Re: [Tracker] trackerd's disk access slows down system



On Mon, 2007-08-20 at 10:41 +0200, Tassilo Horn wrote:
Hi all,

I use tracker (0.6.1) on my laptop system and I really like it.  But one
thing that annoys me sometimes is that tracker can slow down my system
considerably if there are a lot of files that need indexing.

For example I like it to index the source code of some projects I'm
involved, where a huge part of it is auto-generated from some schema
definitions.  So whenever the generator ran about 1000 files have to be
indexed anew.

Till that is done I hardly can start applications or do other stuff that
needs disk access.

Is it possible to tell trackerd to always let other applications have
precedence in disk access?  Or would another IO scheduler attenuate my
problems a bit?  Currently I use the CFQ scheduler.

I think that would be a very useful feature for anyone with a system
where the slow disk is the bottleneck.  I think that's the case for
nearly every notebook.

Thanks in advance,

we are working on improving this dramatically by replacing our current
indexer backend (qdbm) with the latest sqlite. QDBM suffers from massive
fragmentation when indexing many GBs of text/source code as we only
optimise after initial indexing has finished (we really should optimize
the index more frequently but sqlite is non-fragmenting so wont bloat up
like qdbm does nor will it require any optimizing)

In the meantime, for best results it is wise to use our crawl directory
facility with source code such that it indexes them without slowing down
your compiling or check out









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