Re: [Tracker] Next version in Cvs - needs testing



Laurent Aguerreche wrote:
Le dimanche 30 juillet 2006 Ã 21:50 +0100, Jamie McCracken a Ãcrit :
Laurent Aguerreche wrote:

As you can see, on the first line it worked! Before, I just remove my
~/.Tracker directory because I had some database corruptions (kill
trackerd with "kill -9" seems to be a bad idea!).
they are not database corruption. If tables are left open and you kill the process they are marked as crash and will be automatically checked (and repaired if necessary) when you restart tracker.

Hum... okay. It's scary to see that!

060730 23:27:02 [ERROR] mysql_embedded: Table
'/home/laurent/.Tracker/data/mysql/proc' is marked as crashed and should
be repaired
060730 23:27:02 [Warning] Checking table:
'/home/laurent/.Tracker/data//mysql/proc'
060730 23:27:02 [ERROR] mysql_embedded: Table
'/home/laurent/.Tracker/data/tracker/Services' is marked as crashed and
should be repaired
060730 23:27:02 [Warning] Checking table:
'/home/laurent/.Tracker/data//tracker/Services'



these are all harmless warnings

You will see - Database corruption detected - repairing table x if there was any corruption (mysql is self healing so it should auto repair any)


A stupid question: is it possible to see the table contents? And how?

yes if you have mysql server installed and you copy the db files to its data directory then you can use the mysql cli program to do sql queries.



In tracker.log, I have
at the last line:

30 Jul 2006, 21:30:34:135 - Executing search with params Files, hello.
512, 0, 030 Jul 2006, 21:30:34:167 - search returned no results

So it only received my first search request...
I cannot replicate them Im afraid. Please check you do not have two versions of tracker (one in /usr/bin and one in /usr/local/bin or something use "whereis trackerd" and "whereis tracker-search")


I do not have duplicate versions installed.

But more fun: I kill trackerd and then launched it again. Trackerd
restarted to indexing files (why it has stop its work before?!) and
tracker-search seems to work now. Some mutex problems? Is it possible
that a research blocks indexing and then next researches? Is it possible
that indexing stops on a file because is too big, special, etc. and that
this blocking keeps trackerd in a state where it isn't able to treat
DBus messages (so researches)?


no tracker is fully multithreaded so there should never be any significant blocking. Its forbidden to do any heavy activity in the main thread and all the hard work is done in separate threads.

if you deleted ~/.Tracker then you have deleted the database and it will rebuild it (and reindex everything).

You should never have to delete ~/.Tracker




--
Mr Jamie McCracken
http://jamiemcc.livejournal.com/




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]