Version:

Collecting Per-Database Data

radar runs the per-database collectors for each accessible database and stores their files in databases/{dbname}/. radar does not connect to a database on which the role lacks the CONNECT privilege. For that database, radar reports one error that names the missing privilege. The following table describes the files that these collectors produce for each database:

File	Source	Description
`bloat.tsv`	`pg_stats` heuristic	Estimates table bloat with a heuristic.
`extensions.tsv`	`pg_extension`	Lists the installed extensions.
`funcs.tsv`	`pg_proc WHERE prokind='f'`	Lists the functions.
`indexes.tsv`	`pg_index` + `pg_stat_all_indexes`	Lists the top 1000 indexes by size with the semantic key, validity, scan counters and size.
`languages.tsv`	`pg_language`	Lists the procedural languages.
`operators.tsv`	`pg_operator`	Lists the operators.
`partitioned_tables.tsv`	`pg_partitioned_table`	Lists the partitioned tables (PostgreSQL 10 and later).
`partitions.tsv`	`pg_inherits`	Lists the partition relationships.
`pgstattuple.tsv`	`pgstattuple_approx()`	Reports bloat when the `pgstattuple` extension is installed, with exact dead-tuple figures and approximate live-tuple and free-space figures. The collector is disabled by default because it reads every heap page that the visibility map does not mark all-visible, in every user table; run radar with `--include pgstattuple` to enable the collector.
`procs.tsv`	`pg_proc WHERE prokind='p'`	Lists the procedures (PostgreSQL 11 and later).
`publication_tables.tsv`	`pg_publication_tables`	Lists the tables in publications.
`publications.tsv`	`pg_publication`	Lists the logical replication publications.
`schemas.tsv`	`pg_namespace`	Lists the schemas.
`sequences.tsv`	`pg_sequences`	Lists the sequences with their last, minimum and maximum values (PostgreSQL 10 and later).
`stat_database.tsv`	`pg_stat_database`	Reports per-database statistics.
`statistics.tsv`	`pg_statistic_ext`	Lists the extended statistics (PostgreSQL 10 and later).
`subscription_tables.tsv`	`pg_subscription_rel`	Reports the subscription relation states.
`table_freeze_age.tsv`	`pg_class`	Lists the top 1000 relations by transaction-ID and multixact freeze age. The ranking uses age rather than size, so the file includes a small table that holds the wraparound horizon back where `tables.tsv` would miss that table. The file also covers the `pg_catalog` and `pg_toast` relations that `tables.tsv` excludes. The query selects only relations whose `relkind` is `r`, `m` or `t` (ordinary tables, materialized views and TOAST tables).
`tables.tsv`	`pg_class` + `pg_stat_all_tables`	Lists the top 1000 tables by size with persistence, options, heap and table sizes, toast mapping, dead-tup counters, vacuum and analyze timestamps, per-table vacuum and analyze ages in seconds, and transaction-ID and multixact freeze ages. The freeze ages are empty for partitioned tables, which hold no tuples of their own. The query requires PostgreSQL 13 or later.
`triggers.tsv`	`pg_trigger`	Lists the triggers.
`types.tsv`	`pg_type`	Lists the data types.