Greetings,

Firstly, I am not sure if I am posting this in the right place. If not, please move this topic appropriately.

I am experiencing a lot of cpu/memory usage by baloo_file_extractor (/usr/lib/baloo_file_extractor).
After killing it with

Code: Select all

kill `pgrep baloo_file`

the resources are back to normal and my PC is much quieter. CPU temps drop by about 10-15 C.

I read that baloo_file_extractor is a file indexing tool. I have disabled file indexing in System Settings > File Search and restarted. It still runs just as much as before.

So what do /usr/lib/baloo_file{,_extractor} actually do?


I am running the plasma version of Manjaro. I am fully up to date with the following information from kinfo:
Operating System: Manjaro Linux
KDE Plasma Version: 5.21.5
KDE Frameworks Version: 5.82.0
Qt Version: 5.15.2
Kernel Version: 5.10.36-2-MANJARO
OS Type: 64-bit
Graphics Platform: X11
Processors: 24 × AMD Ryzen 9 3900X 12-Core Processor
Memory: 62.8 GiB of RAM
Graphics Processor: GeForce RTX 2070 SUPER/PCIe/SSE2

Please let me know if you need any further information to help track down this issue.

Thank you in advance for your help/advice!

So what do /usr/lib/baloo_file{,_extractor} actually do?

baloo_file is the process that gets told of changes to your files - it asks inotify to let it know - and records the filename and external attributes to the index (for those folders you've asked it to index, that is...)

If you've selected "content indexing", that is, you want to be able to search for things based on text they contain, baloo_file manages a queue of files that need to be "fully indexed" and feeds these files, a bunch at a time, to the baloo_file_extractor process. Baloo_file_extractor reads each file and writes info about "what words" are "used where" back into the index. That can be a major job if (re)indexing a big file.

Please let me know if you need any further information to help track down this issue.

See if you can see what is happening....

If you run "balooctl status", this should show you how many files baloo is keeping track of plus the number it thinks it still needs to index. A "balooctl indexSize' shows the amount of data in the index plus the size of the index file. Both useful things to know.

If you run "balooctl monitor" in a window, this should list the files "as they are indexed". Do you see files being indexed? Do the same files reappear? (repeatedly, or get listed every time you log on?). Do you have *loads* of files to index?

What can go wrong?

Historically, quite a lot of things If you indexing a log file, every time something is appended to it, it is reindexed. There are files where the indexing fails and you find yourself reindexing them every time you log on. If you are writing crash dumps somewhere you are indexing and baloo crashes when indexing the dump, then you've got real trouble (yes, apparently that happened). It can be a gradual, step by step process to work out the root cause.

The "Big Red Button", as in delete all and start again, is "balooctl purge". You can run this in one window and "balooctl monitor" in another to watch the reindexing. Of course, if you have a load of files, this can take some time.

... killing baloo_file ... the resources are back to normal and my PC is much quieter. CPU temps drop by about 10-15 C.

A bit curious as baloo_file_extractor "would be" a single process, running on a single core. It shouldn't make such an impact on a system of your capabilities. What does a process monitor (htop or whatever...) say?

I read that baloo_file_extractor is a file indexing tool. I have disabled file indexing in System Settings > File Search and restarted. It still runs just as much as before.
If you look at the configuration file

Code: Select all

~/.config/baloofilerc

You should see a line:

Code: Select all

Indexing-Enabled=false

If you've got that, then baloo should not start.

Could also be worth having a look at: https://community.kde.org/Baloo/Debugging