Bug 415132

SUMMARY
Baloo seems to be working in general, returns valid search results for indexed file content.

Baloo File Indexer is running
Indexer state: Indexing file content
Total files indexed: 446,139
Files waiting for content indexing: 1
Files failed to index: 100
Current size of index is 26.64 GiB

But the xorg-session.log is around 0,5 TB now and growing, full off lines like
org.kde.baloo.engine: Transaction::commit MDB_BAD_TXN: Transaction must abort, has a child, or is invalid

Performance of MD Raid 1 goes down dew to that.

STEPS TO REPRODUCE
1. Have 400000+ files that can be indexed
2. 
3. 

OBSERVED RESULT
Log goes to eat all FS (EXT4) space on home drive

EXPECTED RESULT
Log does not eat all home drive FS space due to logging, baloo not complaining about "org.kde.baloo.engine: Transaction::commit MDB_BAD_TXN: Transaction must abort, has a child, or is invalid"

SOFTWARE/OS VERSIONS
KDE Plasma Version: 5.17.4
KDE Frameworks Version: 5.64.0
Qt Version: 5.13.2

ADDITIONAL INFORMATION
64 GiB RAM, gcc-9.2.0, Gentoo ~amd64, fs.inotify.max_user_watches = 1024000


It took my desktop around 14 days to index all on the relatively slow SW RAID 1 made from two 4TB 5900 RPM hard disks. I would really not like to delete the database and start the indexing from scratch. A fix is much preferred. Can I fix the database on my system, if it is broken? I will not share its content.