Fixed upgrade_table_schema/upgrade_runtime_migration bugs related to hive_post_tags table data migration.

Removed creation of hive_posts (tags_ids) GIN indexes from upgrade code, due to new tags implementation

Removed obsolete creation of hivemind_app.hive_db_patch_level table from upgrade script (this table is created by regular hivemind schema)

Prerequisites to support database upgrade spawned directly by Hivemind python implementation (what allows to safely reference upgrade .sql scripts)

Marcin authored 11 months ago and Dan Notestein committed 11 months ago

also added trick to force execute deleted_post_tags CTE
because previously postgres optimize it and delete was not executed
what leads to duplicate pair (tag_id, post_id) in hive_post_tags table

Marcin authored 11 months ago and Dan Notestein committed 11 months ago

make faster api bridge_get_ranked_post for tag

Added insert condition to prevent hive_db_patch_level pollution with records having specified same git revision

After breaking massive sync, when hivemind was restarted
it started to sync from block 1, because current_block_num
was not updated

tags_ids index replaced by 2 partial ones specific to live_posts and live post-or-comments conditions

Konrad Botor authored 1 year ago and Bartek Wrona committed 1 year ago

Updated build_data.sh to properly remove invalid HAF replay and to wait if another replay is in progress

Cleanup inside db_upgrade.sh script, which right now is dedicated only optional (performed manually) hivemind table schema upgrades

SQL scripts containing definitions of runtime code (functions/views etc) must be ready for reinstallation each time install_app is spawned

Reimplemented main query inside bridge_get_ranked_post_by_created_for_community to enforce using correct index

Eliminated not needed JOIN to hive_accounts table inside hivemind_app.muted_accounts_by_id_view definition