Re: Long Standing Problem...

On Aug 1, 2011, at 4:27 PM, Paul Fan wrote:

>> the mds process dies with a variety of ugly errors, access violation, assert
>> errors and
>> a variety of other excuses…
> 
> The following works for us whenever we have an 
> out-of-control mds/mdworker process.
> 
> As root, do 
> 
> launchctl unload -w /System/Library/LaunchDaemons/com.apple.metadata.mds.plist
>   (may take a few seconds to complete)
> 
> Remove all user folders in
>  /var/tmp/mds  if any.
> 
> Remove all folders in  
>  /var/db/Spotlight-V100/Store-V1/Stores/
> 
> Wait a few seconds and do
> launchctl load -w /System/Library/LaunchDaemons/com.apple.metadata.mds.plist
> 
> Spotlight will now be very busy reindexing everything
> for a while.


How does this differ from the result you get from:

mdutil -E /


Actually on my Snow Leopard system I'm not seeing a /var/db/Spotlight* directory at all, nor a /var/tmp/mds folder.

mdutil -i off /	will disable indexing without shutting down the daemon but maybe restarting the daemon is what you're after with the above.

It might be redundant to also remove /.Spotlight-V100 in addition to mdutil -E, I'm not sure. But if I were to do that I would either disable indexing with -i or I'd kill the daemon with launchctl. I'm sure it doesn't like the index just vanishing.

Chris