You are not logged in.

#1 2007-11-08 12:21:31

DonVla
Member
From: Bonn, Germany
Registered: 2007-06-07
Posts: 997

beagle, tracker, etc

hello,

yesterday i found some time time to play around with metadata search engines, and i have to say, though i like how they work, they are barerly useable. i heard tracker should be the "best" search engine, i've tried to install it, but 0.6.3 doesn't work with new sqlite3 version. i downgraded sqlite3, but still doesn't work. strigi didn't compile at all. finally i got kdemod-beagle/ kerry working.
after about 7 hours or more (!!!) indexing (45GB mp3 files and 5-10 GB rest, including /var/lib/pacman with many small files) i finally tried it. i was impressed. the search results were good and fast. but, now the annyoing part, the beagled uses almost all the time about 15% cpu and the index daemon uses 30%. on a 1.2ghz celeron the system gets quite sluggish. that doesn't make much sense only for a search helper.
however, has anybody tested the other apps? is tracker better (uses less resources) than beagle?
any other recommendations?

vlad

Offline

#2 2007-11-08 12:57:38

buttons
Member
From: NJ, USA
Registered: 2007-08-04
Posts: 620

Re: beagle, tracker, etc

Tracker is the only one worth using, and I found that it crashed every other time it ran.  Eventually I gave up.

You can try the google desktop search for linux, however you should understand that besides consuming vast resources (a la beagle), it will also take a while to index.  By that, I mean days.

When tracker works, it's pretty nice, though.
EDIT: And yes, consumes almost no resources.

I just make do with find and locate these days.

Last edited by buttons (2007-11-08 12:58:07)


Cthulhu For President!

Offline

#3 2007-11-08 15:54:22

tjb0647
Member
From: Kongsberg, Norway
Registered: 2007-11-05
Posts: 4

Re: beagle, tracker, etc

I was able to install the kdemod-beagle/kerry search engine, after making the mistake to index the root directory...effectively locking me out of my normal user since the home directory was full after leaving it running over night, I regrouped and took a look at the documentation. 

I have since only indexed the locations of interest:
/bin
/etc
/opt
/usr
/var
/home/[usr]
and two NFS drives consisting of photos, videos, mp3s totally over 500 GB...

After a few hours I was able to find photos, videos, programs, files quickly and efficiently.  The downside of beagle is it does require approximately 5-10% of the space for its own index, but as of today (2 days running solid) my .beagle folder is 983 MB and you do have to clean out the logs (can be down with cron though).

I did try beagle in Ubuntu 7.10 and was not happy, it crashed and consumed to much cpu to be "crawling".  But with Arch and the kdemod, I could not be happier.

btw... my cpu history while writing I had an average of 3% for core 1 and 5% for core 2 on a Pentium D 3.00 GHz.

Offline

#4 2007-11-09 16:35:22

tjb0647
Member
From: Kongsberg, Norway
Registered: 2007-11-05
Posts: 4

Re: beagle, tracker, etc

Ok... well I screwed myself with my previous post.  Got home from work and my /home partition was FULL!  Looks like it is time to try something else out.

Offline

#5 2007-11-10 04:01:38

sinister99
Member
Registered: 2007-04-10
Posts: 136

Re: beagle, tracker, etc

For beagle, I found I definitely needed a daily cron job to delete all the GIGs of logs it created.  Got rid of it, waste of resources.

I'm now using recoll (in AUR).  Seems to be the only one I could get working, and works well enough, though it does leave a few things to be desired...

Offline

#6 2007-11-10 05:09:48

bender02
Member
From: UK
Registered: 2007-02-04
Posts: 1,328

Re: beagle, tracker, etc

Well, I've been using recoll for some time now, and I like it. Didn't use it for indexing mp3's and pictures; only pdf's and textfiles, and for that it works great. It's in the community.

Offline

#7 2007-11-10 06:11:44

mintcoffee
Member
From: Waterloo, ON
Registered: 2007-10-05
Posts: 120
Website

Re: beagle, tracker, etc

sinister99 wrote:

For beagle, I found I definitely needed a daily cron job to delete all the GIGs of logs it created.  Got rid of it, waste of resources.

I'm now using recoll (in AUR).  Seems to be the only one I could get working, and works well enough, though it does leave a few things to be desired...

Funny how you switched from beagle to recoll, a full-text search. Recoll consumes an insane amount for its index. Beagle works okay for me, the key is to index only your document directories. Basically, nothing outside of your home directory.

Last edited by mintcoffee (2007-11-10 06:12:04)


Arch on a Thinkpad T400s

Offline

#8 2007-11-10 14:22:35

sinister99
Member
Registered: 2007-04-10
Posts: 136

Re: beagle, tracker, etc

mintcoffee wrote:
sinister99 wrote:

For beagle, I found I definitely needed a daily cron job to delete all the GIGs of logs it created.  Got rid of it, waste of resources.

I'm now using recoll (in AUR).  Seems to be the only one I could get working, and works well enough, though it does leave a few things to be desired...

Funny how you switched from beagle to recoll, a full-text search. Recoll consumes an insane amount for its index. Beagle works okay for me, the key is to index only your document directories. Basically, nothing outside of your home directory.

I'm indexing ~750GB of docs, mp3s, and misc files, and my index is 950mb.  The index alone was larger with beagle, not including the daily 2-50 (not exaggerating, once filled my home directory) gigs of useless logs.

Beagle is also a full-text search

Last edited by sinister99 (2007-11-10 14:25:44)

Offline

#9 2007-11-11 20:19:42

DonVla
Member
From: Bonn, Germany
Registered: 2007-06-07
Posts: 997

Re: beagle, tracker, etc

thanks bender02,

recoll was a good idea. i've installed it yesterday and it's pretty good. indexing took only about 1,5 hours for the same amount of data beagle needed 7. reindexing can be run as cronjob, that's also great - no annoying daemon. and another cool feature is that you can have multiple databases for different directories which can be reindexed at different times.
indexing of ogg/ flac and image files didn't work out of the box. you need following filters:
for image files you need this: http://www.lesbonscomptes.com/recoll/filters/rclimg ,
for ogg files thsi script: http://pastebin.archlinux.org/18973 ,
for flac files this script: http://pastebin.archlinux.org/18974.
just place this scripts as rclogg, rclflac under /usr/share/recoll/filters or the filter directory specified in recoll.conf.
actually i haven't found this scripts on the recoll but on the linuxquestion homepage (http://www.linuxquestions.org/questions … page3.html) as a patch for a slackware package.
however, they work great.

vlad

Offline

Board footer

Powered by FluxBB