This bug refers to the Z39.50 "reservoir" which cannot be [easily] emptied. A quick look in MySQL shows that we have: mysql> SELECT COUNT(*) FROM import_records; +----------+ | COUNT(*) | +----------+ | 148357 | +----------+ 1 row in set (0.01 sec) and the exact same in import_biblios (none in either import_items or import_record_matches) -- this relates to 29,410 valid biblios in our catalogue. mysql> SHOW TABLE STATUS LIKE 'import_records'; tells me that this table has a Data_length of 1040508672 or just over 1 Gbyte. Is there a downside to deleting this data? [TRUNCATE TABLE import_records; same for import_biblios] *Limited* testing in my sandbox has shown no downside or side effects. The upside is a measurable improvement in the search time from koha-admin/cgi-bin/koha/cataloguing/addbooks.pl, and speed/size improvements for backups. My cataloguers tell me that they never use the "reservoir" as it only corresponds to historical data from previous multi-Z39.50 server searches from which they already selected the "best fit" and thought that they had discarded the rest. As an aside, I'd be curious to know how other libraries use this data. Thanks and best regards, Paul