https://bugs.kde.org/show_bug.cgi?id=460460

--- Comment #3 from tagwer...@innerjoin.org ---
(In reply to Adam Fontenot from comment #2)
> ... the act of checking on Baloo's status is what triggers both
> memory use and i/o thrashing followed by bloating the database? That's ...
> insidious ...
I tend to agree :-)

"baloo status" locks the file while it tries to enumerate the of files it
should index; while the file is locked, updates are committed as "appends"; the
database balloons in size and memory use goes up with it. It's a vicious circle
and particularly bad when baloo's trying to delete loads of records.

> ... This is part of what I'm advocating here; there should be no work done by 
> Baloo
> that is invisible to `balooctl` or is unkillable by the user ...
... and agree here too.

When it starts up "baloo_file" scans through the filesystem looking for
unindexed or updated files. At the moment it does that in memory, only
committing the results at the end of the process. That means "balooctl" doesn't
see the work done when it is looking in the database file. I'd say "baloo_file"
ought to update to the database - every 15 seconds seems OK to me - so that
"balooctl" can show what's happening

> ... I decided to clear up some free space in order to be able to export 
> files. These were mostly
> a bunch of large source code directories ...
That definitely sounds like the trigger.

> ... content indexing disabled for a while (due to the innumerable bugs
> I've encountered with it) ...
> I am indeed using BTRFS.
If you are interested, have a look at:

    https://bugs.kde.org/show_bug.cgi?id=402154#c12

With BTFRS and multiple subvols, it can be that you see a different "minor
device number" after a reboot. Baloo tracks files in its index by "DocID" and
which is a concatenation of the device number and inode, if the device number
changes on reboot then baloo thinks it has a completely new set of files to
index.

Does it make sense to close this as a duplicate of Bug 437754?

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to