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
In the wake of commit 5b9312378, there's no particular reason
for this restriction (previously, it was problematic because of
the implied rowtype reference). A simple constraint on a whole-row
Var probably isn't that useful, but conceivably somebody would want
to pass one to a function that extracts a partitioning key. Besides
which, we're expending much more code to enforce the restriction than
we save by having it, since the latter quantity is now zero.
So drop the restriction.
Amit Langote
Discussion: https://postgr.es/m/CA+HiwqFUzjfj9HEsJtYWcr1SgQ_=iCAvQ=O2Sx6aQxoDu4OiHw@mail.gmail.com
Formerly the rd_partkey and rd_partdesc data structures were always
populated immediately when a relcache entry was built or rebuilt.
This patch changes things so that they are populated only when they
are first requested. (Hence, callers *must* now always use
RelationGetPartitionKey or RelationGetPartitionDesc; just fetching
the pointer directly is no longer acceptable.)
This seems to have some performance benefits, but the main reason to do
it is that it eliminates a recursive-reload failure that occurs if the
partkey or partdesc expressions contain any references to the relation's
rowtype (as discovered by Amit Langote). In retrospect, since loading
these data structures might result in execution of nearly-arbitrary code
via eval_const_expressions, it was a dumb idea to require that to happen
during relcache entry rebuild.
Also, fix things so that old copies of a relcache partition descriptor
will be dropped when the cache entry's refcount goes to zero. In the
previous coding it was possible for such copies to survive for the
lifetime of the session, as I'd complained of in a previous discussion.
(This management technique still isn't perfect, but it's better than
before.) Improve the commentary explaining how that works and why
it's safe to hand out direct pointers to these relcache substructures.
In passing, improve RelationBuildPartitionDesc by using the same
memory-context-parent-swap approach used by RelationBuildPartitionKey,
thereby making it less dependent on strong assumptions about what
partition_bounds_copy does. Avoid doing get_rel_relkind in the
critical section, too.
Patch by Amit Langote and Tom Lane; Robert Haas deserves some credit
for prior work in the area, too. Although this is a pre-existing
problem, no back-patch: the patch seems too invasive to be safe to
back-patch, and the bug it fixes is a corner case that seems
relatively unlikely to cause problems in the field.
Discussion: https://postgr.es/m/CA+HiwqFUzjfj9HEsJtYWcr1SgQ_=iCAvQ=O2Sx6aQxoDu4OiHw@mail.gmail.com
Discussion: https://postgr.es/m/CA+TgmoY3bRmGB6-DUnoVy5fJoreiBJ43rwMrQRCdPXuKt4Ykaw@mail.gmail.com
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
When maintaining or merging patches, one of the most common sources
for conflicts are the list of objects in makefiles. Especially when
the split across lines has been changed on both sides, which is
somewhat common due to attempting to stay below 80 columns, those
conflicts are unnecessarily laborious to resolve.
By splitting, and alphabetically sorting, OBJS style lines into one
object per line, conflicts should be less frequent, and easier to
resolve when they still occur.
Author: Andres Freund
Discussion: https://postgr.es/m/20191029200901.vww4idgcxv74cwes@alap3.anarazel.de
Commit 6f6b99d13 stuck an INFO message into the fast path for
checking partition constraints, for no very good reason except
that it made it easy for the regression tests to verify that
that path was taken. Assorted later patches did likewise,
increasing the unsuppressable-chatter level from ALTER TABLE
even more. This isn't good for the user experience, so let's
drop these messages down to DEBUG1 where they belong. So as
not to have a loss of test coverage, create a TAP test that
runs the relevant queries with client_min_messages = DEBUG1
and greps for the expected messages.
This testing method is a bit brute-force --- in particular,
it duplicates the execution of a fair amount of the core
create_table and alter_table tests. We experimented with
other solutions, but running any significant amount of
standard testing with client_min_messages = DEBUG1 seems
to have a lot of output-stability pitfalls, cf commits
bbb96c370 and 5655565c0. Possibly at some point we'll look
into whether we can reduce the amount of test duplication.
Backpatch into v12, because some of these messages are new
in v12 and we don't really want to ship it that way.
Sergei Kornilov
Discussion: https://postgr.es/m/81911511895540@web58j.yandex.ru
Discussion: https://postgr.es/m/4859321552643736@myt5-02b80404fd9e.qloud-c.yandex.net
For most uses of acl.h the details of how "Acl" internally looks like
are irrelevant. It might make sense to move a lot of the
implementation details into a separate header at a later point.
The main motivation of this change is to avoid including fmgr.h (via
array.h, which needs it for exposed structs) in a lot of files that
otherwise don't need it. A subsequent commit will remove the fmgr.h
include from a lot of files.
Directly include utils/array.h and utils/expandeddatum.h from the
files that need them, but previously included them indirectly, via
acl.h.
Author: Andres Freund
Discussion: https://postgr.es/m/20190803193733.g3l3x3o42uv4qj7l@alap3.anarazel.de
We only need to invoke constraint exclusion on partitioned tables when
they are a partition, and they themselves contain a default partition;
it's not necessary otherwise, and it's expensive, so avoid it. Also, we
were trying once for each clause separately, but we can do it for all
the clauses at once.
While at it, centralize setting of RelOptInfo->partition_qual instead of
computing it in slightly different ways in different places.
Per complaints from Simon Riggs about 4e85642d935e; reviewed by Yuzuko
Hosoya, Kyotaro Horiguchi.
Author: Amit Langote. I (Álvaro) again mangled the patch somewhat.
Discussion: https://postgr.es/m/CANP8+j+tMCY=nEcQeqQam85=uopLBtX-2vHiLD2bbp7iQQUKpA@mail.gmail.com
In the wake of commit 1cff1b95a, the result of list_concat no longer
shares the ListCells of the second input. Therefore, we can replace
"list_concat(x, list_copy(y))" with just "list_concat(x, y)".
To improve call sites that were list_copy'ing the first argument,
or both arguments, invent "list_concat_copy()" which produces a new
list sharing no ListCells with either input. (This is a bit faster
than "list_concat(list_copy(x), y)" because it makes the result list
the right size to start with.)
In call sites that were not list_copy'ing the second argument, the new
semantics mean that we are usually leaking the second List's storage,
since typically there is no remaining pointer to it. We considered
inventing another list_copy variant that would list_free the second
input, but concluded that for most call sites it isn't worth worrying
about, given the relative compactness of the new List representation.
(Note that in cases where such leakage would happen, the old code
already leaked the second List's header; so we're only discussing
the size of the leak not whether there is one. I did adjust two or
three places that had been troubling to free that header so that
they manually free the whole second List.)
Patch by me; thanks to David Rowley for review.
Discussion: https://postgr.es/m/11587.1550975080@sss.pgh.pa.us
We were applying constraint exclusion on the partition constraint when
generating pruning steps for a clause, but only for the rather
restricted situation of them being boolean OR operators; however it is
possible to have differently shaped clauses that also benefit from
constraint exclusion. This applies particularly to the default
partition since their constraints are in essence a long list of OR'ed
subclauses ... but it applies to other cases too. So in certain cases
we're scanning partitions that we don't need to.
Remove the specialized code in OR clauses, and add a generally
applicable test of the clause refuting the partition constraint; mark
the whole pruning operation as contradictory if it hits.
This has the unwanted side-effect of testing some (most? all?)
constraints more than once if constraint_exclusion=on. That seems
unavoidable as far as I can tell without some additional work, but
that's not the recommended setting for that parameter anyway.
However, because this imposes additional processing cost for all
queries using partitioned tables, I decided not to backpatch this
change.
Author: Amit Langote, Yuzuko Hosoya, Álvaro Herrera
Reviewers: Shawn Wang, Thibaut Madeleine, Yoshikazu Imai, Kyotaro
Horiguchi; they were also uncredited reviewers for commit 489247b0e6.
Discussion: https://postgr.es/m/9bb31dfe-b0d0-53f3-3ea6-e64b811424cf@lab.ntt.co.jp
When querying a partitioned table containing a default partition, we
were wrongly deciding to include it in the scan too early in the
process, failing to exclude it in some cases. If we reinterpret the
PruneStepResult.scan_default flag slightly, we can do a better job at
detecting that it can be excluded. The change is that we avoid setting
the flag for that pruning step unless the step absolutely requires the
default partition to be scanned (in contrast with the previous
arrangement, which was to set it unless the step was able to prune it).
So get_matching_partitions() must explicitly check the partition that
each returned bound value corresponds to in order to determine whether
the default one needs to be included, rather than relying on the flag
from the final step result.
Author: Yuzuko Hosoya <hosoya.yuzuko@lab.ntt.co.jp>
Reviewed-by: Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>
Discussion: https://postgr.es/m/00e601d4ca86$932b8bc0$b982a340$@lab.ntt.co.jp
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
match_clause_to_partition_key incorrectly would return
PARTCLAUSE_UNSUPPORTED if a bool qual could not be matched to the current
partition key. This was a problem, as it causes the calling function to
discard the qual and not try to match it to any other partition key. If
there was another partition key which did match this qual, then the qual
would not be checked again and we could fail to prune some partitions.
The worst this could do was to cause partitions not to be pruned when they
could have been, so there was no danger of incorrect query results here.
Fix this by changing match_boolean_partition_clause to have it return a
PartClauseMatchStatus rather than a boolean value. This allows it to
communicate if the qual is unsupported or if it just does not match this
particular partition key, previously these two cases were treated the
same. Now, if match_clause_to_partition_key is unable to match the qual
to any other qual type then we can simply return the value from the
match_boolean_partition_clause call so that the calling function properly
treats the qual as either unmatched or unsupported.
Reported-by: Rares Salcudean
Reviewed-by: Amit Langote
Backpatch-through: 11 where partition pruning was introduced
Discussion: https://postgr.es/m/CAHp_FN2xwEznH6oyS0hNTuUUZKp5PvegcVv=Co6nBXJ+mC7Y5w@mail.gmail.com
We forgot to map column numbers to/from the default partition for
various operations, leading to valid cases failing with spurious
errors, such as
ERROR: attribute N of type some_partition has been dropped
It was also possible that the search for conflicting rows in the default
partition when attaching another partition would fail to detect some.
Secondarily, it was also possible that such a search should be skipped
(because the constraint was implied) but wasn't.
Fix all this by mapping column numbers when necessary.
Reported by: Daniel Wilches
Author: Amit Langote
Discussion: https://postgr.es/m/15873-8c61945d6b3ef87c@postgresql.org
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
Previously, gen_partprune_steps() always built executor pruning steps
using all suitable clauses, including those containing PARAM_EXEC
Params. This meant that the pruning steps were only completely safe
for executor run-time (scan start) pruning. To prune at executor
startup, we had to ignore the steps involving exec Params. But this
doesn't really work in general, since there may be logic changes
needed as well --- for example, pruning according to the last operator's
btree strategy is the wrong thing if we're not applying that operator.
The rules embodied in gen_partprune_steps() and its minions are
sufficiently complicated that tracking their incremental effects in
other logic seems quite impractical.
Short of a complete redesign, the only safe fix seems to be to run
gen_partprune_steps() twice, once to create executor startup pruning
steps and then again for run-time pruning steps. We can save a few
cycles however by noting during the first scan whether we rejected
any clauses because they involved exec Params --- if not, we don't
need to do the second scan.
In support of this, refactor the internal APIs in partprune.c to make
more use of passing information in the GeneratePruningStepsContext
struct, rather than as separate arguments.
This is, I hope, the last piece of our response to a bug report from
Alan Jackson. Back-patch to v11 where this code came in.
Discussion: https://postgr.es/m/FAD28A83-AC73-489E-A058-2681FA31D648@tvsquared.com
gen_prune_steps_from_opexps's notion of how to do this was overly
complicated and underly correct.
Per discussion of a report from Alan Jackson (though this fixes only one
aspect of that problem). Back-patch to v11 where this code came in.
Amit Langote
Discussion: https://postgr.es/m/FAD28A83-AC73-489E-A058-2681FA31D648@tvsquared.com
Cross-type comparison operators in a btree or hash opclass might be
only stable not immutable (this is true of timestamp vs. timestamptz
for example). partprune.c ignored this possibility and would perform
plan-time pruning with them anyway, possibly leading to wrong answers
if the environment changed between planning and execution.
To fix, teach gen_partprune_steps() to do things differently when
creating plan-time pruning steps vs. run-time pruning steps.
analyze_partkey_exprs() also needs an extra check, which is rather
annoying but now is not the time to restructure things enough to
avoid that.
While at it, simplify the logic for the plan-time case a little
by insisting that the comparison value be a Const and nothing else.
This relies on the assumption that eval_const_expressions will have
reduced any immutable expression to a Const; which is not quite
100% true, but certainly any case that comes up often enough to be
interesting should have simplification logic there.
Also improve a bunch of inadequate/obsolete/wrong comments.
Per discussion of a report from Alan Jackson (though this fixes only one
aspect of that problem). Back-patch to v11 where this code came in.
David Rowley, with some further hacking by me
Discussion: https://postgr.es/m/FAD28A83-AC73-489E-A058-2681FA31D648@tvsquared.com
The original coding of generate_partition_qual() just copied the list
of predicate expressions into the global CacheMemoryContext, making it
effectively impossible to clean up when the owning relcache entry is
destroyed --- the relevant code in RelationDestroyRelation() only managed
to free the topmost List header :-(. This resulted in a session-lifespan
memory leak whenever a table partition's relcache entry is rebuilt.
Fortunately, that's not normally a large data structure, and rebuilds
shouldn't occur all that often in production situations; but this is
still a bug worth fixing back to v10 where the code was introduced.
To fix, put the cached expression tree into its own small memory context,
as we do with other complicated substructures of relcache entries.
Also, deal more honestly with the case that a partition has an empty
partcheck list; while that probably isn't a case that's very interesting
for production use, it's legal.
In passing, clarify comments about how partitioning-related relcache
data structures are managed, and add some Asserts that we're not leaking
old copies when we overwrite these data fields.
Amit Langote and Tom Lane
Discussion: https://postgr.es/m/7961.1552498252@sss.pgh.pa.us
If we need ordered output from a scan of a partitioned table, but
the ordering matches the partition ordering, then we don't need to
use a MergeAppend to combine the pre-ordered per-partition scan
results: a plain Append will produce the same results. This
both saves useless comparison work inside the MergeAppend proper,
and allows us to start returning tuples after istarting up just
the first child node not all of them.
However, all is not peaches and cream, because if some of the
child nodes have high startup costs then there will be big
discontinuities in the tuples-returned-versus-elapsed-time curve.
The planner's cost model cannot handle that (yet, anyway).
If we model the Append's startup cost as being just the first
child's startup cost, we may drastically underestimate the cost
of fetching slightly more tuples than are available from the first
child. Since we've had bad experiences with over-optimistic choices
of "fast start" plans for ORDER BY LIMIT queries, that seems scary.
As a klugy workaround, set the startup cost estimate for an ordered
Append to be the sum of its children's startup costs (as MergeAppend
would). This doesn't really describe reality, but it's less likely
to cause a bad plan choice than an underestimated startup cost would.
In practice, the cases where we really care about this optimization
will have child plans that are IndexScans with zero startup cost,
so that the overly conservative estimate is still just zero.
David Rowley, reviewed by Julien Rouhaud and Antonin Houska
Discussion: https://postgr.es/m/CAKJS1f-hAqhPLRk_RaSFTgYxd=Tz5hA7kQ2h4-DhJufQk8TGuw@mail.gmail.com
Previously, the planner created RangeTblEntry and RelOptInfo structs
for every partition of a partitioned table, even though many of them
might later be deemed uninteresting thanks to partition pruning logic.
This incurred significant overhead when there are many partitions.
Arrange to postpone creation of these data structures until after
we've processed the query enough to identify restriction quals for
the partitioned table, and then apply partition pruning before not
after creation of each partition's data structures. In this way
we need not open the partition relations at all for partitions that
the planner has no real interest in.
For queries that can be proven at plan time to access only a small
number of partitions, this patch improves the practical maximum
number of partitions from under 100 to perhaps a few thousand.
Amit Langote, reviewed at various times by Dilip Kumar, Jesper Pedersen,
Yoshikazu Imai, and David Rowley
Discussion: https://postgr.es/m/9d7c5112-cb99-6a47-d3be-cf1ee6862a1d@lab.ntt.co.jp
These instances were apparently trying to carry the const qualifier
from the arguments through the complex casts, but for that the const
qualifier was misplaced.
Postpone most of the effort of constructing PartitionedRelPruneInfos
until after we have found out whether run-time pruning is needed at all.
This costs very little duplicated effort (basically just an extra
find_base_rel() call per partition) and saves quite a bit when we
can't do run-time pruning.
Also, merge the first loop (for building relid_subpart_map) into
the second loop, since we don't need the map to be valid during
that loop.
Amit Langote
Discussion: https://postgr.es/m/9d7c5112-cb99-6a47-d3be-cf1ee6862a1d@lab.ntt.co.jp
This adds a flag "deterministic" to collations. If that is false,
such a collation disables various optimizations that assume that
strings are equal only if they are byte-wise equal. That then allows
use cases such as case-insensitive or accent-insensitive comparisons
or handling of strings with different Unicode normal forms.
This functionality is only supported with the ICU provider. At least
glibc doesn't appear to have any locales that work in a
nondeterministic way, so it's not worth supporting this for the libc
provider.
The term "deterministic comparison" in this context is from Unicode
Technical Standard #10
(https://unicode.org/reports/tr10/#Deterministic_Comparison).
This patch makes changes in three areas:
- CREATE COLLATION DDL changes and system catalog changes to support
this new flag.
- Many executor nodes and auxiliary code are extended to track
collations. Previously, this code would just throw away collation
information, because the eventually-called user-defined functions
didn't use it since they only cared about equality, which didn't
need collation information.
- String data type functions that do equality comparisons and hashing
are changed to take the (non-)deterministic flag into account. For
comparison, this just means skipping various shortcuts and tie
breakers that use byte-wise comparison. For hashing, we first need
to convert the input string to a canonical "sort key" using the ICU
analogue of strxfrm().
Reviewed-by: Daniel Verite <daniel@manitou-mail.org>
Reviewed-by: Peter Geoghegan <pg@bowt.ie>
Discussion: https://www.postgresql.org/message-id/flat/1ccc668f-4cbc-0bef-af67-450b47cdfee7@2ndquadrant.com
Some buildfarm members using CLOBBER_CACHE_ALWAYS have been having OOM
problems of late. Commit 2455ab488 addressed this problem by recovering
space transiently used within RelationBuildPartitionDesc, but it turns
out that leaves quite a lot on the table, because other subroutines of
RelationBuildDesc also leak memory like mad. Let's move the temp-context
management into RelationBuildDesc so that leakage from the other
subroutines is also recovered.
I examined this issue by arranging for postgres.c to dump the size of
MessageContext just before resetting it in each command cycle, and
then running the update.sql regression test (which is one of the two
that are seeing buildfarm OOMs) with and without CLOBBER_CACHE_ALWAYS.
Before 2455ab488, the peak space usage with CCA was as much as 250MB.
That patch got it down to ~80MB, but with this patch it's about 0.5MB,
and indeed the space usage now seems nearly indistinguishable from a
non-CCA build.
RelationBuildDesc's traditional behavior of not worrying about leaking
transient data is of many years' standing, so I'm pretty hesitant to
change that without more evidence that it'd be useful in a normal build.
(So far as I can see, non-CCA memory consumption is about the same with
or without this change, whuch if anything suggests that it isn't useful.)
Hence, configure the patch so that we recover space only when
CLOBBER_CACHE_ALWAYS or CLOBBER_CACHE_RECURSIVELY is defined. However,
that choice can be overridden at compile time, in case somebody would
like to do some performance testing and try to develop evidence for
changing that decision.
It's possible that we ought to back-patch this change, but in the
absence of back-branch OOM problems in the buildfarm, I'm not in
a hurry to do that.
Discussion: https://postgr.es/m/CA+TgmoY3bRmGB6-DUnoVy5fJoreiBJ43rwMrQRCdPXuKt4Ykaw@mail.gmail.com
The idea was to generate all the junk in a destroyable subcontext rather
than leaking it in the caller's context, but partition_bounds_create was
still being called in the caller's context, allowing plenty of scope for
leakage. Also, get_rel_relkind() was still being called in the rel's
rd_pdcxt, creating a risk of session-lifespan memory wastage.
Simplify the logic a bit while at it. Also, reduce rd_pdcxt to
ALLOCSET_SMALL_SIZES, since it seems likely to not usually be big.
Probably something like this needs to be back-patched into v11,
but for now let's get some buildfarm testing on this.
Discussion: https://postgr.es/m/15943.1552601288@sss.pgh.pa.us
In normal builds, this isn't very important, because the leaks go
into fairly short-lived contexts, but under CLOBBER_CACHE_ALWAYS,
this can result in leaking hundreds of megabytes into MessageContext,
which probably explains recent failures on hyrax.
This may or may not be the best long-term strategy for dealing
with this leak, but we can change it later if we come up with
something better. For now, do this to make the buildfarm green
again (hopefully). Commit 898e5e3290
seems to have exacerbated this problem for reasons that are not
quite clear, but I don't believe it's actually the cause.
Discussion: http://postgr.es/m/CA+TgmoY3bRmGB6-DUnoVy5fJoreiBJ43rwMrQRCdPXuKt4Ykaw@mail.gmail.com
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
We still require AccessExclusiveLock on the partition itself, because
otherwise an insert that violates the newly-imposed partition
constraint could be in progress at the same time that we're changing
that constraint; only the lock level on the parent relation is
weakened.
To make this safe, we have to cope with (at least) three separate
problems. First, relevant DDL might commit while we're in the process
of building a PartitionDesc. If so, find_inheritance_children() might
see a new partition while the RELOID system cache still has the old
partition bound cached, and even before invalidation messages have
been queued. To fix that, if we see that the pg_class tuple seems to
be missing or to have a null relpartbound, refetch the value directly
from the table. We can't get the wrong value, because DETACH PARTITION
still requires AccessExclusiveLock throughout; if we ever want to
change that, this will need more thought. In testing, I found it quite
difficult to hit even the null-relpartbound case; the race condition
is extremely tight, but the theoretical risk is there.
Second, successive calls to RelationGetPartitionDesc might not return
the same answer. The query planner will get confused if lookup up the
PartitionDesc for a particular relation does not return a consistent
answer for the entire duration of query planning. Likewise, query
execution will get confused if the same relation seems to have a
different PartitionDesc at different times. Invent a new
PartitionDirectory concept and use it to ensure consistency. This
ensures that a single invocation of either the planner or the executor
sees the same view of the PartitionDesc from beginning to end, but it
does not guarantee that the planner and the executor see the same
view. Since this allows pointers to old PartitionDesc entries to
survive even after a relcache rebuild, also postpone removing the old
PartitionDesc entry until we're certain no one is using it.
For the most part, it seems to be OK for the planner and executor to
have different views of the PartitionDesc, because the executor will
just ignore any concurrently added partitions which were unknown at
plan time; those partitions won't be part of the inheritance
expansion, but invalidation messages will trigger replanning at some
point. Normally, this happens by the time the very next command is
executed, but if the next command acquires no locks and executes a
prepared query, it can manage not to notice until a new transaction is
started. We might want to tighten that up, but it's material for a
separate patch. There would still be a small window where a query
that started just after an ATTACH PARTITION command committed might
fail to notice its results -- but only if the command starts before
the commit has been acknowledged to the user. All in all, the warts
here around serializability seem small enough to be worth accepting
for the considerable advantage of being able to add partitions without
a full table lock.
Although in general the consequences of new partitions showing up
between planning and execution are limited to the query not noticing
the new partitions, run-time partition pruning will get confused in
that case, so that's the third problem that this patch fixes.
Run-time partition pruning assumes that indexes into the PartitionDesc
are stable between planning and execution. So, add code so that if
new partitions are added between plan time and execution time, the
indexes stored in the subplan_map[] and subpart_map[] arrays within
the plan's PartitionedRelPruneInfo get adjusted accordingly. There
does not seem to be a simple way to generalize this scheme to cope
with partitions that are removed, mostly because they could then get
added back again with different bounds, but it works OK for added
partitions.
This code does not try to ensure that every backend participating in
a parallel query sees the same view of the PartitionDesc. That
currently doesn't matter, because we never pass PartitionDesc
indexes between backends. Each backend will ignore the concurrently
added partitions which it notices, and it doesn't matter if different
backends are ignoring different sets of concurrently added partitions.
If in the future that matters, for example because we allow writes in
parallel query and want all participants to do tuple routing to the same
set of partitions, the PartitionDirectory concept could be improved to
share PartitionDescs across backends. There is a draft patch to
serialize and restore PartitionDescs on the thread where this patch
was discussed, which may be a useful place to start.
Patch by me. Thanks to Alvaro Herrera, David Rowley, Simon Riggs,
Amit Langote, and Michael Paquier for discussion, and to Alvaro
Herrera for some review.
Discussion: http://postgr.es/m/CA+Tgmobt2upbSocvvDej3yzokd7AkiT+PvgFH+a9-5VV1oJNSQ@mail.gmail.com
Discussion: http://postgr.es/m/CA+TgmoZE0r9-cyA-aY6f8WFEROaDLLL7Vf81kZ8MtFCkxpeQSw@mail.gmail.com
Discussion: http://postgr.es/m/CA+TgmoY13KQZF-=HNTrt9UYWYx3_oYOQpu9ioNT49jGgiDpUEA@mail.gmail.com
This is similar in spirit to the existing partbounds.c file in the
same directory, except that there's a lot less code in the new file
created by this commit. Pending work in this area proposes to add a
bunch more code related to PartitionDescs, though, and this will give
us a good place to put it.
Discussion: http://postgr.es/m/CA+TgmoZUwPf_uanjF==gTGBMJrn8uCq52XYvAEorNkLrUdoawg@mail.gmail.com
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
Move a few very simple node-creation and node-type-testing functions
from the planner's clauses.c to nodes/makefuncs and nodes/nodeFuncs.
There's nothing planner-specific about them, as evidenced by the
number of other places that were using them.
While at it, rename and_clause() etc to is_andclause() etc, to clarify
that they are node-type-testing functions not node-creation functions.
And use "static inline" implementations for the shortest ones.
Also, modify flatten_join_alias_vars() and some subsidiary functions
to take a Query not a PlannerInfo to define the join structure that
Vars should be translated according to. They were only using the
"parse" field of the PlannerInfo anyway, so this just requires removing
one level of indirection. The advantage is that now parse_agg.c can
use flatten_join_alias_vars() without the horrid kluge of creating an
incomplete PlannerInfo, which will allow that file to be decoupled from
relation.h in a subsequent patch.
Discussion: https://postgr.es/m/11460.1548706639@sss.pgh.pa.us
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
This commit moves expand_inherited_tables and underlings from
optimizer/prep/prepunionc.c to optimizer/utils/inherit.c.
Also, all of the AppendRelInfo-based expression manipulation routines
are moved to optimizer/utils/appendinfo.c.
No functional code changes. One exception is the introduction of
make_append_rel_info, but that's still just moving around code.
Also, stop including <limits.h> in prepunion.c, which no longer needs
it since 3fc6e2d7f5. I (Álvaro) noticed this because Amit was copying
that to inherit.c, which likewise doesn't need it.
Author: Amit Langote
Discussion: https://postgr.es/m/3be67028-a00a-502c-199a-da00eec8fb6e@lab.ntt.co.jp
Bruce Momjian
Michael Paquier
Tom Lane
Etsuro Fujita
Amit Kapila
Andres Freund
Alvaro Herrera
David Rowley
Peter Eisentraut
Robert Haas