Spotlight not working correctly

I have been wondering if I could get around the problem by switching to mountain lion, however my son's Macbook Pro seems to have a similar problem with Time Machine. I left it all night, and it was still preparing for backup. Unfortunately Spotless didn't work, so I couldn't see if it was related to Spotlight.

I am in a bit of a bind now, because I have Spotlight turned off so I can back things up, but that means that I can't find anything in the Finder, or even in Mail. Turning Spotlight back on is hopeless, because it never finishes, and every time you turn it off, it starts again from the beginning.

So any suggestions most welcome!

I screwed around with this all weekend. After installing 10.7.5 over 10.7.4 on both my Mac Pro (1,1) and MacBook Pro (2,2) it caused Spotlight to have ridiculous indexing times and Time Machine issues. The only "solution" that worked was downgrading to 10.7.4.

I happened to save a copy of the 10.7.4 install dmg to a USB drive when I originally installed 10.7.4 on my Mac Pro. I'm not sure how it would work on my MacBook Pro though. I haven't tried it yet. But constantly indexing or indexing off makes using Mail.app miserable.

I'm once again re-installing 10.7.5 over 10.7.5 on my MacBook Pro. But I doubt it will help.

Anyone have experience using the MacPro install dmg on a MacBook Pro?

This article from Apple indicates that installing 10.7.4 from the Mac Pro dmg on to the MacBook Pro would probably not be successful.

http://support.apple.com/kb/HT4718?viewlocale=en_US&locale=en_US

Also had much drama with iPad 2 and iOS 6. App store would download updates, just get stuck on installing then it would pop-up and dismiss errors. After a third restore it is working correctly (for now?).

Apple really has done a horrible job with 10.7.4 and iOS 6. Very poor quality control. iOS 6 I'm a little more forgivable of but in a "10.x.5" release? Come on. There is no excuse for this ridiculous bug this far into the development cycle.

Forum rant: They might want to update the spell check dictionary on their (um.. l a m e ) 3rd party forum software. iOS and MacBook Pro aren't in it. Neither are iPhone or iPad. A remember me on same page login wouldn't hurt. And yes, I have made these suggestions via forum feedback to no avail.

Good luck to everyone dealing with this drama.

Apple really has done a horrible job with 10.7.4 and iOS 6. Very poor quality control. iOS 6 I'm a little more forgivable of but in a "10.x.5" release? Come on. There is no excuse for this ridiculous bug this far into the development cycle.

But, if they didn't release it when they did, the toy users would have rioted, since NOT delaying iPhone 5's release is what it was all about, not to mention that the market would have lowered Apple's share price. 😀

FYI update on my Spotlight drama. Just spent 1.5 hours on the phone with a Senior Apple Care tech today. (FYI he claimed to know nothing about this issue.) We performed all of the tricks I'd already read here in the forum pages...everything from making HD private then not-private to all the various Terminal commands. Nothing seemed to help. Finally he emailed me a "Capture Data" program. I ran the Capture Data program and email him the results. He told me that would be sent to the engineers and he would call me back in 1-3 days with some sort of solution/suggestion/advice. I have no solutions to offer but maybe I will when he calls back? If you have Apple Care, call them and ask about it. Ask them to send you the Capture Data program so they can determine what is causing this drama. And get them to send results to the engineers. Now, I realize not everyone has Apple Care. I get that. So as soon as the Apple Care guy calls me...I will report what I have learned. That's all I've got for now.

ThisApple discussion thread suggests that it might be a (recently) incompatible mdimporter plugin from a 3rd party application. This works in one person's case. Later in the thread I go through deleting various preference files, as well as temporarily relocated all 3rd party apps that have mdimporter plugins. Consistently I'm finding that mdworker is getting hung up on the sandbox cache, and cannot "talk to" lsboxd.

I might have a working "fix". Based on the constant stalling on sandbox-cache.db I decided to take a chance and delete all folders found in /var/folders using the command. What I actually did, in order:

Restored Evernote, Colloquy, LibreOffice back to /Applications (from the Trash).

Quit all applications except Terminal, issued the following two commands:

sudo mdutil -i off /

sudo mdutil -E /

sudo rm -Rf /var/folders/*

Reboot.

sudo mdutil -i on /

There were no further stalls on sandbox-cache.db according to opensnoop, and no further lsboxd complaints in Console. mds and mdworker processes have been very busy, typically 40-100% CPU, unlike before. And now the index estimate is variably saying 6 to 19 minutes for completion instead of 3 days to 5 weeks. So hopefully in an hour or so (the typical amount of time for my system to do a Spotlight index rebuild) I will know if this has really worked.

Now, blowing away the folders in /var/folders may have other consequences, possibly negative, for sandboxing. But the files therein appear to be recreated from scratch after reboot to no ill effect or complaints in Console. But let's just say, the suffering bleeding edgers may end up bleeding a lot more by trying this. Any volunteers?

Also posted this response in : Apple Support Communities > Mac OS & System Software > OS X Mountain Lion > Re: Spotlight forever indexing

Interesting results. I get to 10 seconds remaining in the indexing, with the status bar all the way to the end, and then it hangs!

The last items in Console are:

Oct 1 17:42:11 mingisapsycho mdworker[403]: ENSpotlightImporter:importFileAtPath:attributes:error: to find object id from path /Users/chris/Library/Containers/com.evernote.Evernote/Data/Library/CoreData/com .evernote.Evernote/6EE7073B-BB36-4AD6-8091-A15073CD960B/ENSyncState/_records/0/p 1.ennote

Oct 1 17:43:15 mingisapsycho com.apple.security.pboxd[544]: kCGErrorIllegalArgument: CGSGetWindowTags: Invalid window 0xab

Oct 1 17:43:15 mingisapsycho com.apple.security.pboxd[544]: kCGErrorFailure: Set a breakpoint @ CGErrorBreakpoint() to catch errors as they are logged.

Oct 1 17:43:15 mingisapsycho com.apple.security.pboxd[544]: kCGErrorIllegalArgument: CGSSetHideOnDeact: error getting window tags

Oct 1 17:43:15 mingisapsycho com.apple.security.pboxd[544]: kCGErrorIllegalArgument: CGSCopyWindowColorSpace: Invalid window 0xab

Oct 1 17:44:45 mingisapsycho com.apple.mdworker.pool.0[550]: PSSniffer error: Undefined error: 0

Oct 1 17:44:45 mingisapsycho com.apple.mdworker.pool.0[550]: PSSniffer error: No such file or directory

Oct 1 17:45:15: --- last message repeated 2 times ---

Oct 1 17:45:18 mingisapsycho com.apple.mdworker.pool.0[550]: I/O error : encoder error

Oct 1 17:49:31: --- last message repeated 3 times ---

Oct 1 17:49:31 mingisapsycho com.apple.mdworker.isolation.0[592]: Incorrect start/end range ordering; fixing.

Oct 1 17:50:18 mingisapsycho com.apple.mdworker.pool.0[582]: I/O error : encoder error

Oct 1 17:50:19 mingisapsycho mdworker[585]: Unable to talk to lsboxd

There are almost 100 message like the very first line about ENSpotlightImporter related to Evernote. I don't know if this is related to the failure but it happens right at the very end so it seems possible. 99.9% of the indexing took about 45 minutes. The last 10 seconds has been the status for 15 minutes, and hasn't changed yet.

OK about 20-25 minutes later it gave up and the indexing is finished. A remarkable improvement over 5 days or 5 weeks.

What I don't know is if this is actually fixed, or if the problem has now recurred, and the sandbox-cache.db is "corrupted" again. So what I've done for now is is place /Users/chris/Library/Containers in Spotlight > Privacy to keep it from being scanned. I'll use mdutil -E / to force another reindexing without changing anything else and see if this recovers at the end faster or not.

File a bug reportwith Apple. Be sure to include what you've found. Sandboxing, security, and importers are totally hosed, but heXX, they released iPhone 5 on time.

I know I'll sound like a complete moron, nevertheless -- what is Sandbox?

Same thing. Last 10 seconds, there's a hang that lasts for about 15 minutes. It does not exactly coincide with:

Oct 1 18:58:10 mingisapsycho com.apple.mdworker.isolation.0[797]: Incorrect start/end range ordering; fixing.

Oct 1 18:58:49 mingisapsycho com.apple.mdworker.pool.0[787]: I/O error : encoder error

Oct 1 18:58:50 mingisapsycho mdworker[790]: Unable to talk to lsboxd

But that's when the hangs begin. During the hang mds and mdworker aren't doing much of anything. mds uses 0.2% CPU and a sample report indicates it's mostly hung on libsystem_kernel.dylib. I think we've got a bug. I don't think in every instance this 10.7.5 and 10.8.2 Spotlight issue is about bad importers. I've eliminated that as a possibility, unless of course, one of Apple's own importers has gone bad.

Anyway, blowing away the folders in /var/folders, then rebooting, and then either 'mdutil -E /' or adding your Mac OS volume in privacy to prevent searches, then promptly removing it, appears to fix the problem. For me.

Previously, I was waiting 3 days for rebuilds that still said (variably) 5 days to 5 weeks to complete. Today it took ~45 minutes after doing the above.

Sample report for mds when "hung" 2602 Thread_737 DispatchQueue_1: com.apple.main-thread (serial)

+ 2602 ??? (in mds) load address 0x1097e5000 + 0x4444 [0x1097e9444]

+ 2602 ??? (in mds) load address 0x1097e5000 + 0xdbf9 [0x1097f2bf9]

+ 2602 CFRunLoopRunSpecific (in CoreFoundation) + 230 [0x7fff83db5486]

+ 2602 __CFRunLoopRun (in CoreFoundation) + 1204 [0x7fff83db5c74]

+ 2602 __CFRunLoopServiceMachPort (in CoreFoundation) + 188 [0x7fff83dad50c]

+ 2602 mach_msg (in libsystem_kernel.dylib) + 73 [0x7fff8d2b5d71]

+ 2602 mach_msg_trap (in libsystem_kernel.dylib) + 10 [0x7fff8d2b667a]

2602 Thread_797 DispatchQueue_2: com.apple.libdispatch-manager (serial)

+ 2602 _dispatch_mgr_thread (in libdispatch.dylib) + 54 [0x7fff82f3d316]

+ 2599 _dispatch_mgr_invoke (in libdispatch.dylib) + 923 [0x7fff82f3e786]

+ ! 2599 kevent (in libsystem_kernel.dylib) + 10 [0x7fff8d2b87e6]

+ 2 _dispatch_mgr_invoke (in libdispatch.dylib) + 926,923 [0x7fff82f3e789,0x7fff82f3e786]

+ 1 _dispatch_mgr_invoke (in libdispatch.dylib) + 931 [0x7fff82f3e78e]

+ 1 __error (in libsystem_c.dylib) + 0 [0x7fff89b6ac88]

2602 Thread_862

+ 2602 thread_start (in libsystem_c.dylib) + 13 [0x7fff89b19b75]

+ 2602 _pthread_start (in libsystem_c.dylib) + 335 [0x7fff89b168bf]

+ 2602 ??? (in mds) load address 0x1097e5000 + 0x962f [0x1097ee62f]

+ 2602 read (in libsystem_kernel.dylib) + 10 [0x7fff8d2b8af2]

2602 Thread_143919

+ 2602 start_wqthread (in libsystem_c.dylib) + 13 [0x7fff89b19b85]

+ 2602 _pthread_wqthread (in libsystem_c.dylib) + 758 [0x7fff89b18594]

+ 2602 __workq_kernreturn (in libsystem_kernel.dylib) + 10 [0x7fff8d2b8192]

2602 Thread_145921

+ 2602 start_wqthread (in libsystem_c.dylib) + 13 [0x7fff89b19b85]

+ 2602 _pthread_wqthread (in libsystem_c.dylib) + 758 [0x7fff89b18594]

+ 2602 __workq_kernreturn (in libsystem_kernel.dylib) + 10 [0x7fff8d2b8192]

2212 Thread_146823

+ 2212 start_wqthread (in libsystem_c.dylib) + 13 [0x7fff89b19b85]

+ 2211 _pthread_wqthread (in libsystem_c.dylib) + 758 [0x7fff89b18594]

+ ! 2211 __workq_kernreturn (in libsystem_kernel.dylib) + 10 [0x7fff8d2b8192]

+ 1 _pthread_wqthread (in libsystem_c.dylib) + 763 [0x7fff89b18599]

345 Thread_146872

345 start_wqthread (in libsystem_c.dylib) + 13 [0x7fff89b19b85]

345 _pthread_wqthread (in libsystem_c.dylib) + 758 [0x7fff89b18594]

345 __workq_kernreturn (in libsystem_kernel.dylib) + 10 [0x7fff8d2b8192]

Please let us know if anything else breaks. I would really like to get Spotlight working again, but I don't want to bleed too much 😉

Nothing else has broken. I don't have any sandbox or unusual security warnings in Console.

Is there anything else that uses the contents of /var/folders?

It looks like it's just for caching, so it would seem disposable. But anything actively using it probably won't like its cache files vanishing. I'm seeing references that iTunes uses it for downloading apps, not your downloads folder. Once the download is complete, it moves the apps into position (it may be an installer package that's being downloaded). I have previously downloaded one App Store app, months ago, and even though I deleted all /var/folders folders, that app still works.

However, what I notice is that this app is not up to date, and the App Store application is saying no updates are available. If I go to Purchases, it sees the apps I've purchased and there it indicates an update for one of the apps is available. I have no idea if the discrepancy between Purchases and Updates in the App Store app has anything to do with removing /var/folders or if it's a bug, or what.

So if I try to update this app I get a cryptic message saying "To update this application, sign in to the account you used to purchase it." But I've already done this. I only have one App Store account. If I sign out and back in, same message. Not sure if it's related.