File Indexing service (e.g.Recoll/Tracker)?

[victorhooi]

I'm curious if there's any interest in including a file-indexing service (like OSX's Spotlight) in FreeNAS, whether as part of the core, or as a plugin?

From what I've read, Tracker (https://wiki.gnome.org/Projects/Tracker) and Recoll (http://www.lesbonscomptes.com/recoll/) are the two most active indexing project on the Linux side.

I don't know about Tracker, but I believe Recoll also does work on FreeBSD:

http://portsmon.freebsd.org/portoverview.py?category=deskutils&portname=recoll

What are your thoughts?

 

[anodos]

FreeNAS doesn't have a file manager. You can always let your client machine index the shares on FreeNAS.

 

[cyberjock]

There is by some users, it just hasn't been implemented (assuming those even work).

 

[flanello]

Pherhaps DocFetcher (http://docfetcher.sourceforge.net/en/index.html) is something for you.
It is not an indexing service for FreeNAS, but an indexing software for the clients that can work with the shares. Especially the portable mode is useful. It is possible to place the software with its index on the share and use it from different clients.

 

[victorhooi]

Hi,

@anodos - Neither Tracker or Recoll have anything to do with a file-manager - they're a background indexing services.

I believe in both cases you could create the index on the server, and then export it to desktop clients.

Recoll also has a web-interface:

https://github.com/koniu/recoll-webui

@cyberjock - Oh, do you mean people have already tried doing it? =) That'd be pretty cool. Do you know anybody that might be able to share their experiences?

@flanello - Hmm, DocFetcher looks interesting, I'll look into it. However, I think you could do the same thing with either Tracker or Recoll as well - as in, have it installed locally, and point it to a network drive.

However, my idea was to have the indexing/search engine service run once on the remote server (reducing the network latency issues), and having it run once, rather than every user needing to index all the files all over again.

Regards,
Victor

 

[cyberjock]

What I meant is that there is interest for many people, but people haven't implemented a seamless windows indexer, yet.

The only two good options I know of (for varying values of "good")

1. Force the workstations to do indexing of the server themselves.
2. Use the FreeBSD index tool, but it requires the user to use the CLI, which most people will find non-obvious to use.


I've never seen this DocFetcher before, so I might have to give that a whirl just to see how good/bad it is.

 

[anodos]

Neither Tracker or Recoll have anything to do with a file-manager - they're a background indexing service.

I'm pretty sure that I read that they have gtk, qt4, x11 dependencies.

 

[victorhooi]

Hi,

Recoll is certainly able to run in a headless mode - e.g.:

http://www.freelists.org/post/recoll-user/Headless-setup-possible,1

Those packages are for the Recoll desktop application, which provides a UI to search the index. You can use the web GUI I linked above instead.

You can compile it without X/QT4 support, or you could just not run the X components.

Regards,
Victor

 

[flanello]

@victorhooi: I use docfetcher in the portable mode on external drives. The software and the index are on the external drives, thus I can use the searching on every pc (without generating the index). Probably the same is possible on a network drive:
The software and the index would be stored on the server. Thus every client starting the software uses the same index. Docfetcher seems to have an headless index updater, probably it can be run directly on FreeNAS in a jail or a virtual machine. But I haven't tested all this. Perhaps it can be a tool for home use, it is not a tool for a business use case with many clients.

The headless mode of recoll sounds interesting for a installation on a server.

 

[superwad]

Have you tried slocate? You'll need to set a crontab to keep the db up to date though, and I don't know if it does file contents. Or generally "find" should work, although it's not indexed.