Add a GUC that acts as a multiplier on work_mem. It gets applied when
sizing executor node hash tables that were previously size constrained
using work_mem alone.
The new GUC can be used to preferentially give hash-based nodes more
memory than the generic work_mem limit. It is intended to enable admin
tuning of the executor's memory usage. Overall system throughput and
system responsiveness can be improved by giving hash-based executor
nodes more memory (especially over sort-based alternatives, which are
often much less sensitive to being memory constrained).
The default value for hash_mem_multiplier is 1.0, which is also the
minimum valid value. This means that hash-based nodes continue to apply
work_mem in the traditional way by default.
hash_mem_multiplier is generally useful. However, it is being added now
due to concerns about hash aggregate performance stability for users
that upgrade to Postgres 13 (which added disk-based hash aggregation in
commit 1f39bce0). While the old hash aggregate behavior risked
out-of-memory errors, it is nevertheless likely that many users actually
benefited. Hash agg's previous indifference to work_mem during query
execution was not just faster; it also accidentally made aggregation
resilient to grouping estimate problems (at least in cases where this
didn't create destabilizing memory pressure).
hash_mem_multiplier can provide a certain kind of continuity with the
behavior of Postgres 12 hash aggregates in cases where the planner
incorrectly estimates that all groups (plus related allocations) will
fit in work_mem/hash_mem. This seems necessary because hash-based
aggregation is usually much slower when only a small fraction of all
groups can fit. Even when it isn't possible to totally avoid hash
aggregates that spill, giving hash aggregation more memory will reliably
improve performance (the same cannot be said for external sort
operations, which appear to be almost unaffected by memory availability
provided it's at least possible to get a single merge pass).
The PostgreSQL 13 release notes should advise users that increasing
hash_mem_multiplier can help with performance regressions associated
with hash aggregation. That can be taken care of by a later commit.
Author: Peter Geoghegan
Reviewed-By: Álvaro Herrera, Jeff Davis
Discussion: https://postgr.es/m/20200625203629.7m6yvut7eqblgmfo@alap3.anarazel.de
Discussion: https://postgr.es/m/CAH2-WzmD%2Bi1pG6rc1%2BCjc4V6EaFJ_qSuKCCHVnH%3DoruqD-zqow%40mail.gmail.com
Backpatch: 13-, where disk-based hash aggregation was introduced.
Note: This GUC was originally named enable_hashagg_disk when it appeared
in commit 1f39bce0, which added disk-based hash aggregation. It was
subsequently renamed in commit 92c58fd9.
Author: Peter Geoghegan
Reviewed-By: Jeff Davis, Álvaro Herrera
Discussion: https://postgr.es/m/9d9d1e1252a52ea1bad84ea40dbebfd54e672a0f.camel%40j-davis.com
Backpatch: 13-, where disk-based hash aggregation was introduced.
Eliminate enable_groupingsets_hash_disk, which was primarily useful
for testing grouping sets that use HashAgg and spill. Instead, hack
the table stats to convince the planner to choose hashed aggregation
for grouping sets that will spill to disk. Suggested by Melanie
Plageman.
Rename enable_hashagg_disk to hashagg_avoid_disk_plan, and invert the
meaning of on/off. The new name indicates more strongly that it only
affects the planner. Also, the word "avoid" is less definite, which
should avoid surprises when HashAgg still needs to use the
disk. Change suggested by Justin Pryzby, though I chose a different
GUC name.
Discussion: https://postgr.es/m/CAAKRu_aisiENMsPM2gC4oUY1hHG3yrCwY-fXUg22C6_MJUwQdA%40mail.gmail.com
Discussion: https://postgr.es/m/20200610021544.GA14879@telsasoft.com
Backpatch-through: 13
Thomas Munro fixed a longstanding annoyance in pg_bsd_indent, that
it would misformat lines containing IsA() macros on the assumption
that the IsA() call should be treated like a cast. This improves
some other cases involving field/variable names that match typedefs,
too. The only places that get worse are a couple of uses of the
OpenSSL macro STACK_OF(); we'll gladly take that trade-off.
Discussion: https://postgr.es/m/20200114221814.GA19630@alvherre.pgsql
Includes some manual cleanup of places that pgindent messed up,
most of which weren't per project style anyway.
Notably, it seems some people didn't absorb the style rules of
commit c9d297751, because there were a bunch of new occurrences
of function calls with a newline just after the left paren, all
with faulty expectations about how the rest of the call would get
indented.
When estimating the number of pre-sorted groups in cost_incremental_sort
we must not pass Vars with varno 0 to estimate_num_groups, which would
cause failues in find_base_rel. This may happen when sorting output of
set operations, thanks to generate_append_tlist.
Unlike recurse_set_operations we can't easily access the original target
list, so if we find any Vars with varno 0, we fall back to the default
estimate DEFAULT_NUM_DISTINCT.
Reported-by: Justin Pryzby
Discussion: https://postgr.es/m/20200411214639.GK2228%40telsasoft.com
Incremental Sort is an optimized variant of multikey sort for cases when
the input is already sorted by a prefix of the requested sort keys. For
example when the relation is already sorted by (key1, key2) and we need
to sort it by (key1, key2, key3) we can simply split the input rows into
groups having equal values in (key1, key2), and only sort/compare the
remaining column key3.
This has a number of benefits:
- Reduced memory consumption, because only a single group (determined by
values in the sorted prefix) needs to be kept in memory. This may also
eliminate the need to spill to disk.
- Lower startup cost, because Incremental Sort produce results after each
prefix group, which is beneficial for plans where startup cost matters
(like for example queries with LIMIT clause).
We consider both Sort and Incremental Sort, and decide based on costing.
The implemented algorithm operates in two different modes:
- Fetching a minimum number of tuples without check of equality on the
prefix keys, and sorting on all columns when safe.
- Fetching all tuples for a single prefix group and then sorting by
comparing only the remaining (non-prefix) keys.
We always start in the first mode, and employ a heuristic to switch into
the second mode if we believe it's beneficial - the goal is to minimize
the number of unnecessary comparions while keeping memory consumption
below work_mem.
This is a very old patch series. The idea was originally proposed by
Alexander Korotkov back in 2013, and then revived in 2017. In 2018 the
patch was taken over by James Coleman, who wrote and rewrote most of the
current code.
There were many reviewers/contributors since 2013 - I've done my best to
pick the most active ones, and listed them in this commit message.
Author: James Coleman, Alexander Korotkov
Reviewed-by: Tomas Vondra, Andreas Karlsson, Marti Raudsepp, Peter Geoghegan, Robert Haas, Thomas Munro, Antonin Houska, Andres Freund, Alexander Kuzmenkov
Discussion: https://postgr.es/m/CAPpHfdscOX5an71nHd8WSUH6GNOCf=V7wgDaTXdDd9=goN-gfA@mail.gmail.com
Discussion: https://postgr.es/m/CAPpHfds1waRZ=NOmueYq0sx1ZSCnt+5QJvizT8ndT2=etZEeAQ@mail.gmail.com
While performing hash aggregation, track memory usage when adding new
groups to a hash table. If the memory usage exceeds work_mem, enter
"spill mode".
In spill mode, new groups are not created in the hash table(s), but
existing groups continue to be advanced if input tuples match. Tuples
that would cause a new group to be created are instead spilled to a
logical tape to be processed later.
The tuples are spilled in a partitioned fashion. When all tuples from
the outer plan are processed (either by advancing the group or
spilling the tuple), finalize and emit the groups from the hash
table. Then, create new batches of work from the spilled partitions,
and select one of the saved batches and process it (possibly spilling
recursively).
Author: Jeff Davis
Reviewed-by: Tomas Vondra, Adam Lee, Justin Pryzby, Taylor Vesely, Melanie Plageman
Discussion: https://postgr.es/m/507ac540ec7c20136364b5272acbcd4574aa76ef.camel@j-davis.com
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
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
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
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
Add support function requests for estimating the selectivity, cost,
and number of result rows (if a SRF) of the target function.
The lack of a way to estimate selectivity of a boolean-returning
function in WHERE has been a recognized deficiency of the planner
since Berkeley days. This commit finally fixes it.
In addition, non-constant estimates of cost and number of output
rows are now possible. We still fall back to looking at procost
and prorows if the support function doesn't service the request,
of course.
To make concrete use of the possibility of estimating output rowcount
for SRFs, this commit adds support functions for array_unnest(anyarray)
and the integer variants of generate_series; the lack of plausible
rowcount estimates for those, even when it's obvious to a human,
has been a repeated subject of complaints. Obviously, much more
could now be done in this line, but I'm mostly just trying to get
the infrastructure in place.
Discussion: https://postgr.es/m/15193.1548028093@sss.pgh.pa.us
In place of three separate but interrelated lists (indexclauses,
indexquals, and indexqualcols), an IndexPath now has one list
"indexclauses" of IndexClause nodes. This holds basically the same
information as before, but in a more useful format: in particular, there
is now a clear connection between an indexclause (an original restriction
clause from WHERE or JOIN/ON) and the indexquals (directly usable index
conditions) derived from it.
We also change the ground rules a bit by mandating that clause commutation,
if needed, be done up-front so that what is stored in the indexquals list
is always directly usable as an index condition. This gets rid of repeated
re-determination of which side of the clause is the indexkey during costing
and plan generation, as well as repeated lookups of the commutator
operator. To minimize the added up-front cost, the typical case of
commuting a plain OpExpr is handled by a new special-purpose function
commute_restrictinfo(). For RowCompareExprs, generating the new clause
properly commuted to begin with is not really any more complex than before,
it's just different --- and we can save doing that work twice, as the
pretty-klugy original implementation did.
Tracking the connection between original and derived clauses lets us
also track explicitly whether the derived clauses are an exact or lossy
translation of the original. This provides a cheap solution to getting
rid of unnecessary rechecks of boolean index clauses, which previously
seemed like it'd be more expensive than it was worth.
Another pleasant (IMO) side-effect is that EXPLAIN now always shows
index clauses with the indexkey on the left; this seems less confusing.
This commit leaves expand_indexqual_conditions() and some related
functions in a slightly messy state. I didn't bother to change them
any more than minimally necessary to work with the new data structure,
because all that code is going to be refactored out of existence in
a follow-on patch.
Discussion: https://postgr.es/m/22182.1549124950@sss.pgh.pa.us
The old name of this file was never a very good indication of what it
was for. Now that there's also access/relation.h, we have a potential
confusion hazard as well, so let's rename it to something more apropos.
Per discussion, "pathnodes.h" is reasonable, since a good fraction of
the file is Path node definitions.
While at it, tweak a couple of other headers that were gratuitously
importing relation.h into modules that don't need it.
Discussion: https://postgr.es/m/7719.1548688728@sss.pgh.pa.us
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
The fact that "SELECT expression" has no base relations has long been a
thorn in the side of the planner. It makes it hard to flatten a sub-query
that looks like that, or is a trivial VALUES() item, because the planner
generally uses relid sets to identify sub-relations, and such a sub-query
would have an empty relid set if we flattened it. prepjointree.c contains
some baroque logic that works around this in certain special cases --- but
there is a much better answer. We can replace an empty FROM clause with a
dummy RTE that acts like a table of one row and no columns, and then there
are no such corner cases to worry about. Instead we need some logic to
get rid of useless dummy RTEs, but that's simpler and covers more cases
than what was there before.
For really trivial cases, where the query is just "SELECT expression" and
nothing else, there's a hazard that adding the extra RTE makes for a
noticeable slowdown; even though it's not much processing, there's not
that much for the planner to do overall. However testing says that the
penalty is very small, close to the noise level. In more complex queries,
this is able to find optimizations that we could not find before.
The new RTE type is called RTE_RESULT, since the "scan" plan type it
gives rise to is a Result node (the same plan we produced for a "SELECT
expression" query before). To avoid confusion, rename the old ResultPath
path type to GroupResultPath, reflecting that it's only used in degenerate
grouping cases where we know the query produces just one grouped row.
(It wouldn't work to unify the two cases, because there are different
rules about where the associated quals live during query_planner.)
Note: although this touches readfuncs.c, I don't think a catversion
bump is required, because the added case can't occur in stored rules,
only plans.
Patch by me, reviewed by David Rowley and Mark Dilger
Discussion: https://postgr.es/m/15944.1521127664@sss.pgh.pa.us
Up to now we've not worried much about joins where the join key is a
relation's CTID column, reasoning that storing a table's CTIDs in some
other table would be pretty useless. However, there are use-cases for
this sort of query involving self-joins, so that argument doesn't really
hold water.
This patch allows generating plans for joins on CTID that use a nestloop
with inner TidScan, similar to what we might do with an index on the join
column. This is the most efficient way to join when the outer side of
the nestloop is expected to yield relatively few rows.
This change requires upgrading tidpath.c and the generated TidPaths
to work with RestrictInfos instead of bare qual clauses, but that's
long-postponed technical debt anyway.
Discussion: https://postgr.es/m/17443.1545435266@sss.pgh.pa.us
"rescanratio" was computed as 1 + rescanned-tuples / total-inner-tuples,
which is sensible if it's to be multiplied by total-inner-tuples or a cost
value corresponding to scanning all the inner tuples. But in reality it
was (mostly) multiplied by inner_rows or a related cost, numbers that take
into account the possibility of stopping short of scanning the whole inner
relation thanks to a limited key range in the outer relation. This'd
still make sense if we could expect that stopping short would result in a
proportional decrease in the number of tuples that have to be rescanned.
It does not, however. The argument that establishes the validity of our
estimate for that number is independent of whether we scan all of the inner
relation or stop short, and experimentation also shows that stopping short
doesn't reduce the number of rescanned tuples. So the correct calculation
is 1 + rescanned-tuples / inner_rows, and we should be sure to multiply
that by inner_rows or a corresponding cost value.
Most of the time this doesn't make much difference, but if we have
both a high rescan rate (due to lots of duplicate values) and an outer
key range much smaller than the inner key range, then the error can
be significant, leading to a large underestimate of the cost associated
with rescanning.
Per report from Vijaykumar Jain. This thinko appears to go all the way
back to the introduction of the rescan estimation logic in commit
70fba7043, so back-patch to all supported branches.
Discussion: https://postgr.es/m/CAE7uO5hMb_TZYJcZmLAgO6iD68AkEK6qCe7i=vZUkCpoKns+EQ@mail.gmail.com
In final_cost_hashjoin(), commit 9c7f5229a allowed inner_unique cases
to follow a code path previously used only for SEMI/ANTI joins; but it
neglected to fix an if-test within that path that assumed SEMI and ANTI
were the only possible cases. This resulted in a wrong value for
hashjointuples, and an ensuing bad cost estimate, for inner_unique normal
joins. Fortunately, for inner_unique normal joins we can assume the number
of joined tuples is the same as for a SEMI join; so there's no need for
more code, we just have to invert the test to check for ANTI not SEMI.
It turns out that in two contrib tests in which commit 9c7f5229a
changed the plan expected for a query, the change was actually wrong
and induced by this estimation error, not by any real improvement.
Hence this patch also reverts those changes.
Per report from RK Korlapati. Backpatch to v10 where the error was
introduced.
David Rowley
Discussion: https://postgr.es/m/CA+SNy03bhq0fodsfOkeWDCreNjJVjsdHwUsb7AG=jpe0PtZc_g@mail.gmail.com
We include <float.h> in every place that needs isnan(), because MSVC
used to require it. However, since MSVC 2013 that's no longer necessary
(cf. commit cec8394b5c), so we can retire the inclusion to a
version-specific stanza in win32_port.h, where it doesn't need to
pollute random .c files. The header is of course still needed in a few
places for other reasons.
I (Álvaro) removed float.h from a few more files than in Emre's original
patch. This doesn't break the build in my system, but we'll see what
the buildfarm has to say about it all.
Author: Emre Hasegeli
Discussion: https://postgr.es/m/CAE2gYzyc0+5uG+Cd9-BSL7NKC8LSHLNg1Aq2=8ubjnUwut4_iw@mail.gmail.com
This controls both plan-time and execution-time new-style partition
pruning. While finer-grain control is possible (maybe using an enum GUC
instead of boolean), there doesn't seem to be much need for that.
This new parameter controls partition pruning for all queries:
trivially, SELECT queries that affect partitioned tables are naturally
under its control since they are using the new technology. However,
while UPDATE/DELETE queries do not use the new code, we make the new GUC
control their behavior also (stealing control from
constraint_exclusion), because it is more natural, and it leads to a
more natural transition to the future in which those queries will also
use the new pruning code.
Constraint exclusion still controls pruning for regular inheritance
situations (those not involving partitioned tables).
Author: David Rowley
Review: Amit Langote, Ashutosh Bapat, Justin Pryzby, David G. Johnston
Discussion: https://postgr.es/m/CAKJS1f_0HwsxJG9m+nzU+CizxSdGtfe6iF_ykPYBiYft302DCw@mail.gmail.com
On further reflection, commit e5d83995e didn't go far enough: pretty much
everywhere in the planner that examines a clause's is_pushed_down flag
ought to be changed to use the more complicated behavior where we also
check the clause's required_relids. Otherwise we could make incorrect
decisions about whether, say, a clause is safe to use as a hash clause.
Some (many?) of these places are safe as-is, either because they are
never reached while considering a parameterized path, or because there
are additional checks that would reject a pushed-down clause anyway.
However, it seems smarter to just code them all the same way rather
than rely on easily-broken reasoning of that sort.
In support of that, invent a new macro RINFO_IS_PUSHED_DOWN that should
be used in place of direct tests on the is_pushed_down flag.
Like the previous patch, back-patch to all supported branches.
Discussion: https://postgr.es/m/f8128b11-c5bf-3539-48cd-234178b2314d@proxel.se
If the partition keys of input relation are part of the GROUP BY
clause, all the rows belonging to a given group come from a single
partition. This allows aggregation/grouping over a partitioned
relation to be broken down * into aggregation/grouping on each
partition. This should be no worse, and often better, than the normal
approach.
If the GROUP BY clause does not contain all the partition keys, we can
still perform partial aggregation for each partition and then finalize
aggregation after appending the partial results. This is less certain
to be a win, but it's still useful.
Jeevan Chalke, Ashutosh Bapat, Robert Haas. The larger patch series
of which this patch is a part was also reviewed and tested by Antonin
Houska, Rajkumar Raghuwanshi, David Rowley, Dilip Kumar, Konstantin
Knizhnik, Pascal Legrand, and Rafia Sabih.
Discussion: http://postgr.es/m/CAM2+6=V64_xhstVHie0Rz=KPEQnLJMZt_e314P0jaT_oJ9MR8A@mail.gmail.com
Commit 1804284042 established that single-batch
parallel-aware hash joins could create one large shared hash table using the
combined work_mem budget of all participants. The costing accidentally
assumed that parallel-oblivious hash joins could also do that. The
documentation for initial_cost_hashjoin() also failed to mention the new
argument. Repair.
Author: Thomas Munro
Reported-By: Antonin Houska
Reviewed-By: Antonin Houska
Discussion: https://postgr.es/m/12441.1513935950%40localhost
Previously, Append didn't charge anything at all, and MergeAppend
charged only cpu_operator_cost, about half the value used here. This
change might make MergeAppend plans slightly more likely to be chosen
than before, since this commit increases the assumed cost for Append
-- with default values -- by 0.005 per tuple but MergeAppend by only
0.0025 per tuple. Since the comparisons required by MergeAppend are
costed separately, it's not clear why MergeAppend needs to be
otherwise more expensive than Append, so hopefully this is OK.
Prior to partition-wise join, it didn't really matter whether or not
an Append node had any cost of its own, because every plan had to use
the same number of Append or MergeAppend nodes and in the same places.
Only the relative cost of Append vs. MergeAppend made a difference.
Now, however, it is possible to avoid some of the Append nodes using a
partition-wise join, so it's worth making an effort. Pending patches
for partition-wise aggregate care too, because an Append of Aggregate
nodes will incur the Append overhead fewer times than an Aggregate
over an Append. Although in most cases this change will favor the use
of partition-wise techniques, it does the opposite when the join
cardinality is greater than the sum of the input cardinalities. Since
this situation arises in an existing regression test, I [rhaas]
adjusted it to keep the overall plan shape approximately the same.
Jeevan Chalke, per a suggestion from David Rowley. Reviewed by
Ashutosh Bapat. Some changes by me. The larger patch series of which
this patch is a part was also reviewed and tested by Antonin Houska,
Rajkumar Raghuwanshi, David Rowley, Dilip Kumar, Konstantin Knizhnik,
Pascal Legrand, Rafia Sabih, and me.
Discussion: http://postgr.es/m/CAKJS1f9UXdk6ZYyqbJnjFO9a9hyHKGW7B=ZRh-rxy9qxfPA5Gw@mail.gmail.com
To make this work, tuplesort.c and logtape.c must also support
parallelism, so this patch adds that infrastructure and then applies
it to the particular case of parallel btree index builds. Testing
to date shows that this can often be 2-3x faster than a serial
index build.
The model for deciding how many workers to use is fairly primitive
at present, but it's better than not having the feature. We can
refine it as we get more experience.
Peter Geoghegan with some help from Rushabh Lathia. While Heikki
Linnakangas is not an author of this patch, he wrote other patches
without which this feature would not have been possible, and
therefore the release notes should possibly credit him as an author
of this feature. Reviewed by Claudio Freire, Heikki Linnakangas,
Thomas Munro, Tels, Amit Kapila, me.
Discussion: http://postgr.es/m/CAM3SWZQKM=Pzc=CAHzRixKjp2eO5Q0Jg1SoFQqeXFQ647JiwqQ@mail.gmail.com
Discussion: http://postgr.es/m/CAH2-Wz=AxWqDoVvGU7dq856S4r6sJAj6DBn7VMtigkB33N5eyg@mail.gmail.com
- Remove unnecessary #include mistakenly added in execnodes.h.
- Fix mistake in comment in choose_next_subplan_for_leader.
- Adjust row estimates in cost_append for a possibly-different
parallel divisor.
- Clamp row estimates in cost_append after operations that may
not produce integers.
Amit Kapila, with cosmetic adjustments by me.
Discussion: http://postgr.es/m/CAA4eK1+qcbeai3coPpRW=GFCzFeLUsuY4T-AKHqMjxpEGZBPQg@mail.gmail.com
Introduce parallel-aware hash joins that appear in EXPLAIN plans as Parallel
Hash Join with Parallel Hash. While hash joins could already appear in
parallel queries, they were previously always parallel-oblivious and had a
partial subplan only on the outer side, meaning that the work of the inner
subplan was duplicated in every worker.
After this commit, the planner will consider using a partial subplan on the
inner side too, using the Parallel Hash node to divide the work over the
available CPU cores and combine its results in shared memory. If the join
needs to be split into multiple batches in order to respect work_mem, then
workers process different batches as much as possible and then work together
on the remaining batches.
The advantages of a parallel-aware hash join over a parallel-oblivious hash
join used in a parallel query are that it:
* avoids wasting memory on duplicated hash tables
* avoids wasting disk space on duplicated batch files
* divides the work of building the hash table over the CPUs
One disadvantage is that there is some communication between the participating
CPUs which might outweigh the benefits of parallelism in the case of small
hash tables. This is avoided by the planner's existing reluctance to supply
partial plans for small scans, but it may be necessary to estimate
synchronization costs in future if that situation changes. Another is that
outer batch 0 must be written to disk if multiple batches are required.
A potential future advantage of parallel-aware hash joins is that right and
full outer joins could be supported, since there is a single set of matched
bits for each hashtable, but that is not yet implemented.
A new GUC enable_parallel_hash is defined to control the feature, defaulting
to on.
Author: Thomas Munro
Reviewed-By: Andres Freund, Robert Haas
Tested-By: Rafia Sabih, Prabhat Sahu
Discussion:
https://postgr.es/m/CAEepm=2W=cOkiZxcg6qiFQP-dHUe09aqTrEMM7yJDrHMhDv_RA@mail.gmail.comhttps://postgr.es/m/CAEepm=37HKyJ4U6XOLi=JgfSHM3o6B-GaeO-6hkOmneTDkH+Uw@mail.gmail.com
When we create an Append node, we can spread out the workers over the
subplans instead of piling on to each subplan one at a time, which
should typically be a bit more efficient, both because the startup
cost of any plan executed entirely by one worker is paid only once and
also because of reduced contention. We can also construct Append
plans using a mix of partial and non-partial subplans, which may allow
for parallelism in places that otherwise couldn't support it.
Unfortunately, this patch doesn't handle the important case of
parallelizing UNION ALL by running each branch in a separate worker;
the executor infrastructure is added here, but more planner work is
needed.
Amit Khandekar, Robert Haas, Amul Sul, reviewed and tested by
Ashutosh Bapat, Amit Langote, Rafia Sabih, Amit Kapila, and
Rajkumar Raghuwanshi.
Discussion: http://postgr.es/m/CAJ3gD9dy0K_E8r727heqXoBmWZ83HwLFwdcaSSmBQ1+S+vRuUQ@mail.gmail.com
For some reason, we have never accounted for either the evaluation cost
or the selectivity of filter conditions attached to Agg and Group nodes
(which, in practice, are always conditions from a HAVING clause).
Applying our regular selectivity logic to post-grouping conditions is a
bit bogus, but it's surely better than taking the selectivity as 1.0.
Perhaps someday the extended-statistics mechanism can be taught to provide
statistics that would help us in getting non-default estimates here.
Per a gripe from Benjamin Coutu. This is surely a bug fix, but I'm
hesitant to back-patch because of the prospect of destabilizing existing
plan choices. Given that it took us this long to notice the bug, it's
probably not hurting too many people in the field.
Discussion: https://postgr.es/m/20968.1509486337@sss.pgh.pa.us
Instead of joining two partitioned tables in their entirety we can, if
it is an equi-join on the partition keys, join the matching partitions
individually. This involves teaching the planner about "other join"
rels, which are related to regular join rels in the same way that
other member rels are related to baserels. This can use significantly
more CPU time and memory than regular join planning, because there may
now be a set of "other" rels not only for every base relation but also
for every join relation. In most practical cases, this probably
shouldn't be a problem, because (1) it's probably unusual to join many
tables each with many partitions using the partition keys for all
joins and (2) if you do that scenario then you probably have a big
enough machine to handle the increased memory cost of planning and (3)
the resulting plan is highly likely to be better, so what you spend in
planning you'll make up on the execution side. All the same, for now,
turn this feature off by default.
Currently, we can only perform joins between two tables whose
partitioning schemes are absolutely identical. It would be nice to
cope with other scenarios, such as extra partitions on one side or the
other with no match on the other side, but that will have to wait for
a future patch.
Ashutosh Bapat, reviewed and tested by Rajkumar Raghuwanshi, Amit
Langote, Rafia Sabih, Thomas Munro, Dilip Kumar, Antonin Houska, Amit
Khandekar, and by me. A few final adjustments by me.
Discussion: http://postgr.es/m/CAFjFpRfQ8GrQvzp3jA2wnLqrHmaXna-urjm_UY9BqXj=EaDTSA@mail.gmail.com
Discussion: http://postgr.es/m/CAFjFpRcitjfrULr5jfuKWRPsGUX0LQ0k8-yG0Qw2+1LBGNpMdw@mail.gmail.com
Allowing arrays with a domain type as their element type was left un-done
in the original domain patch, but not for any very good reason. This
omission leads to such surprising results as array_agg() not working on
a domain column, because the parser can't identify a suitable output type
for the polymorphic aggregate.
In order to fix this, first clean up the APIs of coerce_to_domain() and
some internal functions in parse_coerce.c so that we consistently pass
around a CoercionContext along with CoercionForm. Previously, we sometimes
passed an "isExplicit" boolean flag instead, which is strictly less
information; and coerce_to_domain() didn't even get that, but instead had
to reverse-engineer isExplicit from CoercionForm. That's contrary to the
documentation in primnodes.h that says that CoercionForm only affects
display and not semantics. I don't think this change fixes any live bugs,
but it makes things more consistent. The main reason for doing it though
is that now build_coercion_expression() receives ccontext, which it needs
in order to be able to recursively invoke coerce_to_target_type().
Next, reimplement ArrayCoerceExpr so that the node does not directly know
any details of what has to be done to the individual array elements while
performing the array coercion. Instead, the per-element processing is
represented by a sub-expression whose input is a source array element and
whose output is a target array element. This simplifies life in
parse_coerce.c, because it can build that sub-expression by a recursive
invocation of coerce_to_target_type(). The executor now handles the
per-element processing as a compiled expression instead of hard-wired code.
The main advantage of this is that we can use a single ArrayCoerceExpr to
handle as many as three successive steps per element: base type conversion,
typmod coercion, and domain constraint checking. The old code used two
stacked ArrayCoerceExprs to handle type + typmod coercion, which was pretty
inefficient, and adding yet another array deconstruction to do domain
constraint checking seemed very unappetizing.
In the case where we just need a single, very simple coercion function,
doing this straightforwardly leads to a noticeable increase in the
per-array-element runtime cost. Hence, add an additional shortcut evalfunc
in execExprInterp.c that skips unnecessary overhead for that specific form
of expression. The runtime speed of simple cases is within 1% or so of
where it was before, while cases that previously required two levels of
array processing are significantly faster.
Finally, create an implicit array type for every domain type, as we do for
base types, enums, etc. Everything except the array-coercion case seems
to just work without further effort.
Tom Lane, reviewed by Andrew Dunstan
Discussion: https://postgr.es/m/9852.1499791473@sss.pgh.pa.us
These variables are only ever written to in assertion-enabled builds,
and the latest Microsoft compilers complain about such variables in
non-assertion-enabled builds.
Apparently they don't worry so much about variables that are written to
but not read from, so most of our PG_USED_FOR_ASSERTS_ONLY variables
don't cause the problem.
Discussion: https://postgr.es/m/7800.1505950322@sss.pgh.pa.us
The executor is capable of splitting buckets during a hash join if
too much memory is being used by a small number of buckets. However,
this only helps if a bucket's population is actually divisible; if
all the hash keys are alike, the tuples still end up in the same
new bucket. This can result in an OOM failure if there are enough
inner keys with identical hash values. The planner's cost estimates
will bias it against choosing a hash join in such situations, but not
by so much that it will never do so. To mitigate the OOM hazard,
explicitly estimate the hash bucket space needed by just the inner
side's most common value, and if that would exceed work_mem then
add disable_cost to the hash cost estimate.
This approach doesn't account for the possibility that two or more
common values would share the same hash value. On the other hand,
work_mem is normally a fairly conservative bound, so that eating
two or more times that much space is probably not going to kill us.
If we have no stats about the inner side, ignore this consideration.
There was some discussion of making a conservative assumption, but that
would effectively result in disabling hash join whenever we lack stats,
which seems like an overreaction given how seldom the problem manifests
in the field.
Per a complaint from David Hinkle. Although this could be viewed
as a bug fix, the lack of similar complaints weighs against back-
patching; indeed we waited for v11 because it seemed already rather
late in the v10 cycle to be making plan choice changes like this one.
Discussion: https://postgr.es/m/32013.1487271761@sss.pgh.pa.us
Add missing infrastructure for this node type, notably in ruleutils.c where
its lack could demonstrably cause EXPLAIN to fail. Add outfuncs/readfuncs
support. (outfuncs support is useful today for debugging purposes. The
readfuncs support may never be needed, since at present it would only
matter for parallel query and NextValueExpr should never appear in a
parallelizable query; but it seems like a bad idea to have a primnode type
that isn't fully supported here.) Teach planner infrastructure that
NextValueExpr is a volatile, parallel-unsafe, non-leaky expression node
with cost cpu_operator_cost. Given its limited scope of usage, there
*might* be no live bug today from the lack of that knowledge, but it's
certainly going to bite us on the rear someday. Teach pg_stat_statements
about the new node type, too.
While at it, also teach cost_qual_eval() that MinMaxExpr, SQLValueFunction,
XmlExpr, and CoerceToDomain should be charged as cpu_operator_cost.
Failing to do this for SQLValueFunction was an oversight in my commit
0bb51aa96. The others are longer-standing oversights, but no time like the
present to fix them. (In principle, CoerceToDomain could have cost much
higher than this, but it doesn't presently seem worth trying to examine the
domain's constraints here.)
Modify execExprInterp.c to execute NextValueExpr as an out-of-line
function; it seems quite unlikely to me that it's worth insisting that
it be inlined in all expression eval methods. Besides, providing the
out-of-line function doesn't stop anyone from inlining if they want to.
Adjust some places where NextValueExpr support had been inserted with the
aid of a dartboard rather than keeping it in the same order as elsewhere.
Discussion: https://postgr.es/m/23862.1499981661@sss.pgh.pa.us
Don't move parenthesized lines to the left, even if that means they
flow past the right margin.
By default, BSD indent lines up statement continuation lines that are
within parentheses so that they start just to the right of the preceding
left parenthesis. However, traditionally, if that resulted in the
continuation line extending to the right of the desired right margin,
then indent would push it left just far enough to not overrun the margin,
if it could do so without making the continuation line start to the left of
the current statement indent. That makes for a weird mix of indentations
unless one has been completely rigid about never violating the 80-column
limit.
This behavior has been pretty universally panned by Postgres developers.
Hence, disable it with indent's new -lpl switch, so that parenthesized
lines are always lined up with the preceding left paren.
This patch is much less interesting than the first round of indent
changes, but also bulkier, so I thought it best to separate the effects.
Discussion: https://postgr.es/m/E1dAmxK-0006EE-1r@gemulon.postgresql.org
Discussion: https://postgr.es/m/30527.1495162840@sss.pgh.pa.us
Change pg_bsd_indent to follow upstream rules for placement of comments
to the right of code, and remove pgindent hack that caused comments
following #endif to not obey the general rule.
Commit e3860ffa4d wasn't actually using
the published version of pg_bsd_indent, but a hacked-up version that
tried to minimize the amount of movement of comments to the right of
code. The situation of interest is where such a comment has to be
moved to the right of its default placement at column 33 because there's
code there. BSD indent has always moved right in units of tab stops
in such cases --- but in the previous incarnation, indent was working
in 8-space tab stops, while now it knows we use 4-space tabs. So the
net result is that in about half the cases, such comments are placed
one tab stop left of before. This is better all around: it leaves
more room on the line for comment text, and it means that in such
cases the comment uniformly starts at the next 4-space tab stop after
the code, rather than sometimes one and sometimes two tabs after.
Also, ensure that comments following #endif are indented the same
as comments following other preprocessor commands such as #else.
That inconsistency turns out to have been self-inflicted damage
from a poorly-thought-through post-indent "fixup" in pgindent.
This patch is much less interesting than the first round of indent
changes, but also bulkier, so I thought it best to separate the effects.
Discussion: https://postgr.es/m/E1dAmxK-0006EE-1r@gemulon.postgresql.org
Discussion: https://postgr.es/m/30527.1495162840@sss.pgh.pa.us
The new indent version includes numerous fixes thanks to Piotr Stefaniak.
The main changes visible in this commit are:
* Nicer formatting of function-pointer declarations.
* No longer unexpectedly removes spaces in expressions using casts,
sizeof, or offsetof.
* No longer wants to add a space in "struct structname *varname", as
well as some similar cases for const- or volatile-qualified pointers.
* Declarations using PG_USED_FOR_ASSERTS_ONLY are formatted more nicely.
* Fixes bug where comments following declarations were sometimes placed
with no space separating them from the code.
* Fixes some odd decisions for comments following case labels.
* Fixes some cases where comments following code were indented to less
than the expected column 33.
On the less good side, it now tends to put more whitespace around typedef
names that are not listed in typedefs.list. This might encourage us to
put more effort into typedef name collection; it's not really a bug in
indent itself.
There are more changes coming after this round, having to do with comment
indentation and alignment of lines appearing within parentheses. I wanted
to limit the size of the diffs to something that could be reviewed without
one's eyes completely glazing over, so it seemed better to split up the
changes as much as practical.
Discussion: https://postgr.es/m/E1dAmxK-0006EE-1r@gemulon.postgresql.org
Discussion: https://postgr.es/m/30527.1495162840@sss.pgh.pa.us
David Rowley found that the "use the smallest per-column selectivity"
heuristic applied in some cases by get_foreign_key_join_selectivity()
was badly off if the FK columns are independent, producing estimates
much worse than we got before that code was added in 9.6.
One case where that heuristic was used was for LEFT and FULL outer joins
with the referenced rel on the outside of the join. But we should not
really need to special-case those here. eqjoinsel() never has had such a
special case; the correction is applied by calc_joinrel_size_estimate()
instead. Let's just estimate such cases like inner joins and rely on that
later adjustment. (I think there was something of a thinko here, in that
the comments seem to be thinking about the selectivity as defined for
semi/anti joins; but that shouldn't apply to left/full joins.) Add a
regression test exercising such a case to show that this is sane in
at least some cases.
The other case where we used that heuristic was for SEMI/ANTI outer joins,
either if the referenced rel was on the outside, or if it was on the inside
but was part of a join within the RHS. In either case, the FK doesn't give
us a lot of traction towards estimating the selectivity. To ensure that
we don't have regressions from what happened before 9.6, let's punt by
ignoring the FK in such cases and applying the traditional selectivity
calculation. (We might be able to improve on that later, but for now
I just want to be sure it's not worse than 9.5.)
Report and patch by David Rowley, simplified a bit by me. Back-patch
to 9.6 where this code was added.
Discussion: https://postgr.es/m/CAKJS1f8NO8oCDcxrteohG6O72uU1saEVT9qX=R8pENr5QWerXw@mail.gmail.com
When costing a nestloop with stop-at-first-inner-match semantics, and a
non-indexscan inner path, final_cost_nestloop() wants to charge the full
scan cost of the inner rel at least once, with additional scans charged
at inner_rescan_run_cost which might be less. However the logic for
doing this effectively assumed that outer_matched_rows is at least 1.
If it's zero, which is not unlikely for a small outer rel, we ended up
charging inner_run_cost plus N times inner_rescan_run_cost, as much as
double the correct charge for an outer rel with only one row that
we're betting won't be matched. (Unless the inner rel is materialized,
in which case it has very small inner_rescan_run_cost and the cost
is not so far off what it should have been.)
The upshot of this was that the planner had a tendency to select plans
that failed to make effective use of the stop-at-first-inner-match
semantics, and that might have Materialize nodes in them even when the
predicted number of executions of the Materialize subplan was only 1.
This was not so obvious before commit 9c7f5229a, because the case only
arose in connection with semi/anti joins where there's not freedom to
reverse the join order. But with the addition of unique-inner joins,
it could result in some fairly bad planning choices, as reported by
Teodor Sigaev. Indeed, some of the test cases added by that commit
have plans that look dubious on closer inspection, and are changed
by this patch.
Fix the logic to ensure that we don't charge for too many inner scans.
I chose to adjust it so that the full-freight scan cost is associated
with an unmatched outer row if possible, not a matched one, since that
seems like a better model of what would happen at runtime.
This is a longstanding bug, but given the lesser impact in back branches,
and the lack of field complaints, I won't risk a back-patch.
Discussion: https://postgr.es/m/CAKJS1f-LzkUsFxdJ_-Luy38orQ+AdEXM5o+vANR+-pHAWPSecg@mail.gmail.com
This extends the castNode() notation introduced by commit 5bcab1114 to
provide, in one step, extraction of a list cell's pointer and coercion to
a concrete node type. For example, "lfirst_node(Foo, lc)" is the same
as "castNode(Foo, lfirst(lc))". Almost half of the uses of castNode
that have appeared so far include a list extraction call, so this is
pretty widely useful, and it saves a few more keystrokes compared to the
old way.
As with the previous patch, back-patch the addition of these macros to
pg_list.h, so that the notation will be available when back-patching.
Patch by me, after an idea of Andrew Gierth's.
Discussion: https://postgr.es/m/14197.1491841216@sss.pgh.pa.us
If there can certainly be no more than one matching inner row for a given
outer row, then the executor can move on to the next outer row as soon as
it's found one match; there's no need to continue scanning the inner
relation for this outer row. This saves useless scanning in nestloop
and hash joins. In merge joins, it offers the opportunity to skip
mark/restore processing, because we know we have not advanced past the
first possible match for the next outer row.
Of course, the devil is in the details: the proof of uniqueness must
depend only on joinquals (not otherquals), and if we want to skip
mergejoin mark/restore then it must depend only on merge clauses.
To avoid adding more planning overhead than absolutely necessary,
the present patch errs in the conservative direction: there are cases
where inner_unique or skip_mark_restore processing could be used, but
it will not do so because it's not sure that the uniqueness proof
depended only on "safe" clauses. This could be improved later.
David Rowley, reviewed and rather heavily editorialized on by me
Discussion: https://postgr.es/m/CAApHDvqF6Sw-TK98bW48TdtFJ+3a7D2mFyZ7++=D-RyPsL76gw@mail.gmail.com
Follow on patch in the multi-variate statistics patch series.
CREATE STATISTICS s1 WITH (dependencies) ON (a, b) FROM t;
ANALYZE;
will collect dependency stats on (a, b) and then use the measured
dependency in subsequent query planning.
Commit 7b504eb282 added
CREATE STATISTICS with n-distinct coefficients. These are now
specified using the mutually exclusive option WITH (ndistinct).
Author: Tomas Vondra, David Rowley
Reviewed-by: Kyotaro HORIGUCHI, Álvaro Herrera, Dean Rasheed, Robert Haas
and many other comments and contributions
Discussion: https://postgr.es/m/56f40b20-c464-fad2-ff39-06b668fac47c@2ndquadrant.com
A QueryEnvironment concept is added, which allows new types of
objects to be passed into queries from parsing on through
execution. At this point, the only thing implemented is a
collection of EphemeralNamedRelation objects -- relations which
can be referenced by name in queries, but do not exist in the
catalogs. The only type of ENR implemented is NamedTuplestore, but
provision is made to add more types fairly easily.
An ENR can carry its own TupleDesc or reference a relation in the
catalogs by relid.
Although these features can be used without SPI, convenience
functions are added to SPI so that ENRs can easily be used by code
run through SPI.
The initial use of all this is going to be transition tables in
AFTER triggers, but that will be added to each PL as a separate
commit.
An incidental effect of this patch is to produce a more informative
error message if an attempt is made to modify the contents of a CTE
from a referencing DML statement. No tests previously covered that
possibility, so one is added.
Kevin Grittner and Thomas Munro
Reviewed by Heikki Linnakangas, David Fetter, and Thomas Munro
with valuable comments and suggestions from many others
This extends the Aggregate node with two new features: HashAggregate
can now run multiple hashtables concurrently, and a new strategy
MixedAggregate populates hashtables while doing sorted grouping.
The planner will now attempt to save as many sorts as possible when
planning grouping sets queries, while not exceeding work_mem for the
estimated combined sizes of all hashtables used. No SQL-level changes
are required. There should be no user-visible impact other than the
new EXPLAIN output and possible changes to result ordering when ORDER
BY was not used (which affected a few regression tests). The
enable_hashagg option is respected.
Author: Andrew Gierth
Reviewers: Mark Dilger, Andres Freund
Discussion: https://postgr.es/m/87vatszyhj.fsf@news-spur.riddles.org.uk
This replaces the old, recursive tree-walk based evaluation, with
non-recursive, opcode dispatch based, expression evaluation.
Projection is now implemented as part of expression evaluation.
This both leads to significant performance improvements, and makes
future just-in-time compilation of expressions easier.
The speed gains primarily come from:
- non-recursive implementation reduces stack usage / overhead
- simple sub-expressions are implemented with a single jump, without
function calls
- sharing some state between different sub-expressions
- reduced amount of indirect/hard to predict memory accesses by laying
out operation metadata sequentially; including the avoidance of
nearly all of the previously used linked lists
- more code has been moved to expression initialization, avoiding
constant re-checks at evaluation time
Future just-in-time compilation (JIT) has become easier, as
demonstrated by released patches intended to be merged in a later
release, for primarily two reasons: Firstly, due to a stricter split
between expression initialization and evaluation, less code has to be
handled by the JIT. Secondly, due to the non-recursive nature of the
generated "instructions", less performance-critical code-paths can
easily be shared between interpreted and compiled evaluation.
The new framework allows for significant future optimizations. E.g.:
- basic infrastructure for to later reduce the per executor-startup
overhead of expression evaluation, by caching state in prepared
statements. That'd be helpful in OLTPish scenarios where
initialization overhead is measurable.
- optimizing the generated "code". A number of proposals for potential
work has already been made.
- optimizing the interpreter. Similarly a number of proposals have
been made here too.
The move of logic into the expression initialization step leads to some
backward-incompatible changes:
- Function permission checks are now done during expression
initialization, whereas previously they were done during
execution. In edge cases this can lead to errors being raised that
previously wouldn't have been, e.g. a NULL array being coerced to a
different array type previously didn't perform checks.
- The set of domain constraints to be checked, is now evaluated once
during expression initialization, previously it was re-built
every time a domain check was evaluated. For normal queries this
doesn't change much, but e.g. for plpgsql functions, which caches
ExprStates, the old set could stick around longer. The behavior
around might still change.
Author: Andres Freund, with significant changes by Tom Lane,
changes by Heikki Linnakangas
Reviewed-By: Tom Lane, Heikki Linnakangas
Discussion: https://postgr.es/m/20161206034955.bh33paeralxbtluv@alap3.anarazel.de
Commit 51ee6f3160 accidentally changed
the behavior around inheritance hierarchies; before, we always
considered parallel paths even for very small inheritance children,
because otherwise an inheritance hierarchy with even one small child
wouldn't be eligible for parallelism. That exception was inadverently
removed; put it back.
In passing, also adjust the degree-of-parallelism comptuation for
index-only scans not to consider the number of heap pages fetched.
Otherwise, we'll avoid parallel index-only scans on tables that are
mostly all-visible, which isn't especially logical.
Robert Haas and Amit Kapila, per a report from Ashutosh Sharma.
Discussion: http://postgr.es/m/CAE9k0PmgSoOHRd60SHu09aRVTHRSs8s6pmyhJKWHxWw9C_x+XA@mail.gmail.com
Fallout from fcec6caafa2: mark a variable in
set_tablefunc_size_estimates as used for asserts only.
Also, the planner_rte_fetch() call is pointless with assertions
disabled, so enclose it in a USE_ASSERT_CHECKING #ifdef; fix the same
problem in set_subquery_size_estimates().
First problem noted by David Rowley, whose compiler is noisier than mine
in this regard.
Like Gather, we spawn multiple workers and run the same plan in each
one; however, Gather Merge is used when each worker produces the same
output ordering and we want to preserve that output ordering while
merging together the streams of tuples from various workers. (In a
way, Gather Merge is like a hybrid of Gather and MergeAppend.)
This works out to a win if it saves us from having to perform an
expensive Sort. In cases where only a small amount of data would need
to be sorted, it may actually be faster to use a regular Gather node
and then sort the results afterward, because Gather Merge sometimes
needs to wait synchronously for tuples whereas a pure Gather generally
doesn't. But if this avoids an expensive sort then it's a win.
Rushabh Lathia, reviewed and tested by Amit Kapila, Thomas Munro,
and Neha Sharma, and reviewed and revised by me.
Discussion: http://postgr.es/m/CAGPqQf09oPX-cQRpBKS0Gq49Z+m6KBxgxd_p9gX8CKk_d75HoQ@mail.gmail.com
The index is scanned by a single process, but then all cooperating
processes can iterate jointly over the resulting set of heap blocks.
In the future, we might also want to support using a parallel bitmap
index scan to set up for a parallel bitmap heap scan, but that's a
job for another day.
Dilip Kumar, with some corrections and cosmetic changes by me. The
larger patch set of which this is a part has been reviewed and tested
by (at least) Andres Freund, Amit Khandekar, Tushar Ahuja, Rafia
Sabih, Haribabu Kommi, Thomas Munro, and me.
Discussion: http://postgr.es/m/CAFiTN-uc4=0WxRGfCzs-xfkMYcSEWUC-Fon6thkJGjkh9i=13A@mail.gmail.com
XMLTABLE is defined by the SQL/XML standard as a feature that allows
turning XML-formatted data into relational form, so that it can be used
as a <table primary> in the FROM clause of a query.
This new construct provides significant simplicity and performance
benefit for XML data processing; what in a client-side custom
implementation was reported to take 20 minutes can be executed in 400ms
using XMLTABLE. (The same functionality was said to take 10 seconds
using nested PostgreSQL XPath function calls, and 5 seconds using
XMLReader under PL/Python).
The implemented syntax deviates slightly from what the standard
requires. First, the standard indicates that the PASSING clause is
optional and that multiple XML input documents may be given to it; we
make it mandatory and accept a single document only. Second, we don't
currently support a default namespace to be specified.
This implementation relies on a new executor node based on a hardcoded
method table. (Because the grammar is fixed, there is no extensibility
in the current approach; further constructs can be implemented on top of
this such as JSON_TABLE, but they require changes to core code.)
Author: Pavel Stehule, Álvaro Herrera
Extensively reviewed by: Craig Ringer
Discussion: https://postgr.es/m/CAFj8pRAgfzMD-LoSmnMGybD0WsEznLHWap8DO79+-GTRAPR4qA@mail.gmail.com
In combination with 569174f1be, which
taught the btree AM how to perform parallel index scans, this allows
parallel index scan plans on btree indexes. This infrastructure
should be general enough to support parallel index scans for other
index AMs as well, if someone updates them to support parallel
scans.
Amit Kapila, reviewed and tested by Anastasia Lubennikova, Tushar
Ahuja, and Haribabu Kommi, and me.
Currently, we only need this logic in order to cost a Bitmap Heap
Scan. But a pending patch for Parallel Bitmap Heap Scan also uses
it to help figure out how many workers to use for the scan, which
has to be determined prior to costing. So, move the logic to
a separate function to make that easier.
Dilip Kumar. The patch series of which this is a part has been
reviewed by Andres Freund, Amit Khendekar, Tushar Ahuja, Rafia
Sabih, Haribabu Kommi, and me; it is not clear from the email
discussion which of those people have looked specifically at this
part.
Discussion: http://postgr.es/m/CAFiTN-v3QYNJEZnnmKCeATuLbN-h9tMVfeEF0+BrouYDqjXgwg@mail.gmail.com
For a partial path, the cardinality estimate needs to reflect the
number of rows we think each worker will see, rather than the total
number of rows; otherwise, costing will go wrong. The previous coding
got this completely wrong for parallel joins.
Unfortunately, this change may destabilize plans for users of 9.6 who
have enabled parallel query, but since 9.6 is still fairly new I'm
hoping expectations won't be too settled yet. Also, this is really a
brown-paper-bag bug, so leaving it unfixed for the entire lifetime of
9.6 seems unwise.
Related reports (whose import I initially failed to recognize) by
Tomas Vondra and Tom Lane.
Discussion: http://postgr.es/m/CA+TgmoaDxZ5z5Kw_oCQoymNxNoVaTCXzPaODcOuao=CzK8dMZw@mail.gmail.com
This case wasn't thought through sufficiently in commit 100340e2d.
It's true that the FK proves that every outer row has a match in the
inner table, but we forgot that some of the inner rows might be filtered
away by WHERE conditions located within the semijoin's RHS.
If the RHS is just one table, we can reasonably take the semijoin
selectivity as equal to the fraction of the referenced table's rows
that are expected to survive its restriction clauses.
If the RHS is a join, it's not clear how much of the referenced table
might get through the join, so fall back to the same rule we were
already using for other outer-join cases: use the minimum of the
regular per-clause selectivity estimates. This gives the same result
as if we hadn't considered the FK at all when there's a single FK
column, but it should still help for multi-column FKs, which is the
case that 100340e2d is really meant to help with.
Back-patch to 9.6 where the previous commit came in.
Discussion: https://postgr.es/m/16149.1481835103@sss.pgh.pa.us
Commit 7a2fe9bd0 improved merge append so that replacement of a tuple
takes log(N) operations, not twice log(N). Since cost_merge_append knew
about that explicitly, we should adjust it. This probably makes little
difference in practice, but the obsolete comment is confusing.
Ideally this would have been put in in 9.3 with the underlying behavior
change; but I'm not going to back-patch it, since there's some small chance
of changing a plan choice that somebody's optimized for.
Thomas Munro
Discussion: <CAEepm=0WQBSvuYcMOUj4Ga4NXpu2J=ejZcE=e=eiTjTX-6_gDw@mail.gmail.com>
cost_rescan assumed that we don't need to rebuild the hash table when
rescanning a hash join. However, that's currently only true for
single-batch joins; for a multi-batch join we must charge full freight.
This probably has escaped notice because we'd be unlikely to put a hash
join on the inside of a nestloop anyway. Nonetheless, it's wrong.
Fix in HEAD, but don't backpatch for fear of destabilizing plans in
stable releases.
I'd been wondering why I was sometimes seeing fractional rowcount
estimates in parallel-query situations, and this seems to be the
reason. (You won't see the fractional parts in EXPLAIN, because it
prints rowcounts with %.0f, but they are apparent in the debugger.)
A fractional rowcount is not any saner for a partial path than any
other kind of path, and it's equally likely to break cost estimation
for higher paths, so apply clamp_row_est() like we do in other places.
VS2013 apparently has a problem with taking the address of a formal
parameter in some cases. We do that elsewhere without trouble, but
in this case the address is being passed to a subroutine that will
probably get inlined, so maybe the combination of those things is
what tickles the bug. Anyway, introducing an extra copy of the
parameter value is enough to work around it. Per trouble report
from Umair Shahid.
Report: <CAM184AcjqKYZSdQqBHDrnENXHhW=mXbUC46QYPJ=nAh0gUHCGA@mail.gmail.com>
This patch provides a new implementation of the logic added by commit
137805f89 and later removed by 77ba61080. It differs from the original
primarily in expending much less effort per joinrel in large queries,
which it accomplishes by doing most of the matching work once per query not
once per joinrel. Hopefully, it's also less buggy and better commented.
The never-documented enable_fkey_estimates GUC remains gone.
There remains work to be done to make the selectivity estimates account
for nulls in FK referencing columns; but that was true of the original
patch as well. We may be able to address this point later in beta.
In the meantime, any error should be in the direction of overestimating
rather than underestimating joinrel sizes, which seems like the direction
we want to err in.
Tomas Vondra and Tom Lane
Discussion: <31041.1465069446@sss.pgh.pa.us>
This terminology provoked widespread complaints. So, instead, rename
the GUC max_parallel_degree to max_parallel_workers_per_gather
(leaving room for a possible future GUC max_parallel_workers that acts
as a system-wide limit), and rename the parallel_degree reloption to
parallel_workers. Rename structure members to match.
These changes create a dump/restore hazard for users of PostgreSQL
9.6beta1 who have set the reloption (or applied the GUC using ALTER
USER or ALTER DATABASE).
This commit reverts 137805f89 as well as the associated commits 015e88942,
5306df283, and 68d704edb. We found multiple bugs in this feature, and
there was concern about possible planner slowdown (though to be fair,
exhibiting a very large slowdown proved difficult). The way forward
requires a considerable rewrite, which may or may not be possible to
accomplish in time for beta2. In my judgment reviewing the rewrite will
be easier to accomplish starting from a clean slate, so let's temporarily
revert what's there now. This also leaves us in a safe state if it turns
out to be necessary to postpone the rewrite to the next development cycle.
Discussion: <20160429102531.GA13701@huehner.biz>
Discussion is still underway as to whether to revert the entire patch
that added this function, but that discussion may not conclude before
beta1. So, in the meantime, let's do at least this much.
David Rowley
Change max_parallel_degree default from 0 to 2. It is possible that
this is not a good idea, or that we should go with 1 worker rather
than 2, but we won't find out without trying it. Along the way,
reword the documentation for max_parallel_degree a little bit to
hopefully make it more clear.
Discussion: 20160420174631.3qjjhpwsvvx5bau5@alap3.anarazel.de
We still use replacement selection for the first run of the sort only
and only when the number of tuples is relatively small. Otherwise,
the first run, and subsequent runs in all cases, are produced using
quicksort. This tends to be faster except perhaps for very small
amounts of working memory.
Peter Geoghegan, reviewed by Tomas Vondra, Jeff Janes, Mithun Cy,
Greg Stark, and me.
In cases where joins use multiple columns we currently assess each join
separately causing gross mis-estimates for join cardinality.
This patch adds use of FK information for the first time into the
planner. When FKs are present and we have multi-column join information,
plan estimates will be drastically improved. Cases with multiple FKs
are handled, though partial matches are ignored currently.
Net effect is substantial performance improvements for joins in many
common cases. Additional planning time is isolated to cases that are
currently performing poorly, measured at 0.08 - 0.15 ms.
Please watch for planner performance regressions; circumstances seem
unlikely but the law of unintended consequences may apply somewhen.
Additional complex tests welcome to prove this before release.
Tests can be performed using SET enable_fkey_estimates = on | off
using scripts provided during Hackers discussions, message id:
552335D9.3090707@2ndquadrant.com
Authors: Tomas Vondra and David Rowley
Reviewed and tested by Simon Riggs, adding comments only
Previously, the planner would reject an index-only scan if any restriction
clause for its table used a column not available from the index, even
if that restriction clause would later be dropped from the plan entirely
because it's implied by the index's predicate. This is a fairly common
situation for partial indexes because predicates using columns not included
in the index are often the most useful kind of predicate, and we have to
duplicate (or at least imply) the predicate in the WHERE clause in order
to get the index to be considered at all. So index-only scans were
essentially unavailable with such partial indexes.
To fix, we have to do detection of implied-by-predicate clauses much
earlier in the planner. This patch puts it in check_index_predicates
(nee check_partial_indexes), meaning it gets done for every partial index,
whereas we previously only considered this issue at createplan time,
so that the work was only done for an index actually selected for use.
That could result in a noticeable planning slowdown for queries against
tables with many partial indexes. However, testing suggested that there
isn't really a significant cost, especially not with reasonable numbers
of partial indexes. We do get a small additional benefit, which is that
cost_index is more accurate since it correctly discounts the evaluation
cost of clauses that will be removed. We can also avoid considering such
clauses as potential indexquals, which saves useless matching cycles in
the case where the predicate columns aren't in the index, and prevents
generating bogus plans that double-count the clause's selectivity when
the columns are in the index.
Tomas Vondra and Kyotaro Horiguchi, reviewed by Kevin Grittner and
Konstantin Knizhnik, and whacked around a little by me
cost_subplan() supposed that the given subplan must have plan_rows > 0,
which as far as I can tell was true until recent refactoring of the
code in createplan.c; but now that code allows the Result for a provably
empty subquery to have plan_rows = 0. Rather than undo that change,
put in a clamp to prevent zero divide.
get_cheapest_fractional_path() likewise supposed that best_path->rows > 0.
This assumption has been wrong for longer. It's actually harmless given
IEEE float math, because a positive value divided by zero gives +Infinity
and compare_fractional_path_costs() will do the right thing with that.
Still, best not to assume that.
final_cost_nestloop() also seems to have some risks in this area, so
borrow the clamping logic already present in the mergejoin cost functions.
Lastly, remove unnecessary clamp_row_est() in planner.c's calls to
get_number_of_groups(). The only thing that function does with path_rows
is pass it to estimate_num_groups() which already has an internal clamp,
so we don't need the extra call; and if we did, the callers are arguably
the wrong place for it anyway.
First two items reported by Piotr Stefaniak, the others are products
of my nosing around for similar problems. No back-patch since there's
no evidence that problems arise in the back branches.
Parallel workers can now partially aggregate the data and pass the
transition values back to the leader, which can combine the partial
results to produce the final answer.
David Rowley, based on earlier work by Haribabu Kommi. Reviewed by
Álvaro Herrera, Tomas Vondra, Amit Kapila, James Sewell, and me.
In commit 19a541143a I did not make PathTarget a subtype of Node,
and embedded a RelOptInfo's reltarget directly into it rather than having
a separately-allocated Node. In hindsight that was misguided
micro-optimization, enabled by the fact that at that point we didn't have
any Paths with custom PathTargets. Now that PathTarget processing has
been fleshed out some more, it's easier to see that it's better to have
PathTarget as an indepedent Node type, even if it does cost us one more
palloc to create a RelOptInfo. So change it while we still can.
This commit just changes the representation, without doing anything more
interesting than that.
I've been saying we needed to do this for more than five years, and here it
finally is. This patch removes the ever-growing tangle of spaghetti logic
that grouping_planner() used to use to try to identify the best plan for
post-scan/join query steps. Now, there is (nearly) independent
consideration of each execution step, and entirely separate construction of
Paths to represent each of the possible ways to do that step. We choose
the best Path or set of Paths using the same add_path() logic that's been
used inside query_planner() for years.
In addition, this patch removes the old restriction that subquery_planner()
could return only a single Plan. It now returns a RelOptInfo containing a
set of Paths, just as query_planner() does, and the parent query level can
use each of those Paths as the basis of a SubqueryScanPath at its level.
This allows finding some optimizations that we missed before, wherein a
subquery was capable of returning presorted data and thereby avoiding a
sort in the parent level, making the overall cost cheaper even though
delivering sorted output was not the cheapest plan for the subquery in
isolation. (A couple of regression test outputs change in consequence of
that. However, there is very little change in visible planner behavior
overall, because the point of this patch is not to get immediate planning
benefits but to create the infrastructure for future improvements.)
There is a great deal left to do here. This patch unblocks a lot of
planner work that was basically impractical in the old code structure,
such as allowing FDWs to implement remote aggregation, or rewriting
plan_set_operations() to allow consideration of multiple implementation
orders for set operations. (The latter will likely require a full
rewrite of plan_set_operations(); what I've done here is only to fix it
to return Paths not Plans.) I have also left unfinished some localized
refactoring in createplan.c and planner.c, because it was not necessary
to get this patch to a working state.
Thanks to Robert Haas, David Rowley, and Amit Kapila for review.
Up to now, there's been an assumption that all Paths for a given relation
compute the same output column set (targetlist). However, there are good
reasons to remove that assumption. For example, an indexscan on an
expression index might be able to return the value of an expensive function
"for free". While we have the ability to generate such a plan today in
simple cases, we don't have a way to model that it's cheaper than a plan
that computes the function from scratch, nor a way to create such a plan
in join cases (where the function computation would normally happen at
the topmost join node). Also, we need this so that we can have Paths
representing post-scan/join steps, where the targetlist may well change
from one step to the next. Therefore, invent a "struct PathTarget"
representing the columns we expect a plan step to emit. It's convenient
to include the output tuple width and tlist evaluation cost in this struct,
and there will likely be additional fields in future.
While Path nodes that actually do have custom outputs will need their own
PathTargets, it will still be true that most Paths for a given relation
will compute the same tlist. To reduce the overhead added by this patch,
keep a "default PathTarget" in RelOptInfo, and allow Paths that compute
that column set to just point to their parent RelOptInfo's reltarget.
(In the patch as committed, actually every Path is like that, since we
do not yet have any cases of custom PathTargets.)
I took this opportunity to provide some more-honest costing of
PlaceHolderVar evaluation. Up to now, the assumption that "scan/join
reltargetlists have cost zero" was applied not only to Vars, where it's
reasonable, but also PlaceHolderVars where it isn't. Now, we add the eval
cost of a PlaceHolderVar's expression to the first plan level where it can
be computed, by including it in the PathTarget cost field and adding that
to the cost estimates for Paths. This isn't perfect yet but it's much
better than before, and there is a way forward to improve it more. This
costing change affects the join order chosen for a couple of the regression
tests, changing expected row ordering.
The core innovation of this patch is the introduction of the concept
of a partial path; that is, a path which if executed in parallel will
generate a subset of the output rows in each process. Gathering a
partial path produces an ordinary (complete) path. This allows us to
generate paths for parallel joins by joining a partial path for one
side (which at the baserel level is currently always a Partial Seq
Scan) to an ordinary path on the other side. This is subject to
various restrictions at present, especially that this strategy seems
unlikely to be sensible for merge joins, so only nested loops and
hash joins paths are generated.
This also allows an Append node to be pushed below a Gather node in
the case of a partitioned table.
Testing revealed that early versions of this patch made poor decisions
in some cases, which turned out to be caused by the fact that the
original cost model for Parallel Seq Scan wasn't very good. So this
patch tries to make some modest improvements in that area.
There is much more to be done in the area of generating good parallel
plans in all cases, but this seems like a useful step forward.
Patch by me, reviewed by Dilip Kumar and Amit Kapila.
This patch reduces pg_am to just two columns, a name and a handler
function. All the data formerly obtained from pg_am is now provided
in a C struct returned by the handler function. This is similar to
the designs we've adopted for FDWs and tablesample methods. There
are multiple advantages. For one, the index AM's support functions
are now simple C functions, making them faster to call and much less
error-prone, since the C compiler can now check function signatures.
For another, this will make it far more practical to define index access
methods in installable extensions.
A disadvantage is that SQL-level code can no longer see attributes
of index AMs; in particular, some of the crosschecks in the opr_sanity
regression test are no longer possible from SQL. We've addressed that
by adding a facility for the index AM to perform such checks instead.
(Much more could be done in that line, but for now we're content if the
amvalidate functions more or less replace what opr_sanity used to do.)
We might also want to expose some sort of reporting functionality, but
this patch doesn't do that.
Alexander Korotkov, reviewed by Petr Jelínek, and rather heavily
editorialized on by me.
Peter Geoghegan
Andres Freund
Peter Eisentraut
Jeff Davis
Tom Lane
Tomas Vondra
Bruce Momjian
Michael Paquier
Alvaro Herrera
Robert Haas
Peter Eisentraut
Andrew Dunstan
Simon Riggs
Kevin Grittner
Andrew Gierth