Re: pg_upgrade and rsync

4 марта 2015 г., в 19:28, Bruce Momjian <bruce@momjian.us> написал(а):

On Wed, Mar  4, 2015 at 01:53:47PM +0300, Vladimir Borodin wrote:

   After running initdb to create the new master, but before running
   pg_upgrade, modify the new master's postgresql.conf and change wal_level
   = hot_standby.  (Don't modify pg_hba.conf at this stage.)



That does not help. The reason is that pg_upgrade sets 'Current wal_level
setting: minimal' in control-file, and it does not depend on what is set in
postgresql.conf before running pg_upgrade. Details could be seen here - http://
pastie.org/9998671.

Well, what is happening is that the pg_resetxlog commands we run inside
pg_upgrade set the pg_controldata file's wal_level to minimal, but as
you saw, starting the server sets the pg_controldata properly.
pg_resetxlog is not changing the WAL files at all, just the control
file.

The workaround for this is to start  and cleanly shut down postgres on master
after running pg_upgrade but before running rsync. After that there would be a
good control-file for streaming replication and rsync to replica can be done.

You are correct that a pg_controldata file is copied over that has
wal_level=minimal, but that should not be a problem.

But it could not be done with --size-only key, because control-file is of fixed
size and rsync would skip it. Or may be everything should be copied with
--size-only and control-file should be copied without this option.

Well, what happens is that there is no _new_ standby pg_controldata
file, so it is copied fully from the new master.  Are you running initdb
to create the new standby --- you shouldn't be doing that as the rsync
will do that for you.  

Also, are you cleanly shutting down all the
servers, or using pg_ctl -m immediate?

--
 Bruce Momjian  <bruce@momjian.us>        http://momjian.us
 EnterpriseDB                             http://enterprisedb.com

 + Everyone has their own god. +