The following bug has been logged online:

Bug reference: 2239
Logged by: Olleg Samoylov
Email address: olleg_s (AT) mail (DOT) ru
PostgreSQL version: 8.1
Operating system: Linux (debian-amd64)
Description: "vacuumdb -a" remove freeze
Details:

template1=# select version();
version

----------------------------------------------------------------------------
------------------------------------
PostgreSQL 8.1.0 on x86_64-pc-linux-gnu, compiled by GCC cc (GCC) 4.0.3
20051111 (prerelease) (Debian 4.0.2-4)
(1 row)
template1=# select * from pg_database where datname='template1';
datname | datdba | encoding | datistemplate | datallowconn | datconnlimit
-----------+--------+----------+---------------+--------------+-------------
-+---------------+--------------+--------------+---------------+-----------+
------------------------
template1 | 10 | 6 | t | t | -1
(1 row)
template1=# vacuum freeze;
VACUUM
template1=# select * from pg_database where datname='template1';
datname | datdba | encoding | datistemplate | datallowconn | datconnlimit
-----------+--------+----------+---------------+--------------+-------------
-+---------------+--------------+--------------+---------------+-----------+
------------------------
template1 | 10 | 6 | t | t | -1
(1 row)

Okey, freezed, now launch vacuumdb from command line:

# vacuumdb -a -U postgres
vacuumdb: vacuuming database "postgres"
VACUUM
vacuumdb: vacuuming database "template1"
VACUUM

Opps, template1 must not be vacuumed.

select * from pg_database where datname='template1';
datname | datdba | encoding | datistemplate | datallowconn | datconnlimit
-----------+--------+----------+---------------+--------------+-------------
-+---------------+--------------+--------------+---------------+-----------+
------------------------
template1 | 10 | 6 | t | t | -1
(1 row)

Unfreezed indeed.

---------------------------(end of broadcast)---------------------------
TIP 1: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to majordomo (AT) postgresql (DOT) org so that your
message can get through to the mailing list cleanly

"Olleg Samoylov" <olleg_s (AT) mail (DOT) ru> writes:
Says who?

If we didn't vacuum template1 then it would be subject to XID wraparound
problems, unless it had never been modified, which is something vacuumdb
can't count on.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 4: Have you searched our list archives?

http://archives.postgresql.org

Tom Lane wrote:
template1 frozen by "vacuum freeze", thus can't be subject of XID
wraparound problems, isn't it?

man vacuum:
IMHO "vacuumdb -a" must don't vacuum database with
datvacuumxid=datfrozenxid.

--
Olleg Samoylov

---------------------------(end of broadcast)---------------------------
TIP 5: don't forget to increase your free space map settings

Olleg Samoylov <olleg (AT) mipt (DOT) ru> writes:
That's not going to work because it will fail to detect whether the
database has been modified since the VACUUM FREEZE command.

In any case, what's the point? As long as you have a routine vacuuming
process in place, it doesn't really matter whether template1 gets
scanned. The only reason VACUUM FREEZE exists at all is to make it
possible to have a non-connectable, non-vacuumable template0.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 5: don't forget to increase your free space map settings