Tom Lane [Fri, 18 Feb 2011 18:11:45 +0000 (13:11 -0500)]
Fix parallel pg_restore to handle comments on POST_DATA items correctly.
The previous coding would try to process all SECTION_NONE items in the
initial sequential-restore pass, which failed if they were dependencies of
not-yet-restored items. Fix by postponing such items into the parallel
processing pass once we have skipped any non-PRE_DATA item.
Back-patch into 9.0; the original parallel-restore coding in 8.4 did not
have this bug, so no need to change it.
Tom Lane [Fri, 18 Feb 2011 16:55:57 +0000 (11:55 -0500)]
One more hack to make contrib upgrades from 9.0 match fresh 9.1 installs.
intarray and tsearch2 both reference core support functions in their GIN
opclasses, and the signatures of those functions changed for 9.1. We added
backwards-compatible pg_proc entries for the functions in order to allow
9.0 dump files to be restored at all, but that hack leaves the opclasses
pointing at pg_proc entries different from what they'd point to if the
contrib modules were installed fresh in 9.1. To forestall any possibility
of future problems, fix the opclasses to match fresh installs via the
expedient of direct UPDATEs on pg_amproc in the update-from-unpackaged
scripts. (Yech ... but the alternatives are worse, or require far more
effort than seems justified right now.)
Note: updating pg_amproc is sufficient because there will be no pg_depend
entries corresponding to these dependencies, since the referenced functions
are all pinned.
Simon Riggs [Fri, 18 Feb 2011 15:07:26 +0000 (15:07 +0000)]
Make a hard state change from catchup to streaming mode.
More useful state change for monitoring purposes, plus a
required change for synchronous replication patch.
Simon Riggs [Fri, 18 Feb 2011 11:31:49 +0000 (11:31 +0000)]
Separate messages for standby replies and hot standby feedback.
Allow messages to be sent at different times, and greatly reduce
the frequency of hot standby feedback. Refactor to allow additional
message types.
Itagaki Takahiro [Fri, 18 Feb 2011 05:04:34 +0000 (14:04 +0900)]
Add transaction-level advisory locks.
They share the same locking namespace with the existing session-level
advisory locks, but they are automatically released at the end of the
current transaction and cannot be released explicitly via unlock
functions.
Alvaro Herrera [Fri, 18 Feb 2011 01:11:50 +0000 (22:11 -0300)]
Convert Postgres arrays to Perl arrays on PL/perl input arguments
More generally, arrays are turned in Perl array references, and row and
composite types are turned into Perl hash references. This is done
recursively, in a way that's natural to every Perl programmer.
To avoid a backwards compatibility hit, the string representation of
each structure is also available if the function requests it.
Authors: Alexey Klyukin and Alex Hunsaker.
Some code cleanups by me.
Tom Lane [Fri, 18 Feb 2011 00:00:49 +0000 (19:00 -0500)]
Fix tsmatchsel() to account properly for null rows.
ts_typanalyze.c computes MCE statistics as fractions of the non-null rows,
which seems fairly reasonable, and anyway changing it in released versions
wouldn't be a good idea. But then ts_selfuncs.c has to account for that.
Failure to do so results in overestimates in columns with a significant
fraction of null documents. Back-patch to 8.4 where this stuff was
introduced.
Tom Lane [Thu, 17 Feb 2011 21:37:34 +0000 (16:37 -0500)]
Fix upgrade of contrib/btree_gist from 9.0.
The initial version of the update-from-unpackaged script neglected to
include the <> operators that were added to the opclasses during 9.1.
To make this script produce the same final state as the regular install
script, use the same ALTER OPERATOR FAMILY trick as in pg_trgm.
Tom Lane [Thu, 17 Feb 2011 20:03:30 +0000 (15:03 -0500)]
Fix contrib/pg_trgm to have smoother updates from 9.0.
Take care of some loose ends in the update-from-unpackaged script, and
apply some ugly hacks to ensure that it produces the same catalog state
as the fresh-install script. Per discussion, this seems like a safer
plan than having two different catalog states that both call themselves
"pg_trgm 1.0", even if it's not immediately clear that the subtle
differences would ever matter.
Also, fix the stub function gin_extract_trgm() so that it works instead
of just bleating. Needed because this function will get called during a
regular dump and reload, if there are any indexes using its opclass.
The user won't have an opportunity to update the extension till later,
so telling him to do so is unhelpful.
Tom Lane [Thu, 17 Feb 2011 00:24:45 +0000 (19:24 -0500)]
Fix bogus test for hypothetical indexes in get_actual_variable_range().
That function was supposing that indexoid == 0 for a hypothetical index,
but that is not likely to be true in any non-toy implementation of an index
adviser, since assigning a fake OID is the only way to know at EXPLAIN time
which hypothetical index got selected. Fix by adding a flag to
IndexOptInfo to mark hypothetical indexes. Back-patch to 9.0 where
get_actual_variable_range() was added.
Tom Lane [Wed, 16 Feb 2011 22:24:06 +0000 (17:24 -0500)]
Add backwards-compatible declarations of some core GIN support functions.
These are needed to support reloading dumps of 9.0 installations containing
contrib/intarray or contrib/tsearch2. Since not only regular dump/reload
but binary upgrade would fail, it seems worth the trouble to carry these
stubs for awhile. Note that the contrib opclasses referencing these
functions will still work fine, since GIN doesn't actually pay any
attention to the declared signature of a support function.
Peter Eisentraut [Wed, 16 Feb 2011 21:06:36 +0000 (23:06 +0200)]
Better support for thread-support flag detection with clang
When testing the stderr produced by various thread-support flags, also
run a compilation in addition to a link, because clang warns on
certain flags when compiling but not when linking.
Simon Riggs [Wed, 16 Feb 2011 19:29:37 +0000 (19:29 +0000)]
Hot Standby feedback for avoidance of cleanup conflicts on standby.
Standby optionally sends back information about oldestXmin of queries
which is then checked and applied to the WALSender's proc->xmin.
GetOldestXmin() is modified slightly to agree with GetSnapshotData(),
so that all backends on primary include WALSender within their snapshots.
Note this does nothing to change the snapshot xmin on either master or
standby. Feedback piggybacks on the standby reply message.
vacuum_defer_cleanup_age is no longer used on standby, though parameter
still exists on primary, since some use cases still exist.
Simon Riggs, review comments from Fujii Masao, Heikki Linnakangas, Robert Haas
Itagaki Takahiro [Wed, 16 Feb 2011 02:19:11 +0000 (11:19 +0900)]
Export the external file reader used in COPY FROM as APIs.
They are expected to be used by extension modules like file_fdw.
There are no user-visible changes.
Itagaki Takahiro
Reviewed and tested by Kevin Grittner and Noah Misch.
Tom Lane [Tue, 15 Feb 2011 23:09:41 +0000 (18:09 -0500)]
Fix corner case for binary upgrade: extension functions in pg_catalog.
Normally, pg_dump summarily excludes functions in pg_catalog from
consideration. However, some extensions may create functions in pg_catalog
(adminpack already does that, and extensions for procedural languages will
likely do it too). In binary-upgrade mode, we have to dump such functions,
or the extension will be incomplete after upgrading. Per experimentation
with adminpack.
Tom Lane [Tue, 15 Feb 2011 20:49:54 +0000 (15:49 -0500)]
Add CheckTableNotInUse calls in DROP TABLE and DROP INDEX.
Recent releases had a check on rel->rd_refcnt in heap_drop_with_catalog,
but failed to cover the possibility of pending trigger events at DROP time.
(Before 8.4 we didn't even check the refcnt.) When the trigger events were
eventually fired, you'd get "could not open relation with OID nnn" errors,
as in recent report from strk. Better to throw a suitable error when the
DROP is attempted.
Tom Lane [Tue, 15 Feb 2011 01:59:42 +0000 (20:59 -0500)]
Rethink naming of contrib/intagg extension.
Initially it was called int_aggregate after the old SQL file, but since
the documentation just says "intagg" and that's also the directory name,
let's conform to that instead.
Simon Riggs [Tue, 15 Feb 2011 00:51:39 +0000 (00:51 +0000)]
PITR can stop at a named restore point when recovery target = time
though must not update the last transaction timestamp.
Plus comment and message cleanup for recent named restore point.
Tom Lane [Mon, 14 Feb 2011 21:07:00 +0000 (16:07 -0500)]
Rearrange extension-related views as per recent discussion.
The original design of pg_available_extensions did not consider the
possibility of version-specific control files. Split it into two views:
pg_available_extensions shows information that is generic about an
extension, while pg_available_extension_versions shows all available
versions together with information that could be version-dependent.
Also, add an SRF pg_extension_update_paths() to assist in checking that
a collection of update scripts provide sane update path sequences.
Tom Lane [Mon, 14 Feb 2011 03:53:00 +0000 (22:53 -0500)]
Assorted fixups for "unpackaged" conversion scripts.
From first pass of testing. Notably, there seems to be no need for
adminpack--unpackaged--1.0.sql because none of the objects that the
old module creates would ever be dumped by pg_dump anyway (they are
all in pg_catalog).
Tom Lane [Mon, 14 Feb 2011 02:24:14 +0000 (21:24 -0500)]
Avoid use of CREATE OR REPLACE FUNCTION in extension installation files.
It was never terribly consistent to use OR REPLACE (because of the lack of
comparable functionality for data types, operators, etc), and
experimentation shows that it's now positively pernicious in the extension
world. We really want a failure to occur if there are any conflicts, else
it's unclear what the extension-ownership state of the conflicted object
ought to be. Most of the time, CREATE EXTENSION will fail anyway because
of conflicts on other object types, but an extension defining only
functions can succeed, with bad results.
Tom Lane [Mon, 14 Feb 2011 01:06:41 +0000 (20:06 -0500)]
Convert contrib modules to use the extension facility.
This isn't fully tested as yet, in particular I'm not sure that the
"foo--unpackaged--1.0.sql" scripts are OK. But it's time to get some
buildfarm cycles on it.
sepgsql is not converted to an extension, mainly because it seems to
require a very nonstandard installation process.
Tom Lane [Sun, 13 Feb 2011 18:03:41 +0000 (13:03 -0500)]
Change the naming convention for extension files to use double dashes.
This allows us to have an unambiguous rule for deconstructing the names
of script files and secondary control files, without having to forbid
extension and version names from containing any dashes. We do have to
forbid them from containing double dashes or leading/trailing dashes,
but neither restriction is likely to bother anyone in practice.
Per discussion, this seems like a better solution overall than the
original design.
Tom Lane [Sat, 12 Feb 2011 21:40:41 +0000 (16:40 -0500)]
Refactor ALTER EXTENSION UPDATE to have cleaner multi-step semantics.
This change causes a multi-step update sequence to behave exactly as if the
updates had been commanded one at a time, including updating the "requires"
dependencies afresh at each step. The initial implementation took the
shortcut of examining only the final target version's "requires" and
changing the catalog entry but once. But on reflection that's a bad idea,
since it could lead to executing old update scripts under conditions
different than they were designed/tested for. Better to expend a few extra
cycles and avoid any surprises.
In the same spirit, if a CREATE EXTENSION FROM operation involves applying
a series of update files, it will act as though the CREATE had first been
done using the initial script's target version and then the additional
scripts were invoked with ALTER EXTENSION UPDATE.
I also removed the restriction about not changing encoding in secondary
control files. The new rule is that a script is assumed to be in whatever
encoding the control file(s) specify for its target version. Since this
reimplementation causes us to read each intermediate version's control
file, there's no longer any uncertainty about which encoding setting would
get applied.
Bruce Momjian [Sat, 12 Feb 2011 14:47:51 +0000 (09:47 -0500)]
Properly handle Win32 paths of 'E:abc', which can be either absolute or
relative, by creating a function path_is_relative_and_below_cwd() to
check for specific requirements. It is unclear if this fixes a security
problem or not but the new code is more robust.
Peter Eisentraut [Sat, 12 Feb 2011 13:54:13 +0000 (15:54 +0200)]
DDL support for collations
- collowner field
- CREATE COLLATION
- ALTER COLLATION
- DROP COLLATION
- COMMENT ON COLLATION
- integration with extensions
- pg_dump support for the above
- dependency management
- psql tab completion
- psql \dO command
Robert Haas [Sat, 12 Feb 2011 13:27:55 +0000 (08:27 -0500)]
Teach ALTER TABLE .. SET DATA TYPE to avoid some table rewrites.
When the old type is binary coercible to the new type and the using
clause does not change the column contents, we can avoid a full table
rewrite, though any indexes on the affected columns will still need
to be rebuilt. This applies, for example, when changing a varchar
column to be of type text.
The prior coding assumed that the set of operations that force a
rewrite is identical to the set of operations that must be propagated
to tables making use of the affected table's rowtype. This is
no longer true: even though the tuples in those tables wouldn't
need to be modified, the data type change invalidate indexes built
using those composite type columns. Indexes on the table we're
actually modifying can be invalidated too, of course, but the
existing machinery is sufficient to handle that case.
Along the way, add some debugging messages that make it possible
to understand what operations ALTER TABLE is actually performing
in these cases.
Tom Lane [Sat, 12 Feb 2011 03:53:43 +0000 (22:53 -0500)]
Clean up installation directory choices for extensions.
Arrange for the control files to be in $SHAREDIR/extension not
$SHAREDIR/contrib, since we're generally trying to deprecate the term
"contrib" and this is a once-in-many-moons opportunity to get rid of it in
install paths. Fix PGXS to install the $EXTENSION file into that directory
no matter what MODULEDIR is set to; a nondefault MODULEDIR should only
affect the script and secondary extension files. Fix the control file
directory parameter to be interpreted relative to $SHAREDIR, to avoid a
surprising disconnect between how you specify that and what you set
MODULEDIR to.
Tom Lane [Sat, 12 Feb 2011 02:25:20 +0000 (21:25 -0500)]
Add support for multiple versions of an extension and ALTER EXTENSION UPDATE.
This follows recent discussions, so it's quite a bit different from
Dimitri's original. There will probably be more changes once we get a bit
of experience with it, but let's get it in and start playing with it.
This is still just core code. I'll start converting contrib modules
shortly.
Robert Haas [Fri, 11 Feb 2011 13:47:38 +0000 (08:47 -0500)]
Tweak find_composite_type_dependencies API a bit more.
Per discussion with Noah Misch, the previous coding, introduced by
my commit 65377e0b9c0e0397b1598b38b6a7fb8b6f740d39 on 2011-02-06,
was really an abuse of RELKIND_COMPOSITE_TYPE, since the caller in
typecmds.c is actually passing the name of a domain. So go back
having a type name argument, but make the first argument a Relation
rather than just a string so we can tell whether it's a table or
a foreign table and emit the proper error message.
Send status updates back from standby server to master, indicating how far
the standby has written, flushed, and applied the WAL. At the moment, this
is for informational purposes only, the values are only shown in
pg_stat_replication system view, but in the future they will also be needed
for synchronous replication.
Extracted from Simon riggs' synchronous replication patch by Robert Haas, with
some tweaking by me.
Magnus Hagander [Thu, 10 Feb 2011 14:09:35 +0000 (15:09 +0100)]
Track last time for statistics reset on databases and bgwriter
Tracks one counter for each database, which is reset whenever
the statistics for any individual object inside the database is
reset, and one counter for the background writer.
Tom Lane [Thu, 10 Feb 2011 04:27:07 +0000 (23:27 -0500)]
Fix improper matching of resjunk column names for FOR UPDATE in subselect.
Flattening of subquery range tables during setrefs.c could lead to the
rangetable indexes in PlanRowMark nodes not matching up with the column
names previously assigned to the corresponding resjunk ctid (resp. tableoid
or wholerow) columns. Typical symptom would be either a "cannot extract
system attribute from virtual tuple" error or an Assert failure. This
wasn't a problem before 9.0 because we didn't support FOR UPDATE below the
top query level, and so the final flattening could never renumber an RTE
that was relevant to FOR UPDATE. Fix by using a plan-tree-wide unique
number for each PlanRowMark to label the associated resjunk columns, so
that the number need not change during flattening.
Per report from David Johnston (though I'm darned if I can see how this got
past initial testing of the relevant code). Back-patch to 9.0.
Tom Lane [Thu, 10 Feb 2011 00:17:33 +0000 (19:17 -0500)]
Fix pg_upgrade to handle extensions.
This follows my proposal of yesterday, namely that we try to recreate the
previous state of the extension exactly, instead of allowing CREATE
EXTENSION to run a SQL script that might create some entirely-incompatible
on-disk state. In --binary-upgrade mode, pg_dump won't issue CREATE
EXTENSION at all, but instead uses a kluge function provided by
pg_upgrade_support to recreate the pg_extension row (and extension-level
pg_depend entries) without creating any member objects. The member objects
are then restored in the same way as if they weren't members, in particular
using pg_upgrade's normal hacks to preserve OIDs that need to be preserved.
Then, for each member object, ALTER EXTENSION ADD is issued to recreate the
pg_depend entry that marks it as an extension member.
In passing, fix breakage in pg_upgrade's enum-type support: somebody didn't
fix it when the noise word VALUE got added to ALTER TYPE ADD. Also,
rationalize parsetree representation of COMMENT ON DOMAIN and fix
get_object_address() to allow OBJECT_DOMAIN.
Tom Lane [Wed, 9 Feb 2011 19:05:34 +0000 (14:05 -0500)]
Rethink order of operations for dumping extension member objects.
My original idea of doing extension member identification during
getDependencies() didn't work correctly: we have to mark member tables as
not-to-be-dumped rather earlier than that, else their subsidiary objects
like indexes get dumped anyway. Rearrange code to mark them early enough.
Tom Lane [Wed, 9 Feb 2011 16:55:32 +0000 (11:55 -0500)]
Implement "ALTER EXTENSION ADD object".
This is an essential component of making the extension feature usable;
first because it's needed in the process of converting an existing
installation containing "loose" objects of an old contrib module into
the extension-based world, and second because we'll have to use it
in pg_dump --binary-upgrade, as per recent discussion.
Loosely based on part of Dimitri Fontaine's ALTER EXTENSION UPGRADE
patch.