDSK /R

    Apologies.

    Ron Martell Duncan B.C. Canada
    --
    Microsoft MVP (1997 - 2006)
    On-Line Help Computer Service
    http://onlinehelp.bc.ca
    Syberfix Remote Computer Repair

    "Anyone who thinks that they are too small to make a difference
    has never been in bed with a mosquito."
     
    Ron Martell, Aug 10, 2006
    #7
  8. Lowly Engineer

    Ron Martell Guest

    No attachment found (which is a good thing - this is a text-only
    newsgroup and attachments are frowned upon).

    You probably mean Event I.D. 51 so I will proceed on that basis.
    Please try to be precise when posting error information as very often
    it is the presence or absence of a single word in the error data that
    is the essential clue to the underlying cause.

    See the following article in the Microsoft Knowledge Base:

    Information about Event ID 51
    Article ID : 244780
    http://support.microsoft.com?kbid=244780


    Good luck


    Ron Martell Duncan B.C. Canada
    --
    Microsoft MVP (1997 - 2006)
    On-Line Help Computer Service
    http://onlinehelp.bc.ca
    Syberfix Remote Computer Repair

    "Anyone who thinks that they are too small to make a difference
    has never been in bed with a mosquito."
     
    Ron Martell, Aug 10, 2006
    #8
  9. EVENT VIEWER tells me that all the errors are TYPE 51
     
    LowlyEngineer, Aug 10, 2006
    #9
  10. Lowly Engineer

    Au79 Guest

    However, Ron, you do not specify the *why* it would be a bad idea to
    segregate the swapfile into its own partition.

    We understand that two operating systems designed entirely differently will
    operate as such; Page swapping nontheless works under the same principles.
     
    Au79, Aug 11, 2006
    #10
  11. Lowly Engineer

    Ron Martell Guest

    #1: Performance. If the system is actually using the swap file (and
    much of what is reported as PF Usage by Windows Task Manager is
    actually "phantom" usage) then moving the swap file to a different
    partition will result in increased travel distance for the disk head
    mechanism each time there is a paging opertion as it shuttles between
    the operating system/application/data file partition and the swap file
    partition.

    #2: Design. Windows uses the swap file (paging file to be
    technically correct) on the boot drive (and only on the boot drive) to
    receive the content of the memory dump. It then renames the file when
    the dump is complete. This is faster than creating a new file for the
    memory dump, and time is often in very short supply when a major error
    occurs. So if the paging file is on a different partition on the
    drive then it will not be possible to capture these memory dumps,
    which can be an important tool for diagnosing the underlying cause of
    an error.

    Note that this is the also the only reason that I am aware of for
    having a Windows XP swap file size that is related to the amount of
    RAM in the computer. If the "complete memory dump" option is
    configured (default) then the paging file has to be at least equal in
    size to the amount of installed RAM so as to receive the content of
    the memory dump.

    Other than this specific instance (and most machines can be safely
    reconfigured to produce a smaller memory dump which is still fully
    adequate for most error tracking purposes) the size of the paging file
    is *inversely* related to the amount of installed RAM - more RAM means
    less paging file and less RAM means more paging file provided all
    other factors are held constant.
    I am not at all familiar with the technical details of Linux memory
    management, except that tracking the source of "old wives tales" such
    as setting your swap file to 1.5 times the amount of RAM usually ends
    up with it originating with Linux/Unix.

    Also am not sure how Linux handles the unused portions of memory
    allocation requests. Windows requires that addresses must be
    assigned to satisfy the full amount of all memory requests, but
    improves efficiency by mapping the unused portions to locations in the
    swap file and using RAM only for the part that is actually used. This
    can often be illustrated by comparing the PF Usage figure reported by
    Windows Task Manager with the actual physical of the swap file. For
    example on my system at this moment with 1 gb of RAM there is PF Usage
    of 545 mb but the actual size of c:\pagefile.sys is only 80 mb. The
    maximum size limit is set at 1 gb.

    I noticed a comment in Wikipedia about a recent Linux improvement
    (2.6) called "swap prefetch". That strikes me as being very similar
    in effect to the pre-emptive swapping introduced in Windows 98 where
    pages eligible to be swapped out would be written to the swap file
    during idle times and also retained in RAM. Then if RAM was
    subsequently for another purposes those pages previously written out
    and not changed could be instantaneously dropped from RAM as they were
    already in the swap file.
    Ron Martell Duncan B.C. Canada
    --
    Microsoft MVP (1997 - 2006)
    On-Line Help Computer Service
    http://onlinehelp.bc.ca
    Syberfix Remote Computer Repair

    "Anyone who thinks that they are too small to make a difference
    has never been in bed with a mosquito."
     
    Ron Martell, Aug 11, 2006
    #11
  12. Lowly Engineer

    Au79 Guest

    Ahh... Are you saying that the page file is written/read in close physical
    proximity to whatever application or process is using it? Why should a
    partition, which could be in the same platter, make performance an issue? I
    doubt very seriously that if there is indeed a performance hit, it would be
    of any significance, or even measurable. At any rate, in the Windows world,
    fragmentation has taken care of the performance question.
    It's a mystery why memory dumps cannot be written outside the boot drive. If
    so, it very much seems like a design flaw.
    Quite true, since Windows 3.1.
    Maybe, but it seems like a mute point.
    Without desiring a flame war, my opinion is that Linux has a smoother
    improvement continuum on memory management that should not be compared with
    Microsoft's development efforts. I trust in the reliability and efficiency
    of my 2.6 Kernel way more than Windows 98/2000/XP Pro.
    Nice treatease on memory management, well done.
     
    Au79, Aug 11, 2006
    #12
  13. Lowly Engineer

    Ron Martell Guest

    Yes, especially with regard to paging out operations, which most often
    occur in conjunction with another activity such as launching a new
    application, which triggers a requirement to free up RAM to load that
    application. So paging out operations will be interspersed with file
    loading operations in most instances.

    Hard drive "seek" times vary quite a bit. A typical PATA or SATA
    drive sold today with an *average* seek time of 8 milliseconds will
    have actual seek times that vary from 2 milliseconds (track to
    adjacent track) to over 20 milliseconds (full stroke seek). The point
    is that the greater the distance between the locations being used by
    two concurrent disk operations, such as loading a new app and paging
    out, the longer it will take.

    It might be considered as such. But that was the design decision and
    as I recall it was based purely on the need for speed. Using an
    existing file was shown to be faster than creating a new file for the
    dump, and as the paging file content is totally dispensable that was
    the choice. Even the potential delay while the hard drive
    repositioned to another partition was considered excessive if it could
    be avoided.

    Ron Martell Duncan B.C. Canada
    --
    Microsoft MVP (1997 - 2006)
    On-Line Help Computer Service
    http://onlinehelp.bc.ca
    Syberfix Remote Computer Repair

    "Anyone who thinks that they are too small to make a difference
    has never been in bed with a mosquito."
     
    Ron Martell, Aug 11, 2006
    #13
  14. Lowly Engineer

    why? Guest

    Guessing your EVENT VIEWER is different from the usual MS supplied one?

    The MS one has Event IDs , so 51 could be that instead of a TYPE.

    More information would be handy, along with the ID number, there should
    also be a message and a source process name. Oops sorry, I already said
    that in an earlier post.

    How to use Event Viewer
    http://support.microsoft.com/kb/308427/
    How to Interpret an Event
    Each log entry is classified by type, and contains header information,
    and a description of the event.

    Try entering some of the info you get from the event log here
    http://www.microsoft.com/technet/support/ee/ee_advanced.aspx

    Events and Errors Message Center
    Results 1 - 8 of 8 for: Event ID: 51;

    A nive easy read, only 8 hits.

    Me
     
    why?, Aug 12, 2006
    #14
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.