https://bugs.kde.org/show_bug.cgi?id=492808
--- Comment #9 from tagwer...@innerjoin.org ---
(In reply to Ellie from comment #8)
> ... I'm using BTRFS for most of my partitions, actually ...
It's possible that was the root cause - although it depends a bit on how it was
configured.
If you pick one of your files on an indexed BTRFS partition at run "stat":
$ stat one-of-your-files.txt
and note down the device number and inode. Then after a reboot, do the same and
compare. Maybe checking after each reboot for a while. You are watching to see
if you get new/different device numbers on each reboot.
If you mount a BTRFS partition, you don't (necessarily) get the same device
number. Baloo, previously, depended on a combination of the device number and
inode as an internal "ID" for the file. If the device number changed, Baloo
thought it had a set of new, unindexed, files and indexed them again. It now
digs deeper to get to the FileSystem ID which is invariant.
This means you might have had "some history" in your index and a load of
reindexing after a reboot. The total writes were bad because of the reindexing,
my guess is you will be OK if you reindex now. Indexing will be faster, the
index size smaller and the impact on the system less.
--
You are receiving this mail because:
You are watching all bug changes.
