Indexed Search with BTRFS (not Baloo)

General system Newbie
1

I am sorry to keep bothering you.
We know there is a bug with Baloo under BTRFS. It reindexes all my data, PDFs with every snapshot. So the laptop is always busy indexing files and index growing forever. Searching for a single unique file in Dolphin it returns several results of this single and unique file (multiple snapshots)

I am on KDE Plasma, Dolphin, Baloo.

I wonder if there is an indexed search tool that plays well with BTRFS? (preferably unde KDE Plasma).

P.S: I do not mind considering some other DE though I love Plasma. But I may consider LXQT, Gnome, MATE, XFCE, Cinnamon if I can do desktop search (preferably in the file manager) under BTRFS.

2

Why would you need indexed search in this day and age at all?

honka_animated-128px-36

I don’t use Baloo and search is pretty damn fast without any index, regardless of filesystem…

For example, searching 3.6 TiB hdd drive full of music (583 115 files) for all .wav files recursively via dolphin’s search without any indexing takes about 12 seconds on my system.

1 Like
3

I have a lot of documents, other than books (PDF), legal, invoices, warranty… lots actually.

Well, I am still alive, I still have a lot of stuff to do and follow up and I do research as well.

How is that? Fast search file content? Let me try Kfind (Ooops, not installed! What you recommend to install?)

4

Oh wait, indexed search means file contents?
I always thought it’s just mechanism to quickly search for files themselves by caching them in some db, which was useful on old hardware back in day…

But if you’re correct - then it’s whole another story.

6

Yes… unfortunately non indexed search is too slow.
I don’t mind a bit slow though.
What you recommend I try (from those in Dolphin… Kfind? Fsearch? they are not indexed… right?)

1 Like
7

Yeah you’re right, i just had different concept of indexed search before https://community.kde.org/Baloo

Not an expert, since i never had a need to do such task, maybe @dalto will have some ideas :upside_down_face:

8

How many documents are we talking about and what formats are they in? Searching file contents without an index might be very slow without an index. Especially if you have a large number of PDFs requiring searching.

9

@dalto
I know I am irritating you. I ask too much!

10

A few hundreds at the moment and mostly PDF. and another .txt set of files (and html) from Zim Notes)

Let us assume +/ 1000 to 1200!

Just curious, if installed Kfind, would it be using Baloo index? Would it create its own index? Would Kfind have problems with BTRFS?

UPDATE:
Just installed Kfind, even restricted search to PDF. Horribly slow to search content! Won’t fit my use!

11

Indexing a whole library of pdf files and others of course can take “forever”, regardless of which one you choose, me thinks.

I’d be interested to learn which crawler would be a best solution to this problem, as well!

12

You could try recoll from the repos. It is a file indexing tool.

I did some testing with it and seems quite fast once the index is complete.

I think he is referring to search time, not index build time. kfind doesn’t use an index I don’t think.

3 Likes
13

No problems with BTRFS? (I hope)
I tried it once and it was OK.
Just want to be sure it will not have the same problem with BTRFS as Baloo.

14

I can’t guarantee it but those problems are pretty specific to the way baloo tracks devices.

If you decide to try it, be sure to check the optional dependencies and ensure you have the ones you want for indexing various formats.

1 Like
15

So kfind could be better when it comes to search in-document keywords? Sorry, I don’t know how to ask ‘better’…

16

Well, I when I start I make sure I start before bed time.
When I wake up in the morning it would be done.

17

I just installed and tried it. It is horribly slow to search inside files.

18

kfind is searching, not indexing. So if you try to do a content search on a large library it is essentially doing the same thing an indexer would do for each search.

I don’t think that is the intended use case for kfind.

1 Like
19

AFAIK Baloo defaults to searching your home folder… This is configurable in KDE System Settings. If you have a folder like /home/$USER/snapshots you can exclude that folder or manually specify folders like ~/Desktop, ~/Downloads.

20

mmm… thinking… so probably it will work? How can we know?

OK… If I converted my disk to BTRFS, can I revert back to EXT4 in case needed?
I would better convert and revert rather than fresh install.
I can fresh install IFF (if and only if it is confirmed it will work)
Hopefully someone using BTRFS + Recoll (or any other) give us a clue.
Maybe someone with BTRFS might be interested to try Recoll and report back. Uninstall it if it didn’t work as expected!

21

I feel like I have mentioned this in several topics already but the issue isn’t caused by indexing snapshots. It is a design flaw in baloo.