Postgresql – is it possible to run postgres with no WAL files being produced

You asked:

how postgreSQL will handle the recovery with a pg_data content containing some files which are inconsistent.

pg_start_backup() ensure the data file is at least as new as the checkpoint. On recovery, the logs are applied.

If the data is old, the log will update it..

If the data is new, the log will have same content. There is no hurt writing it again.

The data are never newer then the log, because the logs are write ahead (WAL).

You asked:

... xfs-freeze ...

xfs-freeze is alike to pg_start_backup(), it don't take a snapshot. You need a volume manager to do that.

You asked:

... why do create tablespace & create database statements are unsupported if the WAL can replay everything?

It is supported, just some little gotcha. See http://www.postgresql.org/docs/8.1/static/backup-online.html :

23.3.5. Caveats

CREATE TABLESPACE commands are WAL-logged with the literal absolute path, and will therefore be replayed as tablespace creations with the same absolute path. This might be undesirable if the log is being replayed on a different machine. It can be dangerous even if the log is being replayed on the same machine, but into a new data directory: the replay will still overwrite the contents of the original tablespace. To avoid potential gotchas of this sort, the best practice is to take a new base backup after creating or dropping tablespaces.

If I recall correctly, PostgreSQL now zeroes the end of the WAL segment after an archive timeout, before archiving the file. (Update): This is true for 9.4; for prior versions you still need to use pg_clearxlogtail utility to zero the end of any WAL segment as part of your archive script before you gzip your archive files.

In 9.4 and above you just need to gzip the file, with no need for pg_clearxlogtail.

The change was made commit 9a20a9b, but it's only in tag REL9_4_BETA1, so the change will first appear in 9.4:

commit 9a20a9b
Author: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Date:   Mon Jul 8 11:23:56 2013 +0300

    Improve scalability of WAL insertions.

    This patch replaces WALInsertLock with a number of WAL insertion slots,
    allowing multiple backends to insert WAL records to the WAL buffers
    concurrently. This is particularly useful for parallel loading large amounts
    of data on a system with many CPUs.

    This has one user-visible change: switching to a new WAL segment with
    pg_switch_xlog() now fills the remaining unused portion of the segment with
    zeros. This potentially adds some overhead, but it has been a very common
    practice by DBA's to clear the "tail" of the segment with an external
    pg_clearxlogtail utility anyway, to make the WAL files compress better.
    With this patch, it's no longer necessary to do that.

BTW, you should really be using an archive timeout, rather than manually switching xlogs. Consider having a streaming replica for additional protection.

Update: As Jeff noted, pg_clearxlogtail hasn't been updated for 9.3. The incompatibility is very simple though; the definition of XLogRecPtr changed from

typedef struct XLogRecPtr
{
        uint32          xlogid;                 /* log file #, 0 based */
        uint32          xrecoff;                /* byte offset of location in log file */
} XLogRecPtr;

to

typedef uint64 XLogRecPtr;

when support for platforms with compilers that don't support 64-bit long integers was removed from PostgreSQL in 9.3. So all we need to do is add a macro or function that tests equality in a way that works for both.

The following patch adds support for 9.3, but is completely untested so use entirely at your own risk. If it eats your data then sets fire to your house, well, that's how things are. Patch here, as an attachment to the linked pgfoundry issue.