On Sun, Jan 22, 2017 at 6:57 AM, Stephen Frost <sfrost@snowman.net> wrote:
Greetings,
* julyanto SUTANDANG (julyanto@equnix.co.id) wrote: > CORRECTION: > > "you might you pg_start_backup to tell the server not to write into the > DATADIR" > > become > > "you might *use* pg_start_backup to tell the server not to write into the > *BASEDIR*, actually server still writes but only to XLOGDIR "
Just to make sure anyone reading the mailing list archives isn't confused, running pg_start_backup does *not* make PG stop writing to BASEDIR (or DATADIR, or anything, really). PG *will* continue to write data into BASEDIR after pg_start_backup has been called.
Correct. Unfortunately it is a very common myth that it does cause PostgreSQL to stop writing to the base dir.
The only thing that pg_start_backup does is identify an entry in the WAL stream, from which point all WAL must be replayed when restoring the backup. All WAL generated from that point (pg_start_backup point) until the pg_stop_backup point *must* be replayed when restoring the backup or the database will not be consistent.
pg_start_backup also forces full_page_writes to be effectively 'on' for the duration of the backup, if it is not already explicitly on (which it usually will already be). This affects pg_xlog, of course, not base. But it is an essential step for people who run with full_page_writes=off, as it ensures that anything in base which got changed mid-copy will be fixed up during replay of the WAL.