Indexing Bug ????

  • Thread starter Thread starter Martin Racette
  • Start date Start date
I run a rule (actually a lot of similar rules) which MOVES an incoming
message to a specified folder. The vista search engine reports this email to
both folders; the correct folder where the email has been moved to and also
to the inbox!!!! As you understand this email doesn’t exist in the inbox
because the rule has moved it. Your suggestion to clear and rebuild the index
is not of any use because the search engine duplicates almost all new
messages.
By the way how can I move the preview pane to the bottom? I didn’t find a
way to do this and most probably it is impossible. The 2.6 desktop search
engine of XP was 100 times better than this of the vista.

I see myself reinstalling the XP and I won't be the only one I believe, for
more reasons than only the search engine.

Regards,
Christos


Ilia Sacson said:
2.6 won't work on Vista, sorry. Does it create a new duplicate for each
message you receive? If it doesn't you can work around this issue by
unchecking/OK/checking Outlook in Indexing Options -> Modify... dialog. It
will reindex the emails and get rid of duplicates, faster than rebuilding
everything.
Ilia

Arend said:
I am having the same problem also with my emails. The search results report
the same email twice. And the funny thing is that the index sees the email
twice IN THE SAME FOLDER. If you go to this folder from outlook you see
only
one email. I urgently need a work around to this problem. Can I install in
vista the WDS 2.6 for XP, which did a perfect job?

Regards,
Christos


Ilia Sacson said:
One more thing - the from location is on NTFS partition, right? No FAT32?

Could you do the following:
- Open Indexing Options, wait until the indexer is idle.
- Run the following command: logman start "test1" -p
"Microsoft-Windows-Search-Core" -o "test1.etl" -ets
- Move a file from one indexed location to another indexed location.
- Wait until the indexer is idle again.
- Run the following command: logman stop "test1" -ets
- Verify that moving created a duplicate. If it didn't - try again.
- Zip and email me the test1.etl file.

The ETL file will contain the sequence of events inside the indexer.
Hopefully it will allow me to undertand your problem better. Note - it
WILL
contain the name of the file you are moving.

Thanks,
Ilia

That is exactly what I mean, whenever I move file from folder to
another,
and the use the search I find file at both location even if they aren't
in
the first location anymore.

No need to be sorry, you just want make sure that we understand what
the
problem is exactly

--
-----
Thank you in advance
Merci a l'avance

Martin
By reproducing I don't mean to find duplicates that exist already, but
to
create a new duplicate now - create file, move file, does it duplicate
in
search results? Sorry for asking that many questions, but we can't
reproduce this issue, even though it seem to be fairly common in the
wild. So I need every possible detail. I wish I could see your gather
log
files, but these contain file names of all the files indexed, that's
personal information.
Thanks,
Ilia


I'm sorry, but I'm not certan that I quite follow you. By default we
only index C:\Users. So you've had C:\Users and (say) D:\Data
indexed,
and the indexer was idle. Then you moved D:\Data\Wrong\foo.txt to
D:\Data\Right\foo.txt, and now if you search for "foo" you find it
in
both locations. Is that correct?

Is it easy to reproduce? If you create a new file in the "wrong"
location now and move it to the "right" one, will you get
duplicates?

I can replicate it every time that I run a search, and yes it does
affect all file type whether I use tag or filename as a reference


Also does the problem affect files of all types?

Well, it's not a fix, it's a workaround. It isn't perfect. Once the
ghosts has been removed, you can add the location back again, that
will
index the new files.

Thank you for followin up on this,
Ilia


Dear Martin,

Can you recall the state of the indexer when you moved the files?
Was
it idle? If not, was it working on a new location you've added via
Control Panel?

It is not nor was it a new location in the Index, the location is
on
another HDD than the boot drive, and the state of the Indexing was
IDLE


As a workaround, you can try excluding the original file location
from the indexing scope. For example, say you index entire D:
drive,
and your files were moved from D:\X to D:\Y. In the control panel,
go
to Modify..., all users, and uncheck D:\Y. If it doesn't exist any
more - create it first. If it exists and is unchecked, check it,
OK,
reopen the dialog and uncheck.

You're idea is not bad but in this case it would not work very well
since I still use the old location to put some new documents that
do
belong in that location

Thanks,
Ilia

Last night the indexing finished, and this mornning I tried the
same
search as before and it still find the files in both location

BTW. I just realised when I ran the search this morning that I'm
using the TAG attribute to search those files

THe event viewer's got nothing releated to Indexing or search,
execpt when I did perform searches

Dear Martin,

I was addressing my reply to "kirk jim". Unless this is your
alter-ego,
please be assured that I haven't had you in mind :) Something
is
definitly
not well with your Vista, and we need to get to the bottom of
it.

1) Could you check if there are any interesting events in the
event viewer?
It's in Control Panel -> Administrative tools -> Event Viewer
(or
you can
Start->Run eventvwr.msc). Search Service events are under Event
Viewer
(Local) -> Windows Logs -> Application. To make the log show
Search events
only, choose "Filter Current Log..." on the Action Panel, and
check "Search"
under Event source. We are interested in Warning or Error
events.

2) If the service or one of its sandbox processes crash, it
creates an entry
in Control Panel -> Problems and Solutions -> View Problem
History. Its
entries would be under Search Indexer, Search Protocol Host and
Search
Filter Host. If the Status column says "Report Sent", then we
have
a memory
snapshots (a.k.a. crash dumps) of your indexer somewhere on our
server.
Right click on one of the entries, choose View Problem Details,
and send me
the BucketID. I'll try to find that particular dump and see
what's
going on.
Note that they contain very little data (stack trace and
minimal
internal
state), so even if it was indexing your tax report when it
crashed
I won't
be able to read it. If the Status is Not Reported, there should
be
an extra
entry in View Problem Details titled "Files that help describe
the
problem".
You can email them to me directly.

There are other, more powerful ways to diagnose the indexer
issues, but they
are likely to expose your data, so let's try these first.

Thanks,
Ilia

I'm sorry if I started a war here, but all I wanted to do was
to
try and find a way to fix this

BTW. you asked if the indexing is done, the answer is NO, I
have
so far over 463670 items indexed, and it still going on. I
have
Vista installed since 1 February, and the indexing was never
finished yet

Dear Cap. Kirk,

I have about 2 millions of items in my index (a few years of
email and lots of source code), your puny 50K of PDFs got
nothing on it. Unless your 200Gb is mostly text (like my
stuff),
item sizes don't matter either since there's not much to
index
in a movie or "other downloads". So, my data corpus is bigger
than yours, and in principle it should take longer to finish
indexing it. It takes about 3 days to rebuild my index, that
should be fairly representative of a move performance, and in
my
opinion it's not too bad. And yes you can opt out. Open
Indexing
Options control panel, Modify..., Show all locations, uncheck
everything and click OK, done. Although I can't imagine my
work
without being able to search.

There are no known issues with file moves, so we'd like to
figure out what Martin run into and if there really is a
bug -
fix it for the next version. Knowing if the index is idle
would
be a great start. I don't see a reason to start a flame war
here, if you need one that is.

Best regards,
Ilia

I recently moved, rename and shifted around 200 gb of files,
that include 50.000 pdf books, 1000 movie files, and
thousands
of other downloads and other files.

Now tell me smarty pants.. how will vista cope with all
those
changes?

I have seen that indexing becomes useless when people have
too
many files and shift them around too much. Not only useless,
it
also slows the computer down, makes the hard drive thrash
for
days if not weeks, and gives you erroneous results!

Indexing should have been available as an ADD on, IF you
wanted
it. Not turned on be default
with no apparent way to turn the frikin thing off.


The index should update it in time. I duplicated your issue
on
two machines...the information was provided as an option.
One one machine with a few thousand files indexed it
updated
in about 2 minutes. On the other with other partitions also
indexed, it took about 15 minutes.
If nothing else is wrong(ref Dave Wood [MS] post)the folder
 
Earlier on this thread I've asked Martin to collect some error logs for me.
I'd greatly appreciate it if you could follow the instructions I gave him
and send me the logs. I'd also need to know what version of Outlook you are
running. You can email me at (e-mail address removed).

Thanks,
Ilia

Arend said:
I run a rule (actually a lot of similar rules) which MOVES an incoming
message to a specified folder. The vista search engine reports this email
to
both folders; the correct folder where the email has been moved to and
also
to the inbox!!!! As you understand this email doesn’t exist in the inbox
because the rule has moved it. Your suggestion to clear and rebuild the
index
is not of any use because the search engine duplicates almost all new
messages.
By the way how can I move the preview pane to the bottom? I didn’t find a
way to do this and most probably it is impossible. The 2.6 desktop search
engine of XP was 100 times better than this of the vista.

I see myself reinstalling the XP and I won't be the only one I believe,
for
more reasons than only the search engine.

Regards,
Christos

....
....
 
What I'm really after, is a predictable Search that can be trusted to
find things. I don't mind if it takes a while; what I don't want is
something that lies to me about what is or is not present.

You will probably never get a predictable search that can be trusted
from Microsoft.

I have not tried it on Vista yet, but I used Agent Ransack for
searches in XP. Did more than an adequate job.
 
Back
Top