Skip to content

Commit 5fd9dfa

Browse files
bmomjianbmomjian
committed
Move pg_stat_statements query jumbling to core.
Add compute_query_id GUC to control whether a query identifier should be computed by the core (off by default). It's thefore now possible to disable core queryid computation and use pg_stat_statements with a different algorithm to compute the query identifier by using a third-party module. To ensure that a single source of query identifier can be used and is well defined, modules that calculate a query identifier should throw an error if compute_query_id specified to compute a query id and if a query idenfitier was already calculated. Discussion: https://postgr.es/m/20210407125726.tkvjdbw76hxnpwfi@nol Author: Julien Rouhaud Reviewed-by: Alvaro Herrera, Nitin Jadhav, Zhihong Yu
1 parent a282ee6 commit 5fd9dfa

File tree

13 files changed

+995
-785
lines changed

13 files changed

+995
-785
lines changed

‎contrib/pg_stat_statements/pg_stat_statements.c

+25-780
Large diffs are not rendered by default.

‎contrib/pg_stat_statements/pg_stat_statements.conf

+1
Original file line numberDiff line numberDiff line change
@@ -1 +1,2 @@
11
shared_preload_libraries = 'pg_stat_statements'
2+
compute_query_id = on

‎doc/src/sgml/config.sgml

+25
Original file line numberDiff line numberDiff line change
@@ -7622,6 +7622,31 @@ COPY postgres_log FROM '/full/path/to/logfile.csv' WITH csv;
76227622
<title>Statistics Monitoring</title>
76237623
<variablelist>
76247624

7625+
<varlistentry id="guc-compute-query-id" xreflabel="compute_query_id">
7626+
<term><varname>compute_query_id</varname> (<type>boolean</type>)
7627+
<indexterm>
7628+
<primary><varname>compute_query_id</varname> configuration parameter</primary>
7629+
</indexterm>
7630+
</term>
7631+
<listitem>
7632+
<para>
7633+
Enables in-core computation of a query identifier. The <xref
7634+
linkend="pgstatstatements"/> extension requires a query identifier
7635+
to be computed. Note that an external module can alternatively
7636+
be used if the in-core query identifier computation method
7637+
isn't acceptable. In this case, in-core computation should
7638+
remain disabled. The default is <literal>off</literal>.
7639+
</para>
7640+
<note>
7641+
<para>
7642+
To ensure that a only one query identifier is calculated and
7643+
displayed, extensions that calculate query identifiers should
7644+
throw an error if a query identifier has already been computed.
7645+
</para>
7646+
</note>
7647+
</listitem>
7648+
</varlistentry>
7649+
76257650
<varlistentry>
76267651
<term><varname>log_statement_stats</varname> (<type>boolean</type>)
76277652
<indexterm>

‎doc/src/sgml/pgstatstatements.sgml

+19-1
Original file line numberDiff line numberDiff line change
@@ -20,6 +20,14 @@
2020
This means that a server restart is needed to add or remove the module.
2121
</para>
2222

23+
<para>
24+
The module will not track statistics unless query
25+
identifiers are calculated. This can be done by enabling <xref
26+
linkend="guc-compute-query-id"/> or using a third-party module that
27+
computes its own query identifiers. Note that all statistics tracked
28+
by this module must be reset if the query identifier method is changed.
29+
</para>
30+
2331
<para>
2432
When <filename>pg_stat_statements</filename> is loaded, it tracks
2533
statistics across all databases of the server. To access and manipulate
@@ -84,7 +92,7 @@
8492
<structfield>queryid</structfield> <type>bigint</type>
8593
</para>
8694
<para>
87-
Internal hash code, computed from the statement's parse tree
95+
Hash code to identify identical normalized queries.
8896
</para></entry>
8997
</row>
9098

@@ -386,6 +394,16 @@
386394
are compared strictly on the basis of their textual query strings, however.
387395
</para>
388396

397+
<note>
398+
<para>
399+
The following details about constant replacement and
400+
<structfield>queryid</structfield> only applies when <xref
401+
linkend="guc-compute-query-id"/> is enabled. If you use an external
402+
module instead to compute <structfield>queryid</structfield>, you
403+
should refer to its documentation for details.
404+
</para>
405+
</note>
406+
389407
<para>
390408
When a constant's value has been ignored for purposes of matching the query
391409
to other queries, the constant is replaced by a parameter symbol, such

‎src/backend/parser/analyze.c

+12-2
Original file line numberDiff line numberDiff line change
@@ -46,6 +46,8 @@
4646
#include "parser/parsetree.h"
4747
#include "rewrite/rewriteManip.h"
4848
#include "utils/builtins.h"
49+
#include "utils/guc.h"
50+
#include "utils/queryjumble.h"
4951
#include "utils/rel.h"
5052

5153

@@ -107,6 +109,7 @@ parse_analyze(RawStmt *parseTree, const char *sourceText,
107109
{
108110
ParseState *pstate = make_parsestate(NULL);
109111
Query *query;
112+
JumbleState *jstate = NULL;
110113

111114
Assert(sourceText != NULL); /* required as of 8.4 */
112115

@@ -119,8 +122,11 @@ parse_analyze(RawStmt *parseTree, const char *sourceText,
119122

120123
query = transformTopLevelStmt(pstate, parseTree);
121124

125+
if (compute_query_id)
126+
jstate = JumbleQuery(query, sourceText);
127+
122128
if (post_parse_analyze_hook)
123-
(*post_parse_analyze_hook) (pstate, query);
129+
(*post_parse_analyze_hook) (pstate, query, jstate);
124130

125131
free_parsestate(pstate);
126132

@@ -140,6 +146,7 @@ parse_analyze_varparams(RawStmt *parseTree, const char *sourceText,
140146
{
141147
ParseState *pstate = make_parsestate(NULL);
142148
Query *query;
149+
JumbleState *jstate = NULL;
143150

144151
Assert(sourceText != NULL); /* required as of 8.4 */
145152

@@ -152,8 +159,11 @@ parse_analyze_varparams(RawStmt *parseTree, const char *sourceText,
152159
/* make sure all is well with parameter types */
153160
check_variable_parameters(pstate, query);
154161

162+
if (compute_query_id)
163+
jstate = JumbleQuery(query, sourceText);
164+
155165
if (post_parse_analyze_hook)
156-
(*post_parse_analyze_hook) (pstate, query);
166+
(*post_parse_analyze_hook) (pstate, query, jstate);
157167

158168
free_parsestate(pstate);
159169

‎src/backend/tcop/postgres.c

+5-1
Original file line numberDiff line numberDiff line change
@@ -668,6 +668,7 @@ pg_analyze_and_rewrite_params(RawStmt *parsetree,
668668
ParseState *pstate;
669669
Query *query;
670670
List *querytree_list;
671+
JumbleState *jstate = NULL;
671672

672673
Assert(query_string != NULL); /* required as of 8.4 */
673674

@@ -686,8 +687,11 @@ pg_analyze_and_rewrite_params(RawStmt *parsetree,
686687

687688
query = transformTopLevelStmt(pstate, parsetree);
688689

690+
if (compute_query_id)
691+
jstate = JumbleQuery(query, query_string);
692+
689693
if (post_parse_analyze_hook)
690-
(*post_parse_analyze_hook) (pstate, query);
694+
(*post_parse_analyze_hook) (pstate, query, jstate);
691695

692696
free_parsestate(pstate);
693697

‎src/backend/utils/misc/Makefile

+1
Original file line numberDiff line numberDiff line change
@@ -22,6 +22,7 @@ OBJS = \
2222
pg_rusage.o \
2323
ps_status.o \
2424
queryenvironment.o \
25+
queryjumble.o \
2526
rls.o \
2627
sampling.o \
2728
superuser.o \

‎src/backend/utils/misc/guc.c

+10
Original file line numberDiff line numberDiff line change
@@ -534,6 +534,7 @@ extern const struct config_enum_entry dynamic_shared_memory_options[];
534534
/*
535535
* GUC option variables that are exported from this module
536536
*/
537+
bool compute_query_id = false;
537538
bool log_duration = false;
538539
bool Debug_print_plan = false;
539540
bool Debug_print_parse = false;
@@ -1458,6 +1459,15 @@ static struct config_bool ConfigureNamesBool[] =
14581459
true,
14591460
NULL, NULL, NULL
14601461
},
1462+
{
1463+
{"compute_query_id", PGC_SUSET, STATS_MONITORING,
1464+
gettext_noop("Compute query identifiers."),
1465+
NULL
1466+
},
1467+
&compute_query_id,
1468+
false,
1469+
NULL, NULL, NULL
1470+
},
14611471
{
14621472
{"log_parser_stats", PGC_SUSET, STATS_MONITORING,
14631473
gettext_noop("Writes parser performance statistics to the server log."),

‎src/backend/utils/misc/postgresql.conf.sample

+1
Original file line numberDiff line numberDiff line change
@@ -596,6 +596,7 @@
596596

597597
# - Monitoring -
598598

599+
#compute_query_id = off
599600
#log_parser_stats = off
600601
#log_planner_stats = off
601602
#log_executor_stats = off

0 commit comments

Comments
 (0)