infra/models: review table indices

[slint]

From running https://github.com/ankane/dexter on a local PostgreSQL instance, the following table indices were recommended:

CREATE INDEX IF NOT EXISTS idx_accounts_user_displayname ON accounts_user USING btree (displayname);
CREATE INDEX IF NOT EXISTS idx_communities_members_group_id ON communities_members USING btree (group_id);
CREATE INDEX IF NOT EXISTS idx_communities_members_user_id ON communities_members USING btree (user_id);
CREATE INDEX IF NOT EXISTS idx_communities_metadata_bucket_id ON communities_metadata USING btree (bucket_id);
CREATE INDEX IF NOT EXISTS idx_files_files_last_check ON files_files USING btree (last_check);
CREATE INDEX IF NOT EXISTS idx_pidstore_pid_pid_value ON pidstore_pid USING btree (pid_value);
CREATE INDEX IF NOT EXISTS idx_rdm_parents_community_request_id ON rdm_parents_community USING btree (request_id);
CREATE INDEX IF NOT EXISTS idx_rdm_records_files_record_id ON rdm_records_files USING btree (record_id);
CREATE INDEX IF NOT EXISTS idx_rdm_records_metadata_bucket_id ON rdm_records_metadata USING btree (bucket_id);
CREATE INDEX IF NOT EXISTS idx_rdm_versions_state_next_draft_id ON rdm_versions_state USING btree (next_draft_id);

This means that in some cases we're performing inefficient SQL queries, missing out on indexed columns (a prime example would be the idx_pidstore_pid_pid_value index on pidstore_pid.pid_value). Most of the statistics collected during record bulk indexing to search, but in general we should try to profile SQL queries and review our tables.

[github-actions]

This issue was automatically marked as stale.

[github-actions]

This issue was automatically marked as stale.