Re: pending patch: Re: Streaming replication and pg_xlogfile_name() - Mailing list pgsql-hackers

From Fujii Masao
Subject Re: pending patch: Re: Streaming replication and pg_xlogfile_name()
Date
Msg-id k2i3f0b79eb1004011953i463dadfdy39a6d22c7c1bdee8@mail.gmail.com
Whole thread Raw
In response to Re: pending patch: Re: Streaming replication and pg_xlogfile_name()  (Robert Haas <robertmhaas@gmail.com>)
Responses Re: pending patch: Re: Streaming replication and pg_xlogfile_name()  (Heikki Linnakangas <heikki.linnakangas@enterprisedb.com>)
List pgsql-hackers
On Fri, Apr 2, 2010 at 2:22 AM, Robert Haas <robertmhaas@gmail.com> wrote:
> Can someone explain to me in plain language what problem this is
> trying to fix?  I'm having trouble figuring it out.

The problem is that pg_xlogfile_name(pg_last_xlog_receive_location()) and
pg_xlogfile_name(pg_last_xlog_replay_location()) might report an inaccurate
WAL file name because currently pg_xlogfile_name() always uses the current
timeline to calculate the WAL file name. For example, even though the last
applied WAL file is 000000010000000000000002, the standby wrongly reports
that 000000000000000000000002 has been applied last.
   postgres=# SELECT l lsn, pg_xlogfile_name(l) filename FROM
pg_last_xlog_replay_location() AS l;       lsn    |         filename   -----------+--------------------------
0/200FF70| 000000000000000000000002   (1 row) 
   $ ls sby/pg_xlog/   000000010000000000000002  000000010000000000000003  archive_status

Regards,

--
Fujii Masao
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center


pgsql-hackers by date:

Previous
From: Pei He
Date:
Subject: Problem of Magic Block in Postgres 8.2
Next
From: Dmitry Fefelov
Date:
Subject: Re: Feature request - function-based deferrable uniques.