Attempting to use CREATE INDEX, DROP INDEX or REINDEX with CONCURRENTLY
on a temporary relation with ON COMMIT actions triggered unexpected
errors because those operations use multiple transactions internally to
complete their work. Here is for example one confusing error when using
ON COMMIT DELETE ROWS:
ERROR: index "foo" already contains data
Issues related to temporary relations and concurrent indexing are fixed
in this commit by enforcing the non-concurrent path to be taken for
temporary relations even if using CONCURRENTLY, transparently to the
user. Using a non-concurrent path does not matter in practice as locks
cannot be taken on a temporary relation by a session different than the
one owning the relation, and the non-concurrent operation is more
effective.
The problem exists with REINDEX since v12 with the introduction of
CONCURRENTLY, and with CREATE/DROP INDEX since CONCURRENTLY exists for
those commands. In all supported versions, this caused only confusing
error messages to be generated. Note that with REINDEX, it was also
possible to issue a REINDEX CONCURRENTLY for a temporary relation owned
by a different session, leading to a server crash.
The idea to enforce transparently the non-concurrent code path for
temporary relations comes originally from Andres Freund.
Reported-by: Manuel Rigger
Author: Michael Paquier, Heikki Linnakangas
Reviewed-by: Andres Freund, Álvaro Herrera, Heikki Linnakangas
Discussion: https://postgr.es/m/CA+u7OA6gP7YAeCguyseusYcc=uR8+ypjCcgDDCTzjQ+k6S9ksQ@mail.gmail.com
Backpatch-through: 9.4
This follows multiple complains from Peter Geoghegan, Andres Freund and
Alvaro Herrera that this issue ought to be dug more before actually
happening, if it happens.
Discussion: https://postgr.es/m/20191226144606.GA5659@alvherre.pgsql
The following renaming is done so as source files related to index
access methods are more consistent with table access methods (the
original names used for index AMs ware too generic, and could be
confused as including features related to table AMs):
- amapi.h -> indexam.h.
- amapi.c -> indexamapi.c. Here we have an equivalent with
backend/access/table/tableamapi.c.
- amvalidate.c -> indexamvalidate.c.
- amvalidate.h -> indexamvalidate.h.
- genam.c -> indexgenam.c.
- genam.h -> indexgenam.h.
This has been discussed during the development of v12 when table AM was
worked on, but the renaming never happened.
Author: Michael Paquier
Reviewed-by: Fabien Coelho, Julien Rouhaud
Discussion: https://postgr.es/m/20191223053434.GF34339@paquier.xyz
Tuple conversion support in tupconvert.c is able to convert rowtypes
between two relations, inner and outer, which are logically equivalent
but have a different ordering or even dropped columns (used mainly for
inheritance tree and partitions). This makes use of attribute mappings,
which are simple arrays made of AttrNumber elements with a length
matching the number of attributes of the outer relation. The length of
the attribute mapping has been treated as completely independent of the
mapping itself until now, making it easy to pass down an incorrect
mapping length.
This commit refactors the code related to attribute mappings and moves
it into an independent facility called attmap.c, extracted from
tupconvert.c. This merges the attribute mapping with its length,
avoiding to try to guess what is the length of a mapping to use as this
is computed once, when the map is built.
This will avoid mistakes like what has been fixed in dc816e58, which has
used an incorrect mapping length by matching it with the number of
attributes of an inner relation (a child partition) instead of an outer
relation (a partitioned table).
Author: Michael Paquier
Reviewed-by: Amit Langote
Discussion: https://postgr.es/m/20191121042556.GD153437@paquier.xyz
If CheckAttributeType() threw an error about the datatype of an
index expression column, it would report an empty column name,
which is pretty unhelpful and certainly not the intended behavior.
I (tgl) evidently broke this in commit cfc5008a5, by not noticing
that the column's attname was used above where I'd placed the
assignment of it.
In HEAD and v12, this is trivially fixable by moving up the
assignment of attname. Before v12 the code is a bit more messy;
to avoid doing substantial refactoring, I took the lazy way out
and just put in two copies of the assignment code.
Report and patch by Amit Langote. Back-patch to all supported
branches.
Discussion: https://postgr.es/m/CA+HiwqFA+BGyBFimjiYXXMa2Hc3fcL0+OJOyzUNjhU4NCa_XXw@mail.gmail.com
We implement ON COMMIT DELETE ROWS by truncating tables marked that
way, which requires also truncating/rebuilding their indexes. But
RelationTruncateIndexes asks the relcache for up-to-date copies of any
index expressions, which may cause execution of eval_const_expressions
on them, which can result in actual execution of subexpressions.
This is a bad thing to have happening during ON COMMIT. Manuel Rigger
reported that use of a SQL function resulted in crashes due to
expectations that ActiveSnapshot would be set, which it isn't.
The most obvious fix perhaps would be to push a snapshot during
PreCommit_on_commit_actions, but I think that would just open the door
to more problems: CommitTransaction explicitly expects that no
user-defined code can be running at this point.
Fortunately, since we know that no tuples exist to be indexed, there
seems no need to use the real index expressions or predicates during
RelationTruncateIndexes. We can set up dummy index expressions
instead (we do need something that will expose the right data type,
as there are places that build index tupdescs based on this), and
just ignore predicates and exclusion constraints.
In a green field it'd likely be better to reimplement ON COMMIT DELETE
ROWS using the same "init fork" infrastructure used for unlogged
relations. That seems impractical without catalog changes though,
and even without that it'd be too big a change to back-patch.
So for now do it like this.
Per private report from Manuel Rigger. This has been broken forever,
so back-patch to all supported branches.
This gives an alternative way of catching exceptions, for the common
case where the cleanup code is the same in the error and non-error
cases. So instead of
PG_TRY();
{
... code that might throw ereport(ERROR) ...
}
PG_CATCH();
{
cleanup();
PG_RE_THROW();
}
PG_END_TRY();
cleanup();
one can write
PG_TRY();
{
... code that might throw ereport(ERROR) ...
}
PG_FINALLY();
{
cleanup();
}
PG_END_TRY();
Discussion: https://www.postgresql.org/message-id/flat/95a822c3-728b-af0e-d7e5-71890507ae0c%402ndquadrant.com
When cancelling REINDEX CONCURRENTLY after swapping the old and new
indexes (for example interruption at step 5), the old index remains
around and is marked as invalid. The old index should also be manually
droppable to clean up the parent relation from any invalid indexes still
remaining. For a partition index reindexed, pg_class.relispartition was
not getting updated, causing the index to not be droppable as DROP INDEX
would look for dependencies in a partition tree, which do not exist
anymore after the swap phase is done.
The fix here is simple: when swapping the old and new indexes, make sure
that pg_class.relispartition is correctly switched, similarly to what is
done for the index name.
Reported-by: Justin Pryzby
Author: Michael Paquier
Discussion: https://postgr.es/m/20191015164047.GA22729@telsasoft.com
Backpatch-through: 12
When swapping the dependencies of the old and new indexes, the code has
been correctly switching all links in pg_depend from the old to the new
index for both referencing and referenced entries. However it forgot
the fact that the new index may itself have existing entries in
pg_depend, like references to the parent table attributes. This
resulted in duplicated entries in pg_depend after running REINDEX
CONCURRENTLY.
Fix this problem by removing any existing entries in pg_depend on the
new index before switching the dependencies of the old index to the new
one. More regression tests are added to check the consistency of
entries in pg_depend for indexes, including partition indexes.
Author: Michael Paquier
Discussion: https://postgr.es/m/20191025064318.GF8671@paquier.xyz
Backpatch-through: 12
text_pattern_ops and its siblings can't be used with nondeterministic
collations, because they use the text_eq operator which will not behave
as bitwise equality if applied with a nondeterministic collation. The
initial implementation of that restriction was to insert a run-time test
in the related comparison functions, but that is inefficient, may throw
misleading errors, and will throw errors in some cases that would work.
It seems sufficient to just prevent the combination during CREATE INDEX,
so do that instead.
Lacking any better way to identify the opclasses involved, we need to
hard-wire tests for them, which requires hand-assigned values for their
OIDs, which forces a catversion bump because they previously had OIDs
that would be assigned automatically. That's slightly annoying in the
v12 branch, but fortunately we're not at rc1 yet, so just do it.
Back-patch to v12 where nondeterministic collations were added.
In passing, run make reformat-dat-files, which found some unrelated
whitespace issues (slightly different ones in HEAD and v12).
Peter Eisentraut, with small corrections by me
Discussion: https://postgr.es/m/22566.1568675619@sss.pgh.pa.us
The progress state was being clobbered once the first index completed
being rebuilt, causing the final phases of the operation not show
anything in the progress view. This was inadvertently broken in
03f9e5cba0, which added progress tracking for REINDEX.
(The reason this bugfix is this small is that I had already noticed this
problem when writing monitoring for CREATE INDEX, and had already worked
around it, as can be seen in discussion starting at
https://postgr.es/m/20190329150218.GA25010@alvherre.pgsql Fixing the
problem is just a matter of fixing one place touched by the REINDEX
monitoring.)
Reported by: Álvaro Herrera
Author: Álvaro Herrera
Discussion: https://postgr.es/m/20190801184333.GA21369@alvherre.pgsql
This portion of the code got forgotten in 7cce159 which has introduced a
new routine to build this node, and this finishes the unification of the
places where IndexInfo is initialized.
Author: Michael Paquier
Discussion: https://postgr.es/m/20190801041322.GA3435@paquier.xyz
When copying the definition of an index rebuilt concurrently for the new
entry, the index information was taken directly from the old index using
the relation cache. In this case, predicates and expressions have
some post-processing to prepare things for the planner, which loses some
information including the collations added in any of them.
This inconsistency can cause issues when attempting for example a table
rewrite, and makes the new indexes rebuilt concurrently inconsistent
with the old entries.
In order to fix the problem, fetch expressions and predicates directly
from the catalog of the old entry, and fill in IndexInfo for the new
index with that. This makes the process more consistent with
DefineIndex(), and the code is refactored with the addition of a routine
to create an IndexInfo node.
Reported-by: Manuel Rigger
Author: Michael Paquier
Discussion: https://postgr.es/m/CA+u7OA5Hp0ra235F3czPom_FyAd-3+XwSJmX95r1+sRPOJc9VQ@mail.gmail.com
Backpatch-through: 12
Originally, Postgres Lists were a more or less exact reimplementation of
Lisp lists, which consist of chains of separately-allocated cons cells,
each having a value and a next-cell link. We'd hacked that once before
(commit d0b4399d8) to add a separate List header, but the data was still
in cons cells. That makes some operations -- notably list_nth() -- O(N),
and it's bulky because of the next-cell pointers and per-cell palloc
overhead, and it's very cache-unfriendly if the cons cells end up
scattered around rather than being adjacent.
In this rewrite, we still have List headers, but the data is in a
resizable array of values, with no next-cell links. Now we need at
most two palloc's per List, and often only one, since we can allocate
some values in the same palloc call as the List header. (Of course,
extending an existing List may require repalloc's to enlarge the array.
But this involves just O(log N) allocations not O(N).)
Of course this is not without downsides. The key difficulty is that
addition or deletion of a list entry may now cause other entries to
move, which it did not before.
For example, that breaks foreach() and sister macros, which historically
used a pointer to the current cons-cell as loop state. We can repair
those macros transparently by making their actual loop state be an
integer list index; the exposed "ListCell *" pointer is no longer state
carried across loop iterations, but is just a derived value. (In
practice, modern compilers can optimize things back to having just one
loop state value, at least for simple cases with inline loop bodies.)
In principle, this is a semantics change for cases where the loop body
inserts or deletes list entries ahead of the current loop index; but
I found no such cases in the Postgres code.
The change is not at all transparent for code that doesn't use foreach()
but chases lists "by hand" using lnext(). The largest share of such
code in the backend is in loops that were maintaining "prev" and "next"
variables in addition to the current-cell pointer, in order to delete
list cells efficiently using list_delete_cell(). However, we no longer
need a previous-cell pointer to delete a list cell efficiently. Keeping
a next-cell pointer doesn't work, as explained above, but we can improve
matters by changing such code to use a regular foreach() loop and then
using the new macro foreach_delete_current() to delete the current cell.
(This macro knows how to update the associated foreach loop's state so
that no cells will be missed in the traversal.)
There remains a nontrivial risk of code assuming that a ListCell *
pointer will remain good over an operation that could now move the list
contents. To help catch such errors, list.c can be compiled with a new
define symbol DEBUG_LIST_MEMORY_USAGE that forcibly moves list contents
whenever that could possibly happen. This makes list operations
significantly more expensive so it's not normally turned on (though it
is on by default if USE_VALGRIND is on).
There are two notable API differences from the previous code:
* lnext() now requires the List's header pointer in addition to the
current cell's address.
* list_delete_cell() no longer requires a previous-cell argument.
These changes are somewhat unfortunate, but on the other hand code using
either function needs inspection to see if it is assuming anything
it shouldn't, so it's not all bad.
Programmers should be aware of these significant performance changes:
* list_nth() and related functions are now O(1); so there's no
major access-speed difference between a list and an array.
* Inserting or deleting a list element now takes time proportional to
the distance to the end of the list, due to moving the array elements.
(However, it typically *doesn't* require palloc or pfree, so except in
long lists it's probably still faster than before.) Notably, lcons()
used to be about the same cost as lappend(), but that's no longer true
if the list is long. Code that uses lcons() and list_delete_first()
to maintain a stack might usefully be rewritten to push and pop at the
end of the list rather than the beginning.
* There are now list_insert_nth...() and list_delete_nth...() functions
that add or remove a list cell identified by index. These have the
data-movement penalty explained above, but there's no search penalty.
* list_concat() and variants now copy the second list's data into
storage belonging to the first list, so there is no longer any
sharing of cells between the input lists. The second argument is
now declared "const List *" to reflect that it isn't changed.
This patch just does the minimum needed to get the new implementation
in place and fix bugs exposed by the regression tests. As suggested
by the foregoing, there's a fair amount of followup work remaining to
do.
Also, the ENABLE_LIST_COMPAT macros are finally removed in this
commit. Code using those should have been gone a dozen years ago.
Patch by me; thanks to David Rowley, Jesper Pedersen, and others
for review.
Discussion: https://postgr.es/m/11587.1550975080@sss.pgh.pa.us
As part of REINDEX CONCURRENTLY, this formerly internal-only error
message becomes potentially user-visible (see regression tests), so
change from errmsg_internal() to errmsg(), and update comment.
This is still using the 2.0 version of pg_bsd_indent.
I thought it would be good to commit this separately,
so as to document the differences between 2.0 and 2.1 behavior.
Discussion: https://postgr.es/m/16296.1558103386@sss.pgh.pa.us
Commits 3dbb317d3 et al failed under CLOBBER_CACHE_ALWAYS testing.
Investigation showed that to reindex pg_class_oid_index, we must
suppress accesses to the index (via SetReindexProcessing) before we call
RelationSetNewRelfilenode, or at least before we do CommandCounterIncrement
therein; otherwise, relcache reloads happening within the CCI may try to
fetch pg_class rows using the index's new relfilenode value, which is as
yet an empty file.
Of course, the point of 3dbb317d3 was that that ordering didn't work
either, because then RelationSetNewRelfilenode's own update of the index's
pg_class row cannot access the index, should it need to.
There are various ways we might have got around that, but Andres Freund
came up with a brilliant solution: for a mapped index, we can really just
skip the pg_class update altogether. The only fields it was actually
changing were relpages etc, but it was just setting them to zeroes which
is useless make-work. (Correct new values will be installed at the end
of index build.) All pg_class indexes are mapped and probably always will
be, so this eliminates the problem by removing work rather than adding it,
always a pleasant outcome. Having taught RelationSetNewRelfilenode to do
it that way, we can revert the code reordering in reindex_index. (But
I left the moved setup code where it was; there seems no reason why it
has to run without use of the old index. If you're trying to fix a
busted pg_class index, you'll have had to disable system index use
altogether to get this far.)
Moreover, this means we don't need RelationSetIndexList at all, because
reindex_relation's hacking to make "REINDEX TABLE pg_class" work is
likewise now unnecessary. We'll leave that code in place in the back
branches, but a follow-on patch will remove it in HEAD.
In passing, do some minor cleanup for commit 5c1560606 (in HEAD only),
notably removing a duplicate newrnode assignment.
Patch by me, using a core idea due to Andres Freund. Back-patch to all
supported branches, as 3dbb317d3 was.
Discussion: https://postgr.es/m/28926.1556664156@sss.pgh.pa.us
When reindexing individual indexes on pg_class it was possible to
either trigger an assertion failure:
TRAP: FailedAssertion("!(!ReindexIsProcessingIndex(((index)->rd_id)))
That's because reindex_index() called SetReindexProcessing() - which
enables an asserts ensuring no index insertions happen into the index
- before calling RelationSetNewRelfilenode(). That not correct for
indexes on pg_class, because RelationSetNewRelfilenode() updates the
relevant pg_class row, which needs to update the indexes.
The are two reasons this wasn't noticed earlier. Firstly the bug
doesn't trigger when reindexing all of pg_class, as reindex_relation
has code "hiding" all yet-to-be-reindexed indexes. Secondly, the bug
only triggers when the the update to pg_class doesn't turn out to be a
HOT update - otherwise there's no index insertion to trigger the
bug. Most of the time there's enough space, making this bug hard to
trigger.
To fix, move RelationSetNewRelfilenode() to before the
SetReindexProcessing() (and, together with some other code, to outside
of the PG_TRY()).
To make sure the error checking intended by SetReindexProcessing() is
more robust, modify CatalogIndexInsert() to check
ReindexIsProcessingIndex() even when the update is a HOT update.
Also add a few regression tests for REINDEXing of system catalogs.
The last two improvements would have prevented some of the issues
fixed in 5c1560606d from being introduced in the first place.
Reported-By: Michael Paquier
Diagnosed-By: Tom Lane and Andres Freund
Author: Andres Freund
Reviewed-By: Tom Lane
Discussion: https://postgr.es/m/20190418011430.GA19133@paquier.xyz
Backpatch: 9.4-, the bug is present in all branches
Up to now, DefineIndex() was responsible for adding attnotnull constraints
to the columns of a primary key, in any case where it hadn't been
convenient for transformIndexConstraint() to mark those columns as
is_not_null. It (or rather its minion index_check_primary_key) did this
by executing an ALTER TABLE SET NOT NULL command for the target table.
The trouble with this solution is that if we're creating the index due
to ALTER TABLE ADD PRIMARY KEY, and the outer ALTER TABLE has additional
sub-commands, the inner ALTER TABLE's operations executed at the wrong
time with respect to the outer ALTER TABLE's operations. In particular,
the inner ALTER would perform a validation scan at a point where the
table's storage might be inconsistent with its catalog entries. (This is
on the hairy edge of being a security problem, but AFAICS it isn't one
because the inner scan would only be interested in the tuples' null
bitmaps.) This can result in unexpected failures, such as the one seen
in bug #15580 from Allison Kaptur.
To fix, let's remove the attempt to do SET NOT NULL from DefineIndex(),
reducing index_check_primary_key's role to verifying that the columns are
already not null. (It shouldn't ever see such a case, but it seems wise
to keep the check for safety.) Instead, make transformIndexConstraint()
generate ALTER TABLE SET NOT NULL subcommands to be executed ahead of
the ADD PRIMARY KEY operation in every case where it can't force the
column to be created already-not-null. This requires only minor surgery
in parse_utilcmd.c, and it makes for a much more satisfying spec for
transformIndexConstraint(): it's no longer having to take it on faith
that someone else will handle addition of NOT NULL constraints.
To make that work, we have to move the execution of AT_SetNotNull into
an ALTER pass that executes ahead of AT_PASS_ADD_INDEX. I moved it to
AT_PASS_COL_ATTRS, and put that after AT_PASS_ADD_COL to avoid failure
when the column is being added in the same command. This incidentally
fixes a bug in the only previous usage of AT_PASS_COL_ATTRS, for
AT_SetIdentity: it didn't work either for a newly-added column.
Playing around with this exposed a separate bug in ALTER TABLE ONLY ...
ADD PRIMARY KEY for partitioned tables. The intent of the ONLY modifier
in that context is to prevent doing anything that would require holding
lock for a long time --- but the implied SET NOT NULL would recurse to
the child partitions, and do an expensive validation scan for any child
where the column(s) were not already NOT NULL. To fix that, invent a
new ALTER subcommand AT_CheckNotNull that just insists that a child
column be already NOT NULL, and apply that, not AT_SetNotNull, when
recursing to children in this scenario. This results in a slightly laxer
definition of ALTER TABLE ONLY ... SET NOT NULL for partitioned tables,
too: that command will now work as long as all children are already NOT
NULL, whereas before it just threw up its hands if there were any
partitions.
In passing, clean up the API of generateClonedIndexStmt(): remove a
useless argument, ensure that the output argument is not left undefined,
update the header comment.
A small side effect of this change is that no-such-column errors in ALTER
TABLE ADD PRIMARY KEY now produce a different message that includes the
table name, because they are now detected by the SET NOT NULL step which
has historically worded its error that way. That seems fine to me, so
I didn't make any effort to avoid the wording change.
The basic bug #15580 is of very long standing, and these other bugs
aren't new in v12 either. However, this is a pretty significant change
in the way ALTER TABLE ADD PRIMARY KEY works. On balance it seems best
not to back-patch, at least not till we get some more confidence that
this patch has no new bugs.
Patch by me, but thanks to Jie Zhang for a preliminary version.
Discussion: https://postgr.es/m/15580-d1a6de5a3d65da51@postgresql.org
Discussion: https://postgr.es/m/1396E95157071C4EBBA51892C5368521017F2E6E63@G08CNEXMBPEKD02.g08.fujitsu.local
In case of a partition index, when swapping the old and new index, we
also need to attach the new index as a partition and detach the old
one. Also, to handle partition indexes, we not only need to change
dependencies referencing the index, but also dependencies of the index
referencing something else. The previous code did this only
specifically for a constraint, but we also need to do this for
partitioned indexes. So instead write a generic function that does it
for all dependencies.
Author: Michael Paquier <michael@paquier.xyz>
Author: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
Discussion: https://www.postgresql.org/message-id/flat/DF4PR8401MB11964EDB77C860078C343BEBEE5A0%40DF4PR8401MB1196.NAMPRD84.PROD.OUTLOOK.COM#154df1fedb735190a773481765f7b874
This uses the same infrastructure that the CREATE INDEX progress
reporting uses. Add a column to pg_stat_progress_create_index to
report the OID of the index being worked on. This was not necessary
for CREATE INDEX, but it's useful for REINDEX.
Also edit the phase descriptions a bit to be more consistent with the
source code comments.
Discussion: https://www.postgresql.org/message-id/ef6a6757-c36a-9e81-123f-13b19e36b7d7%402ndquadrant.com
This uses the progress reporting infrastructure added by c16dc1aca5,
adding support for CREATE INDEX and CREATE INDEX CONCURRENTLY.
There are two pieces to this: one is index-AM-agnostic, and the other is
AM-specific. The latter is fairly elaborate for btrees, including
reportage for parallel index builds and the separate phases that btree
index creation uses; other index AMs, which are much simpler in their
building procedures, have simplistic reporting only, but that seems
sufficient, at least for non-concurrent builds.
The index-AM-agnostic part is fairly complete, providing insight into
the CONCURRENTLY wait phases as well as block-based progress during the
index validation table scan. (The index validation index scan requires
patching each AM, which has not been included here.)
Reviewers: Rahila Syed, Pavan Deolasee, Tatsuro Yamada
Discussion: https://postgr.es/m/20181220220022.mg63bhk26zdpvmcj@alvherre.pgsql
This adds the CONCURRENTLY option to the REINDEX command. A REINDEX
CONCURRENTLY on a specific index creates a new index (like CREATE
INDEX CONCURRENTLY), then renames the old index away and the new index
in place and adjusts the dependencies, and then drops the old
index (like DROP INDEX CONCURRENTLY). The REINDEX command also has
the capability to run its other variants (TABLE, DATABASE) with the
CONCURRENTLY option (but not SYSTEM).
The reindexdb command gets the --concurrently option.
Author: Michael Paquier, Andreas Karlsson, Peter Eisentraut
Reviewed-by: Andres Freund, Fujii Masao, Jim Nasby, Sergei Kornilov
Discussion: https://www.postgresql.org/message-id/flat/60052986-956b-4478-45ed-8bd119e9b9cf%402ndquadrant.com#74948a1044c56c5e817a5050f554ddee
This moves the responsibility for:
- creating the storage necessary for a relation, including creating a
new relfilenode for a relation with existing storage
- non-transactional truncation of a relation
- VACUUM FULL / CLUSTER's rewrite of a table
below tableam.
This is fairly straight forward, with a bit of complexity smattered in
to move the computation of xid / multixid horizons below the AM, as
they don't make sense for every table AM.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
To support building indexes over tables of different AMs, the scans to
do so need to be routed through the table AM. While moving a fair
amount of code, nearly all the changes are just moving code to below a
callback.
Currently the range based interface wouldn't make much sense for non
block based table AMs. But that seems aceptable for now.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
ALTER INDEX .. ATTACH PARTITION fails if the partitioned table where the
index is defined contains more dropped columns than its partition, with
this message:
ERROR: incorrect attribute map
The cause was that one caller of CompareIndexInfo was passing the number
of attributes of the partition rather than the parent, which confused
the length check. Repair.
This can cause pg_upgrade to fail when used on such a database. Leave
some more objects around after regression tests, so that the case is
detected by pg_upgrade test suite.
Remove some spurious empty lines noticed while looking for other cases
of the same problem.
Discussion: https://postgr.es/m/20190326213924.GA2322@alvherre.pgsql
This uses the same progress reporting infrastructure added in commit
c16dc1aca5 and extends it to these
additional cases. We lack the ability to track the internal progress
of sorts and index builds so the information reported is
coarse-grained for some parts of the operation, but it still seems
like a significant improvement over having nothing at all.
Tatsuro Yamada, reviewed by Thomas Munro, Masahiko Sawada, Michael
Paquier, Jeff Janes, Alvaro Herrera, Rafia Sabih, and by me. A fair
amount of polishing also by me.
Discussion: http://postgr.es/m/59A77072.3090401@lab.ntt.co.jp
When DefineIndex recurses to create constraints on partitions, it needs
to use the value returned by index_constraint_create to set up partition
dependencies. However, in the course of fixing the DEPENDENCY_INTERNAL_AUTO
mess, commit 1d92a0c9f7 introduced some code to that function that
clobbered the return value, causing the recorded OID to be of the wrong
object. Close examination of pg_depend after creating the tables leads
to indescribable objects :-( My sin (in commit bdc3d7fa23, while
preparing for DDL deparsing in event triggers) was to use a variable
name for the return value that's typically used for throwaway objects in
dependency-setting calls ("referenced"). Fix by changing the variable
names to match extended practice (the return value is "myself" rather
than "referenced".)
The pg_upgrade test notices the problem (in an indirect way: the pg_dump
outputs are in different order), but only if you create the objects in a
specific way that wasn't being used in the existing tests. Add a stanza
to leave some objects around that shows the bug.
Catversion bump because preexisting databases might have bogus pg_depend
entries.
Discussion: https://postgr.es/m/20190318204235.GA30360@alvherre.pgsql
Too allow table accesses to be not directly dependent on heap, several
new abstractions are needed. Specifically:
1) Heap scans need to be generalized into table scans. Do this by
introducing TableScanDesc, which will be the "base class" for
individual AMs. This contains the AM independent fields from
HeapScanDesc.
The previous heap_{beginscan,rescan,endscan} et al. have been
replaced with a table_ version.
There's no direct replacement for heap_getnext(), as that returned
a HeapTuple, which is undesirable for a other AMs. Instead there's
table_scan_getnextslot(). But note that heap_getnext() lives on,
it's still used widely to access catalog tables.
This is achieved by new scan_begin, scan_end, scan_rescan,
scan_getnextslot callbacks.
2) The portion of parallel scans that's shared between backends need
to be able to do so without the user doing per-AM work. To achieve
that new parallelscan_{estimate, initialize, reinitialize}
callbacks are introduced, which operate on a new
ParallelTableScanDesc, which again can be subclassed by AMs.
As it is likely that several AMs are going to be block oriented,
block oriented callbacks that can be shared between such AMs are
provided and used by heap. table_block_parallelscan_{estimate,
intiialize, reinitialize} as callbacks, and
table_block_parallelscan_{nextpage, init} for use in AMs. These
operate on a ParallelBlockTableScanDesc.
3) Index scans need to be able to access tables to return a tuple, and
there needs to be state across individual accesses to the heap to
store state like buffers. That's now handled by introducing a
sort-of-scan IndexFetchTable, which again is intended to be
subclassed by individual AMs (for heap IndexFetchHeap).
The relevant callbacks for an AM are index_fetch_{end, begin,
reset} to create the necessary state, and index_fetch_tuple to
retrieve an indexed tuple. Note that index_fetch_tuple
implementations need to be smarter than just blindly fetching the
tuples for AMs that have optimizations similar to heap's HOT - the
currently alive tuple in the update chain needs to be fetched if
appropriate.
Similar to table_scan_getnextslot(), it's undesirable to continue
to return HeapTuples. Thus index_fetch_heap (might want to rename
that later) now accepts a slot as an argument. Core code doesn't
have a lot of call sites performing index scans without going
through the systable_* API (in contrast to loads of heap_getnext
calls and working directly with HeapTuples).
Index scans now store the result of a search in
IndexScanDesc->xs_heaptid, rather than xs_ctup->t_self. As the
target is not generally a HeapTuple anymore that seems cleaner.
To be able to sensible adapt code to use the above, two further
callbacks have been introduced:
a) slot_callbacks returns a TupleTableSlotOps* suitable for creating
slots capable of holding a tuple of the AMs
type. table_slot_callbacks() and table_slot_create() are based
upon that, but have additional logic to deal with views, foreign
tables, etc.
While this change could have been done separately, nearly all the
call sites that needed to be adapted for the rest of this commit
also would have been needed to be adapted for
table_slot_callbacks(), making separation not worthwhile.
b) tuple_satisfies_snapshot checks whether the tuple in a slot is
currently visible according to a snapshot. That's required as a few
places now don't have a buffer + HeapTuple around, but a
slot (which in heap's case internally has that information).
Additionally a few infrastructure changes were needed:
I) SysScanDesc, as used by systable_{beginscan, getnext} et al. now
internally uses a slot to keep track of tuples. While
systable_getnext() still returns HeapTuples, and will so for the
foreseeable future, the index API (see 1) above) now only deals with
slots.
The remainder, and largest part, of this commit is then adjusting all
scans in postgres to use the new APIs.
Author: Andres Freund, Haribabu Kommi, Alvaro Herrera
Discussion:
https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.dehttps://postgr.es/m/20160812231527.GA690404@alvherre.pgsql
This introduces the concept of table access methods, i.e. CREATE
ACCESS METHOD ... TYPE TABLE and
CREATE TABLE ... USING (storage-engine).
No table access functionality is delegated to table AMs as of this
commit, that'll be done in following commits.
Subsequent commits will incrementally abstract table access
functionality to be routed through table access methods. That change
is too large to be reviewed & committed at once, so it'll be done
incrementally.
Docs will be updated at the end, as adding them incrementally would
likely make them less coherent, and definitely is a lot more work,
without a lot of benefit.
Table access methods are specified similar to index access methods,
i.e. pg_am.amhandler returns, as INTERNAL, a pointer to a struct with
callbacks. In contrast to index AMs that struct needs to live as long
as a backend, typically that's achieved by just returning a pointer to
a constant struct.
Psql's \d+ now displays a table's access method. That can be disabled
with HIDE_TABLEAM=true, which is mainly useful so regression tests can
be run against different AMs. It's quite possible that this behaviour
still needs to be fine tuned.
For now it's not allowed to set a table AM for a partitioned table, as
we've not resolved how partitions would inherit that. Disallowing
allows us to introduce, if we decide that's the way forward, such a
behaviour without a compatibility break.
Catversion bumped, to add the heap table AM and references to it.
Author: Haribabu Kommi, Andres Freund, Alvaro Herrera, Dimitri Golgov and others
Discussion:
https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.dehttps://postgr.es/m/20160812231527.GA690404@alvherre.pgsqlhttps://postgr.es/m/20190107235616.6lur25ph22u5u5av@alap3.anarazel.dehttps://postgr.es/m/20190304234700.w5tmhducs5wxgzls@alap3.anarazel.de
The original setup for dependencies of partitioned objects had
serious problems:
1. It did not verify that a drop cascading to a partition-child object
also cascaded to at least one of the object's partition parents. Now,
normally a child object would share all its dependencies with one or
another parent (e.g. a child index's opclass dependencies would be shared
with the parent index), so that this oversight is usually harmless.
But if some dependency failed to fit this pattern, the child could be
dropped while all its parents remain, creating a logically broken
situation. (It's easy to construct artificial cases that break it,
such as attaching an unrelated extension dependency to the child object
and then dropping the extension. I'm not sure if any less-artificial
cases exist.)
2. Management of partition dependencies during ATTACH/DETACH PARTITION
was complicated and buggy; for example, after detaching a partition
table it was possible to create cases where a formerly-child index
should be dropped and was not, because the correct set of dependencies
had not been reconstructed.
Less seriously, because multiple partition relationships were
represented identically in pg_depend, there was an order-of-traversal
dependency on which partition parent was cited in error messages.
We also had some pre-existing order-of-traversal hazards for error
messages related to internal and extension dependencies. This is
cosmetic to users but causes testing problems.
To fix#1, add a check at the end of the partition tree traversal
to ensure that at least one partition parent got deleted. To fix#2,
establish a new policy that partition dependencies are in addition to,
not instead of, a child object's usual dependencies; in this way
ATTACH/DETACH PARTITION need not cope with adding or removing the
usual dependencies.
To fix the cosmetic problem, distinguish between primary and secondary
partition dependency entries in pg_depend, by giving them different
deptypes. (They behave identically except for having different
priorities for being cited in error messages.) This means that the
former 'I' dependency type is replaced with new 'P' and 'S' types.
This also fixes a longstanding bug that after handling an internal
dependency by recursing to the owning object, findDependentObjects
did not verify that the current target was now scheduled for deletion,
and did not apply the current recursion level's objflags to it.
Perhaps that should be back-patched; but in the back branches it
would only matter if some concurrent transaction had removed the
internal-linkage pg_depend entry before the recursive call found it,
or the recursive call somehow failed to find it, both of which seem
unlikely.
Catversion bump because the contents of pg_depend change for
partitioning relationships.
Patch HEAD only. It's annoying that we're not fixing #2 in v11,
but there seems no practical way to do so given that the problem
is exactly a poor choice of what entries to put in pg_depend.
We can't really fix that while staying compatible with what's
in pg_depend in existing v11 installations.
Discussion: https://postgr.es/m/CAH2-Wzkypv1R+teZrr71U23J578NnTBt2X8+Y=Odr4pOdW1rXg@mail.gmail.com
We can't allow these pseudo-types to be used as table column types,
because storing an anonymous record value in a table would result
in data that couldn't be understood by other sessions. However,
it seems like there's no harm in allowing the case in a column
definition list that's specifying what a function-returning-record
returns. The data involved is all local to the current session,
so we should be just as able to resolve its actual tuple type as
we are for the function-returning-record's top-level tuple output.
Elvis Pranskevichus, with cosmetic changes by me
Discussion: https://postgr.es/m/11038447.kQ5A9Uj5xi@hammer.magicstack.net
Create a new header optimizer/optimizer.h, which exposes just the
planner functions that can be used "at arm's length", without need
to access Paths or the other planner-internal data structures defined
in nodes/relation.h. This is intended to provide the whole planner
API seen by most of the rest of the system; although FDWs still need
to use additional stuff, and more thought is also needed about just
what selfuncs.c should rely on.
The main point of doing this now is to limit the amount of new
#include baggage that will be needed by "planner support functions",
which I expect to introduce later, and which will be in relevant
datatype modules rather than anywhere near the planner.
This commit just moves relevant declarations into optimizer.h from
other header files (a couple of which go away because everything
got moved), and adjusts #include lists to match. There's further
cleanup that could be done if we want to decide that some stuff
being exposed by optimizer.h doesn't belong in the planner at all,
but I'll leave that for another day.
Discussion: https://postgr.es/m/11460.1548706639@sss.pgh.pa.us
Given these routines are heap specific, and that there will be more
generic visibility support in via table AM, it makes sense to move the
prototypes to heapam.h (routines like HeapTupleSatisfiesVacuum will
not be exposed in a generic fashion, because they are too storage
specific).
Similarly, the code in tqual.c is specific to heap, so moving it into
access/heap/ makes sense.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
This reverts commit c203d6cf8 and some follow-on fixes, completing the
task begun in commit 5d28c9bd7. If that feature is ever resurrected,
the code will look quite a bit different from this, so it seems best
to start from a clean slate.
The v11 branch is not touched; in that branch, the recheck_on_update
storage option remains present, but nonfunctional and undocumented.
Discussion: https://postgr.es/m/20190114223409.3tcvejfhlvbucrv5@alap3.anarazel.de
heapam.h previously was included in a number of widely used
headers (e.g. execnodes.h, indirectly in executor.h, ...). That's
problematic on its own, as heapam.h contains a lot of low-level
details that don't need to be exposed that widely, but becomes more
problematic with the upcoming introduction of pluggable table storage
- it seems inappropriate for heapam.h to be included that widely
afterwards.
heapam.h was largely only included in other headers to get the
HeapScanDesc typedef (which was defined in heapam.h, even though
HeapScanDescData is defined in relscan.h). The better solution here
seems to be to just use the underlying struct (forward declared where
necessary). Similar for BulkInsertState.
Another problem was that LockTupleMode was used in executor.h - parts
of the file tried to cope without heapam.h, but due to the fact that
it indirectly included it, several subsequent violations of that goal
were not not noticed. We could just reuse the approach of declaring
parameters as int, but it seems nicer to move LockTupleMode to
lockoptions.h - that's not a perfect location, but also doesn't seem
bad.
As a number of files relied on implicitly included heapam.h, a
significant number of files grew an explicit include. It's quite
probably that a few external projects will need to do the same.
Author: Andres Freund
Reviewed-By: Alvaro Herrera
Discussion: https://postgr.es/m/20190114000701.y4ttcb74jpskkcfb@alap3.anarazel.de
Previously tables declared WITH OIDS, including a significant fraction
of the catalog tables, stored the oid column not as a normal column,
but as part of the tuple header.
This special column was not shown by default, which was somewhat odd,
as it's often (consider e.g. pg_class.oid) one of the more important
parts of a row. Neither pg_dump nor COPY included the contents of the
oid column by default.
The fact that the oid column was not an ordinary column necessitated a
significant amount of special case code to support oid columns. That
already was painful for the existing, but upcoming work aiming to make
table storage pluggable, would have required expanding and duplicating
that "specialness" significantly.
WITH OIDS has been deprecated since 2005 (commit ff02d0a05280e0).
Remove it.
Removing includes:
- CREATE TABLE and ALTER TABLE syntax for declaring the table to be
WITH OIDS has been removed (WITH (oids[ = true]) will error out)
- pg_dump does not support dumping tables declared WITH OIDS and will
issue a warning when dumping one (and ignore the oid column).
- restoring an pg_dump archive with pg_restore will warn when
restoring a table with oid contents (and ignore the oid column)
- COPY will refuse to load binary dump that includes oids.
- pg_upgrade will error out when encountering tables declared WITH
OIDS, they have to be altered to remove the oid column first.
- Functionality to access the oid of the last inserted row (like
plpgsql's RESULT_OID, spi's SPI_lastoid, ...) has been removed.
The syntax for declaring a table WITHOUT OIDS (or WITH (oids = false)
for CREATE TABLE) is still supported. While that requires a bit of
support code, it seems unnecessary to break applications / dumps that
do not use oids, and are explicit about not using them.
The biggest user of WITH OID columns was postgres' catalog. This
commit changes all 'magic' oid columns to be columns that are normally
declared and stored. To reduce unnecessary query breakage all the
newly added columns are still named 'oid', even if a table's column
naming scheme would indicate 'reloid' or such. This obviously
requires adapting a lot code, mostly replacing oid access via
HeapTupleGetOid() with access to the underlying Form_pg_*->oid column.
The bootstrap process now assigns oids for all oid columns in
genbki.pl that do not have an explicit value (starting at the largest
oid previously used), only oids assigned later by oids will be above
FirstBootstrapObjectId. As the oid column now is a normal column the
special bootstrap syntax for oids has been removed.
Oids are not automatically assigned during insertion anymore, all
backend code explicitly assigns oids with GetNewOidWithIndex(). For
the rare case that insertions into the catalog via SQL are called for
the new pg_nextoid() function can be used (which only works on catalog
tables).
The fact that oid columns on system tables are now normal columns
means that they will be included in the set of columns expanded
by * (i.e. SELECT * FROM pg_class will now include the table's oid,
previously it did not). It'd not technically be hard to hide oid
column by default, but that'd mean confusing behavior would either
have to be carried forward forever, or it'd cause breakage down the
line.
While it's not unlikely that further adjustments are needed, the
scope/invasiveness of the patch makes it worthwhile to get merge this
now. It's painful to maintain externally, too complicated to commit
after the code code freeze, and a dependency of a number of other
patches.
Catversion bump, for obvious reasons.
Author: Andres Freund, with contributions by John Naylor
Discussion: https://postgr.es/m/20180930034810.ywp2c7awz7opzcfr@alap3.anarazel.de
This commit completes the work prepared in 1a0586de36, splitting the
old TupleTableSlot implementation (which could store buffer, heap,
minimal and virtual slots) into four different slot types. As
described in the aforementioned commit, this is done with the goal of
making tuple table slots extensible, to allow for pluggable table
access methods.
To achieve runtime extensibility for TupleTableSlots, operations on
slots that can differ between types of slots are performed using the
TupleTableSlotOps struct provided at slot creation time. That
includes information from the size of TupleTableSlot struct to be
allocated, initialization, deforming etc. See the struct's definition
for more detailed information about callbacks TupleTableSlotOps.
I decided to rename TTSOpsBufferTuple to TTSOpsBufferHeapTuple and
ExecCopySlotTuple to ExecCopySlotHeapTuple, as that seems more
consistent with other naming introduced in recent patches.
There's plenty optimization potential in the slot implementation, but
according to benchmarking the state after this commit has similar
performance characteristics to before this set of changes, which seems
sufficient.
There's a few changes in execReplication.c that currently need to poke
through the slot abstraction, that'll be repaired once the pluggable
storage patchset provides the necessary infrastructure.
Author: Andres Freund and Ashutosh Bapat, with changes by Amit Khandekar
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
Upcoming work intends to allow pluggable ways to introduce new ways of
storing table data. Accessing those table access methods from the
executor requires TupleTableSlots to be carry tuples in the native
format of such storage methods; otherwise there'll be a significant
conversion overhead.
Different access methods will require different data to store tuples
efficiently (just like virtual, minimal, heap already require fields
in TupleTableSlot). To allow that without requiring additional pointer
indirections, we want to have different structs (embedding
TupleTableSlot) for different types of slots. Thus different types of
slots are needed, which requires adapting creators of slots.
The slot that most efficiently can represent a type of tuple in an
executor node will often depend on the type of slot a child node
uses. Therefore we need to track the type of slot is returned by
nodes, so parent slots can create slots based on that.
Relatedly, JIT compilation of tuple deforming needs to know which type
of slot a certain expression refers to, so it can create an
appropriate deforming function for the type of tuple in the slot.
But not all nodes will only return one type of slot, e.g. an append
node will potentially return different types of slots for each of its
subplans.
Therefore add function that allows to query the type of a node's
result slot, and whether it'll always be the same type (whether it's
fixed). This can be queried using ExecGetResultSlotOps().
The scan, result, inner, outer type of slots are automatically
inferred from ExecInitScanTupleSlot(), ExecInitResultSlot(),
left/right subtrees respectively. If that's not correct for a node,
that can be overwritten using new fields in PlanState.
This commit does not introduce the actually abstracted implementation
of different kind of TupleTableSlots, that will be left for a followup
commit. The different types of slots introduced will, for now, still
use the same backing implementation.
While this already partially invalidates the big comment in
tuptable.h, it seems to make more sense to update it later, when the
different TupleTableSlot implementations actually exist.
Author: Ashutosh Bapat and Andres Freund, with changes by Amit Khandekar
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
Michael Paquier
Bruce Momjian
Tom Lane
Amit Kapila
Peter Eisentraut
Alvaro Herrera
Andres Freund
Robert Haas
Peter Eisentraut