Last modified: 2014-03-22 11:22:33 UTC
On enwiki: +------------------+----------------------------+-----------+ | table_name | index_name | rows_read | +------------------+----------------------------+-----------+ | article_feedback | PRIMARY | NULL | | article_feedback | article_feedback_timestamp | NULL | | article_feedback | aa_page_id | NULL | | article_feedback | aa_page_user_token | NULL | +------------------+----------------------------+-----------+ 4 rows in set (0.25 sec) NULL rows_read indicates the indexes havn't been used. Stats collection has been running here for over a week. Same results on S1 master and several slaves. Indexes outweigh data on article_feedback quite a bit, taking up space and increasing write load: +---------------------------------------+--------+-------+--------+ | CONCAT(table_schema, '.', table_name) | rows | DATA | idx | +---------------------------------------+--------+-------+--------+ | enwiki.article_feedback | 69.00M | 7.13G | 26.04G | +---------------------------------------+--------+-------+--------+ Surprising; now I'm suspicious of the stats... Are these indexes really likely to be simply unused? Is there any infrequent maintenance or reporting job that requires them?
(In reply to Sean Pringle from comment #0) > Surprising; now I'm suspicious of the stats... Are these indexes really > likely to be simply unused? > > Is there any infrequent maintenance or reporting job that requires them? No, they're just old tables for an extension disabled years ago (so called AFT/AFTv4). Dario, is there any data which was not yet added to http://datahub.io/en/dataset/wikipedia-article-ratings ?