Postgresql – Meaning of query in pgbouncer.log

Those two columns are the result of

SELECT pg_stat_get_live_tuples(c.oid) AS n_live_tup
     , pg_stat_get_dead_tuples(c.oid) AS n_dead_tup
FROM   pg_class c;

Representing the number of live and dead rows (tuples) in the table.
Find those functions in the manual.

Dead rows are deleted rows that will later be reused for new rows from INSERTs or UPDATEs (the space, not the data). Some dead rows (or reserved free space) can be particularly useful for HOT updates (Heap-Only Tuples) that can reuse space in the same data page efficiently. More on H.O.T.:

• PostgreSQL Initial Database Size
• Release notes for PostgreSQL 8.3 where this feature was introduced.

Or dead rows may be removed by VACUUM FULL (or plain VACUUM if it gets lucky) or similar operations on the table, thereby shrinking the physical table accordingly.

Whenever a row is deleted or updated, the old row version becomes invisible to all other transactions starting after the transaction has been committed. The row is completely dead as soon as there are no more uncommitted older transactions. That is necessary for PostgreSQL's MVCC model to handle concurrency.

Those are just statistics. You need to enable statistics collection in postgresql.conf if you want them to be updated automatically. track_counts should be on by default, though. Bear in mind that statistics are not updated instantaneously. Read more about that in the manual.

I believe the answer here is "not exactly". Those stats are coming from pgbouncer itself, so that don't map directly to the postgres backends. That said, you can track connection, data, and query statistics in postgres, but the information won't align exactly, and will probably be formatted differently (ie. you're more likely to get a counter than a gauge, so you'll need to do some tracking/math yourself)