HDD going crazy

I’ve just finished a fresh install and latest updates, and I can hear the HDD constantly thrashing, just that dull rumbling I know to be the hard drive writing furiously.

But it’s non-stop. What on earth could be the reason? Is there some sort of tool I can install to know what drives are being accessed and even what files?

The strange part is Endeavour OS is installed to my nvme SSD drive, which of course does not make noise. But I do have the home directory mapped to a partition on one of my regular HDD’s, and I have sym links to document folders also on my regular HDDs.

The best educated guess I can think of is search indexing.

Please let me know what info I need to give you guys to help find the answer to this.

Operating System: EndeavourOS 
KDE Plasma Version: 5.27.7
KDE Frameworks Version: 5.108.0
Qt Version: 5.15.10
Kernel Version: 6.4.9-arch1-1 (64-bit)
Graphics Platform: X11
Processors: 8 × Intel® Core™ i7-6700K CPU @ 4.00GHz
Memory: 23.4 GiB of RAM
Graphics Processor: NVIDIA GeForce RTX 2060 SUPER/PCIe/SSE2

Have you disabled baloo?

Edit: It’s in system settings under search.

sudo pacman -S iotop
sudo iotop -a

Then choose with Left / Right arrows DISK WRITE tab, so it would re-arrange table output with most writes on top

I sure see a whole lot of firefox in there, what’s it doing??
I also see sdb5 which is indeed my /home partition.

I found this when searching the “start menu” for “baloo”, is this what you mean?
And looks like it says 100%… BUT, the HDD thrashing noise has also stopped moments after I took this screenshot, did it know I was watching??

Yeah that’s what i though.
Firefox does that, let me try to suggest some stuff to reduce it heavily.

P.S. Although to be fair - it’s not that much it writes yet, at least on your screenshot.

Have a look at the following link for ways of disabling baloo:

https://wiki.archlinux.org/title/Baloo#Disabling_the_indexer

For more: https://wiki.archlinux.org/title/Baloo

1 Like

Well actually now that I spend more time looking at the output after the HDD has stopped making the noise, the numbers appear to be a total record of how much writing has happened over time, not how much writing is currently happening. And since I’m currently using firefox to watch YouTube, that’s probably not that surprising that it’s the only app showing activity.

I’m inclined to think it was the search engine indexing from the other post above…

For Firefox do that in about:config. then reboot it:

'user_pref("browser.cache.disk.enable", false);'
'user_pref("browser.sessionstore.interval", 1800000);'
1 Like

It’s definitely the offender, first thing i do is disable it on any new installation.

To disable do that:

balooctl disable
balooctl purge
balooctl suspend
balooctl status
2 Likes

Thank you, and thank you @ricklinux and @pebcak it looks like it was the indexing after all. I like the idea of indexing giving fast search results but it’s only indexed my home drive and ironically that’s not really where I keep all my files, I have sim-links in there to other drives where my real personal files are. Mostly what’s in there is program config files.

Why do you guys always disable it by default? Is it too impractical to allow it to index everything? In my experience, search indexes end up taking up quite a fair bit of space, perhaps that’s the reason?

I’m in the habit of doing (in Dolphin) “More Search Tools”, “KFind”, in the past I just assumed the built in search never worked - now I see it works fine on my home drive now that it’s finished indexing XD

It’s not giving you faster speed, indexing in that case means it will also be able to search through file contents…Which is a bit too much, why would anyone want that in exchange for heavy disk thrashing?

1 Like

Oh it’s indexing the contents?? I just confirmed this by searching one of the variable names in my code files and it showed up. I feel there may be other ramifications for this kind of convenience on the huge amount of files in my actual data drives. Like it taking days to build huge index buffers of several gigabytes. Lucky it was only my home drive it found. I’ll run those commands to purge and disable it.

Yep, and usual file search is perfectly fine and fast without it (especially on SSD)

1 Like

As discussion shows us it is most likely because of Baloo. It is normal to take sometime to index all files you have.

But there is really one very important question. Are you using BTRFS, Sanpper,… to take timely snapshots of system?

If this is the case, Baloo will never finish indexing as It will reindex each and every snapshot as a separate drive containing all the data. So, if you have a unique single file containing a unique single word, if you search for this specific word it will return to you multiple results all pointing to the same file.

I myself prefer to have my files contents indexed, especially I am having lots of files, reports and notes whether PDF files or research I some times do my self (cherry tree) plus lots of spreadsheets.

Just let us know if this is the case. Are you using BTRFS, Sanpper,…?
I’ve been there by the way and had this issue. (but sorted out now)

1 Like

Yes you can disable baloo from there so it isn’t constantly indexing.

Edit: This is just something you can try. I’m not saying that’s the reason for your issue.

1 Like

I always disable Baloo, it’s utterly useless, in my opinion.

When I need to search for something, I use fd, rg, and pdfgrep. All great tools that do not require indexing.

2 Likes

+1 for these. These are great tools, especially when used together with a fuzzy finder. Very useful for scripting as well.

There’s also rga (ripgrep-all) which extends the scope of rg to include archived files, office documents, pdf files, e-books and such.

3 Likes

Never heard of her. Will have to check it out, sounds good.

A nice script that use rg/rga and fzf :

More simple :

rg --line-number --field-match-separator ' ' -e 'string_to_search'| fzf --preview "cat {1}"
2 Likes