none
Windows search does not index certain files RRS feed

  • Question

  • We have many .doc Word files that are not indexed by Windows Search on Windows 2012R2 server. Some of the Word files are pretty old, up to 13 years old (2004). The files may not be touched since they are approved in a FDA 21CFR11 compliant system.

    We have executed index rebuild numerous times, this has no effect.

    If we copy/paste an original file that is not indexed, then the copied file is indexed immediately. 

    If we open an original file in Word (2016) and just close Word without saving the file, then the file is indexed immediately.

    In Windows 2016 we don’t seem to have this problem. Everything seems to be indexed by index rebuild, with no problems. However, we have not validated this in large scale, since it is a large job, and customer prefer not to migrate to 2016 server right now.

    How can we force indexing of all files (without touching the files) in 2012R2 if rebuild index does not do it?

    Any help is greatly appreciated.

    Many thanks

    Sven Andersen

    Friday, January 20, 2017 3:55 PM

All replies

  • Just a guess on this one ...

    NTFS indexing is handled by notification, meaning that existing files are not indexed until something about them changes.  You seem to have proven that by the way you have gotten some additional files indexed by touching the file in some manner. This changes the last access date even though you did not change the file in any manner. Not sure why it is not indexing when you request a full index, but at least there appears to be a fairly simple way to start the indexing for them.

    A simple script can be written to read every file in a directory (or recurse a directory tree) that simply opens the file to change its access date.  It would not take much to test.  And it does not take much time to run, either.  I wrote a simple PowerShell script that performs my own search. It recurses any directory I enter and searches every line in each file for a string I enter. I process thousands of files in a matter of seconds. And that involves making a comparison on every line in every file I open. So a script to simply open the file would take hardly any time at all.


    . : | : . : | : . tim

    Saturday, January 21, 2017 12:57 PM
  • many thanks Tim

    We found the reason to our problems that have been bugging us for the longest time, and it is the same problems for a lot of other good people in many many corporations.

    Windows Search is not compatible with Data Deduplication.

    https://social.technet.microsoft.com/Forums/office/en-US/d8c9c5b1-fc38-476d-a780-e14910edcf6f/widows-search-not-working-with-data-deduplication?forum=winserverfiles

    https://technet.microsoft.com/en-us/library/hh831454(v=ws.11).aspx

    The bug is still present in Windows 2016:

    https://technet.microsoft.com/en-us/windows-server-docs/storage/data-deduplication/interop

    It is so sad. I guess that Microsoft wont fix. I am sure they can, if we can put a man on moon then Microsoft can fix this. right?

    Thanks, and kind regards

    Sven

    Tuesday, January 24, 2017 11:48 AM
  • I know it seems to be a fine line, but there is a difference between a 'bug' and 'working as designed'.  the facetious definition of a 'bug' is 'an undocumented feature'.  This issue is documented. <grin>  I know it does not do what you want, and you are free to consider it a bug in your environment, but for now it is working exactly how Microsoft has designed it.

    But you are absolutely correct.  Microsoft is able to change it.  Windows deduplication is a relatively new feature compared to indexing.  Generally Microsoft will make changes if it receives enough customer input.  They have established a site, https://windowsserver.uservoice.com/forums/295047-general-feedback, that accepts inputs on a variety of topics.  I know for a fact that Microsoft product managers read these posts.  Not only does the site allow you to post your idea, but it allows others to vote for the idea.  Ideas that get more votes tend to get the product managers' attention.


    . : | : . : | : . tim

    Tuesday, January 24, 2017 1:21 PM
  • Thanks Tim :-)

    , and many thanks for the tip to report an improvement, I have done what you have told me to, but I have no expectations:

    https://windowsserver.uservoice.com/forums/295047-general-feedback/suggestions/17888647-make-windows-search-service-work-with-data-dedupli

    Yes bug is maybe a too strong term. Flaw could be the correct term, even if one may find a note on internet telling it is not supposed to Work. Also the note probably came out after release, so probably never tested except by customers. A popup warning when trying to install both services would have been much better, and save a lot of guys time trying to do the same as me, because they never stumbled accross this blog or the MS note..

    DataDeduplication to my knowledge was introduced in Win2003 OEM version, so I actually believe it is a not so new component.

    Thanks, and kind regards

    Sven Andersen

    Monday, January 30, 2017 4:07 PM