Outlook 2007 compound (OR) instant search break down

[GoodManStuckNMicrosfWorld]

"us.pwc" OR detail OR "through"

fails to find a 5 day old message whose subject has the word through, even though 212 matches DO show up. However the newest of those messages is 8/2/12, nearly a year ago

The message is apparently indexed; instant search DOES see it when

- The first term is shortened from "us.pwc" to "us"

- The second term is removed (OR detail)

- "through" is replaced by subject:"through"

In all three of these, literally hundreds of messages since 8/2/12 appear.

changing "through" to subject:"through" fails though if I simlutaneously change "us.pwc" to "us."

changing "us.pwc" to "us." fails to correct it.

Reversing the 2nd term(OR detail) with the 3rd term("through") fails to correct it.

Is there some quirk or limitation causing this [IMO quite serious] failure, which *silently* misreports results?

This is such an obscure appearing bug that I don't know if it can be reproduced by others analogizing with what I've shown.

I'm just going to try spitballing in order to aid someone reproducing the behavior. For all I know, it may be a waste of time in case it's just index corruption. But in case there's a chance, I'll add:

"john.smith@us.pwc.com" is in some of the messages as From:

The word detailed is in some of the subjects.

As mentioned before, the word through is in numerous messages

There are tens of thousands of messages.

I only have a couple of tiny file folders indexed in addition to my Inbox/Sent/etc..

The indexing status says 38,000 done; 0 left to index.

Note, a month ago I had my corporate helpdesk reinstall WDS, since there appears to be no earthly way provided to regenerate the WDS index (and I'm not an admin on this machine so it's even worse). I would hope that doesn't matter, since varying the search terms causes messages containing "through" in their subject to appear and disappear.

Finally, the only use of quotation marks above is to serve as an operator for search terms. They don't appear in my inbox subjects.

If Microsoft actually were to respond to bug reports I'd have them identify the culprit and at minimum, at MINIMUM, kick up a message box indicating the the results are not complete or accurate. (Of course we'll likely have to wait for new versions for fixes such as this, as simple corrections of Microsoft programming errors/omissions are usually refused for any version but the next or newest - pox on them for that.)

As bumbling inept as Microsoft is with almost everything they touch, I have to admit that instant search is actually extremely competent. Yes, I said it: Microsoft performs competently here. It really actually works. It really works fast. And the richness of search term parsing is more Googlish than Microsoftish. I have discovered many "secret" capabilities there, some of which might not even be documented (or if so, obscurely): OR; double quote processing; beginning of term matching; beginning of exact phrase matching (with closing quote optional) keywords like hasattachment; etc. . It's a remarkable tool; tremendously productive, contrary to most of Office and operating systems; and I've always assumed reliable. Doooh! Help!

TIA for the effort, and even for the time reading this long post. I have great regard for the contributors here.

OL07 / XPSP2 / Microsoft network / network Exchange Server

 

[GoodManStuckNMicrosfWorld]

Hmm, noting the date above on the bad search results, and checking drive C:, outlook.bak is dated 8/20/12. outlook.log is too, as it tells the story of running Inbox Repair Tool (on 8/20/12, I'm left to guess). Everything else in C:\Documents and Settings\MyNameHere\Local Settings\Application Data\Microsoft\Outlook (except for a little file extend.dat) is stamped within the past 4 days. FWIW.

 

[Diane Poremsky]

Did you run scanpst on the data file on that date? That is typically the reason for bak files. The log should give you a clue if you can't recall.

Have you rebuilt the search index at all?

 

[GoodManStuckNMicrosfWorld]

Thank you for answering!

Did you run scanpst on the data file on that date?

Most likely.

Have you rebuilt the search index at all?

That's the big question. As basic as that is, I've unclear on how to do it. Control Panel/indexing options/advanced looks promising though? (XP SP3)

There's also maybe

Outlook > Tools > Instant Search > Search Options > Untick data files

Outlook > Tools > Instant Search > Search Options > Tick data files which you want indexed ?

I gather that instant search uses Windows Desktop Search (WDS) (3.01 I believe). I have the magnifying glass icon in the XP taskbar which appears to be the same as the Control Panel item I described above. AAR that indicates an index location of

C:\Documents and Settings\All Users.WINDOWS\Application Data\Microsoft

I did a search for every file or folder in that subsystem and the newest filedate is 28 days ago (and that's drwatsn32.log !). Maybe Instant Search doesn't strictly using WDS - and now I wonder if it's using it at all. I regularly get search results more recent than 28 days ago. But the speed is so fast that it really appears to be fairly instant. (OTOH maybe file dates are not relevant for this process?)

Weirdness abounds. I just closed and reopened OL and 2000 very old emails just became Junk upon reopening.

Quite a mess here. I know how conscientious you are but this may be one to just let go of since there seems to be corruption in play. Any insight, from any lurker, even restating the obvious based on what I've presented, would help so I can less cluelessly communicate to company HelpDesk.

One more clue in case there's any salvation: a week ago I received 10s of megs in one day and at 2AM got a message from "Microsoft Outlook" with subject "Your mailbox is almost full." At 2AM this morning I got another. FWIW. Maybe this caused the junk-ifying above.

Another clue as to what my setup is: sometimes I see two processes running

searchprotocolhost.exe and searchfilterhost.exe

even when Outlook is closed. They come and go. Sometimes they're not running at all.

Anyway I tried clicking rebuild from control panel. BTW this works without requiring [XP] admin level. It says "Indexing in progress" and 39,777 items indexed even though I just started it. The same 2 files are still chugging, though I see a lot of cycles going to process searchindexer.exe, which appears to be one of the first things XP loads at boot time. (5 minutes later: I note that searchprotocolhost.exe and searchfilterhost.exe and not running. searchindexer.exe has not gotten cycles for at least a minute or 2. Indexing status says 39,972 done, 0 to go. Clearly, no way did it truly rebuild the index.)

You've got me to at least do this. Thanks. But I'm not expecting miracles, and there's a lot of noise and volume in my posts here. I feel bad about imposing that, but figured that as much data as possible gives a better chance of revealing something...

 

[GoodManStuckNMicrosfWorld]

A few minutes later, trying a new search, there's a yellow balloon "Search results may be incomplete because items are still being indexed. Click here for more details." Clicking gives a little "popup" "Outlook is currently indexing your items.

Search results may be incomplete because items are still being indexed.

1,230 items remaining across all open mailboxes"

I guess it's confusing as to whether WDS is indexing, or Outlook is "indexing" meaning synchronizing with WDS indexing?

The search terms atop my original post still give no results past 8/2/2012.

Maybe scanpst.exe is in order. Argggggh.

One more piece of info, recognizing that I've never been able to understand the difference between .PST and .OST usage (and even this work machine appears to have used .PST in the past):

C:\Documents and Settings\MyNameHere\Local Settings\Application Data\Microsoft\Outlook\outlook.ost

2,320,122,880 4/9/2013

 

[Diane Poremsky]

One more clue in case there's any salvation: a week ago I received 10s of megs in one day and at 2AM got a message from "Microsoft Outlook" with subject "Your mailbox is almost full." At 2AM this morning I got another. FWIW. Maybe this caused the junk-ifying above.

This means your Exchange mailbox is full. You need to empty deleted items & junk email or delete large messages.

One more piece of info, recognizing that I've never been able to understand the difference between .PST and .OST usage (and even this work machine appears to have used .PST in the past):
C:\Documents and Settings\MyNameHere\Local Settings\Application Data\Microsoft\Outlook\outlook.ost
2,320,122,880 4/9/2013

pst is a personal folders file, typically used by POP3 accounts or to archive mail. Mail is typically downloaded to the pst and removed from the server.




An ost is an offline file used with Exchange server. A copy of the mailbox is synced to Outlook and cached so you can work offline. Everything in your outlook is also in the mailbox online.

Another clue as to what my setup is: sometimes I see two processes running
searchprotocolhost.exe and searchfilterhost.exe
even when Outlook is closed. They come and go. Sometimes they're not running at all.

That is the windows search - its normal to see them run for a few seconds to minutes then stop.

I guess it's confusing as to whether WDS is indexing, or Outlook is "indexing" meaning synchronizing with WDS indexing?

WDS does the indexing and in Outlook 2010 and older, on Windows 7 and XP, there is one index file for everything. WDS indexes, when you do a search in outlook, it looks in the index.

Weirdness abounds. I just closed and reopened OL and 2000 very old emails just became Junk upon reopening.

That is not uncommon with Exchange accounts when you use a new data file - the junk filter runs on everything that is downloaded. It annoys me to no end with my Exchange accounts. I'm constantly making new profiles for testing and need to check the junk folder every time. It should only happen with new profiles, new ost files, or if you clear offline data and resync (right click on a folder, choose Properties - there is a button to clear offline items).

 

[GoodManStuckNMicrosfWorld]

Wow, thank you, lots of useful and specific answers.

I haven't changed anything in months, though I did yesterday run the command

outlook.exe whileout.pst

I wonder if the "full mailbox" caused the junk reclassification. The notification was 2:00 AM; Outlook was up continuously (as it is sometimes for weeks); I closed Outlook (in contemplation of reindexing while Outlook closed); I opened Outlook, and hello Junk.

As to paring I am going folder by folder (I made more besides inbox, sent, etc.) in reverse size sort, since the only way to get everything on the screen at once is Search, and right now I don't trust it!!

As to that single index file, I haven't located it. MS says it's at C:\Documents and Settings\All Users.WINDOWS\Application Data\Microsoft but I don't see it.

Though I have TMK had no data issues or disk read faults, I'm trying CHKDSK and then SCANPST tonight. I assume SCANPST works on .OST.

Thanks very very much for your comments and advice!

 

[Diane Poremsky]

If scanpst won't work on it, I think outlook 2007 includes scanost. Or, just delete the ost file and let outlook rebuild it.

Do you have Show hidden files and folders enabled? Tools, Options, View tab in Windows Explorer.

 

[GoodManStuckNMicrosfWorld]

If scanpst won't work on it, I think outlook 2007 includes scanost. Or, just delete the ost file and let outlook rebuild it.

There is SCANOST in \office12\ and actually now I remember running that about 6-8 months ago. I'll try it overnight.

Do you have Show hidden files and folders enabled? Tools, Options, View tab in Windows Explorer.

I looked at everything in that tree.

Thanks again!

 

[GoodManStuckNMicrosfWorld]

I ran scanost; no errors shown in log. However instant search term

"us.pwc" OR detail OR "through"

still only returns results from 8/2012 and earlier (and is virtually immediate); yet

"us.pwc" OR detail OR subject:"through"

takes a couple of seconds and *does* correctly find recent messages.

Since rebuilding through Control Panel/indexing options/advanced is fruitless (I just tried again) this appears to be either a bug or I have some other index corruption.

If anyone sees this and reproduces this type of issue please post as I will monitor the thread permanently unless it becomes "closed."

Thank you Diane for your many selfless contributions and your skill and competence in providing guidance on this and everything else. There are other places on the web where one may be rolling the dice following their recommendations but your diligence and accuracy is unparalleled.

 

[Diane Poremsky]

Thanks.

One more thought - are you using cached mode? classic online mode uses the server index for searches.

 

[GoodManStuckNMicrosfWorld]

Thanks.

One more thought - are you using cached mode? classic online mode uses the server index for searches.

Yes. » Use Outlook cached mode or not? Johan's Web Portal seems very helpful understanding that topic. I gather that having an actively updating .OST (evidenced from file date/time) implies offline mode; AAR the checkboxes are on under Accoutn setting/change/etc.

I may have lucked onto something. Anyone having similar issues using Exchange Server 2010 SP2 should consider Multi-Mailbox Search feature returns incorrect results when you perform a complex discovery search in an Exchange Server 2010 environment . Crossing my fingers for a miracle from that since I've struck out otherwise.

 

[GoodManStuckNMicrosfWorld]

This is so frustrating. Right now, as has happened before, "Search results may be incomplete because items are still being indexed. Click here for more details." **(below) Clicking says "146 items remaining across all open mailboxes" and not budging (CPU 98% idle). Indexing status (from the magnifying glass icon in the taskbar) says idle with 38,000 indexed, none left. And when I delete messages from search results, they don't disappear in a second or two as they should; they're sitting there without budging (for past 15 minutes, at least). searchindexer.exe is getting cycles here and there. The only notable thing I know of is a power outage 2 days ago, in case that spurred the revival of this particular malady.

BTW there's nothing exotic going on here AFAIK, like a process that monitors the stock market and sends emails on triggers or such.

If they'd just provide a way to delete the index allowing rebuilding at least I could proceed somewhat knowledgeably. Microsoft product designers are so pathetically clueless, stunningly so.

Diane you've been very helpful but despite your admirable persistence and conscientiousness don't feel obligated to tear further into this. Conventional answers seem ineffective in this case; maybe someone with a breakthrough will stumble onto this.

** (Note, this is a single word search, not a compound one.)

 

[GoodManStuckNMicrosfWorld]

Beating this horse further in case it's revealing: 15 minutes later, the deleted items finally fell off the search but now there are "158 items remaining" instead of 146, though only 1 email has arrived in past 20 minutes and I'll I've done is read and delete. How could the number of items remaining grow (maybe it's the deleted items).

 

[Diane Poremsky]

The index is under C:\ProgramData\Microsoft (or a similar location in xp - this is the windows 7/8 location). Find and delete the edb file.