[Koha-bugs] [Bug 6946] New: Import Batches _update_batch_record_counts has inefficient SQL
bugzilla-daemon at bugs.koha-community.org
bugzilla-daemon at bugs.koha-community.org
Fri Sep 30 04:26:05 CEST 2011
http://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=6946
Bug #: 6946
Summary: Import Batches _update_batch_record_counts has
inefficient SQL
Classification: Unclassified
Change sponsored?: ---
Product: Koha
Version: Rel_3_8
Platform: All
OS/Version: All
Status: NEW
Severity: minor
Priority: P5
Component: Architecture, internals, and plumbing
AssignedTo: ian.walls at bywatersolutions.com
ReportedBy: ian.walls at bywatersolutions.com
QAContact: koha-bugs at lists.koha-community.org
The queries to update the number of biblios and the number of items for any
given import batch are much slower than they need to be.
UPDATE import_batches
SET num_biblios = (SELECT COUNT(*)
FROM import_records
WHERE import_batch_id = import_batches.import_batch_id
AND record_type = 'biblio')
WHERE import_batch_id = ?
UPDATE import_batches
SET num_items = (SELECT COUNT(*)
FROM import_records
JOIN import_items USING (import_record_id)
WHERE import_batch_id = import_batches.import_batch_id
AND record_type = 'biblio')
WHERE import_batch_id = ?")
The "AND record_type = 'biblio'" lines are unnecesary; every record within the
same import_batch_id will have the same record_type, and more importantly,
'biblio' is currently the only supported type.
Removing the line changes the return time of the subquery from over 2 seconds
down to 0.01 seconds (on my machine with 34944 items in a batch). I think this
is going to be faster than adding an index on record_type.
--
Configure bugmail: http://bugs.koha-community.org/bugzilla3/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA Contact for the bug.
More information about the Koha-bugs
mailing list