These are chat archives for collectiveaccess/support

10th
Jan 2018
rsteinert
@rsteinert
Jan 10 2018 05:45
Any idea why MySQL replication gets thrown off when processing ca_sql_search_word_index. I have start with a fresh copy of my database twice now and the each time slave throws an error , "Could not execute Delete_rows event on table ca_sql_search_word_index; Can't find record in 'ca_sql_search_word_index', Error_code: 1032; handler error HA_ERR_KEY_NOT_FOUND;"
CollectiveAccess
@collectiveaccess
Jan 10 2018 12:24
Are you changing data on the slave? There are lots of actions in CA that cause indexing to happen, which will write changes into the ca_sql_search_word_index table. MySQL row-based replication doesn't like when table content on the slave changes out from under it. My guess is that you're using the slave for something and the bit of indexing that's getting performed on the slave only is the cause of the problem,.
You might get around this by not replicating the indexing tables, and then periodically reindexing on the slave side. This has obvious disadvantages though.
rsteinert
@rsteinert
Jan 10 2018 13:40
I have Providence working off the master version and Pawtucket working off the slave, so I am guessing that Pawtucket's activity is changing "ca_sql_search_word_index" on its own? My idea with this set up was to have the master database and to replicate it for use by Pawtucket on another server (a subdomain of our main website). I didn't consider that Pawtucket would change the data as well.