Add memory/disk usage for Material nodes in EXPLAIN

Up until now, there was no ability to easily determine if a Material
node caused the underlying tuplestore to spill to disk or even see how
much memory the tuplestore used if it didn't.

Here we add some new functions to tuplestore.c to query this information
and add some additional output in EXPLAIN ANALYZE to display this
information for the Material node.

There are a few other executor node types that use tuplestores, so we
could also consider adding these details to the EXPLAIN ANALYZE for
those nodes too.  Let's consider those independently from this.  Having
the tuplestore.c infrastructure in to allow that is step 1.

Author: David Rowley
Reviewed-by: Matthias van de Meent, Dmitry Dolgov
Discussion: https://postgr.es/m/CAApHDvp5Py9g4Rjq7_inL3-MCK1Co2CRt_YWFwTU2zfQix0p4A@mail.gmail.com
This commit is contained in:
David Rowley 2024-07-05 14:05:08 +12:00
parent aa86129e19
commit 1eff8279d4
5 changed files with 148 additions and 12 deletions

View File

@ -125,6 +125,7 @@ static void show_sort_info(SortState *sortstate, ExplainState *es);
static void show_incremental_sort_info(IncrementalSortState *incrsortstate,
ExplainState *es);
static void show_hash_info(HashState *hashstate, ExplainState *es);
static void show_material_info(MaterialState *mstate, ExplainState *es);
static void show_memoize_info(MemoizeState *mstate, List *ancestors,
ExplainState *es);
static void show_hashagg_info(AggState *aggstate, ExplainState *es);
@ -2251,6 +2252,9 @@ ExplainNode(PlanState *planstate, List *ancestors,
case T_Hash:
show_hash_info(castNode(HashState, planstate), es);
break;
case T_Material:
show_material_info(castNode(MaterialState, planstate), es);
break;
case T_Memoize:
show_memoize_info(castNode(MemoizeState, planstate), ancestors,
es);
@ -3322,6 +3326,39 @@ show_hash_info(HashState *hashstate, ExplainState *es)
}
}
/*
* Show information on material node, storage method and maximum memory/disk
* space used.
*/
static void
show_material_info(MaterialState *mstate, ExplainState *es)
{
Tuplestorestate *tupstore;
const char *storageType;
int64 spaceUsedKB;
if (!es->analyze)
return;
tupstore = mstate->tuplestorestate;
storageType = tuplestore_storage_type_name(tupstore);
spaceUsedKB = BYTES_TO_KILOBYTES(tuplestore_space_used(tupstore));
if (es->format != EXPLAIN_FORMAT_TEXT)
{
ExplainPropertyText("Storage", storageType, es);
ExplainPropertyInteger("Maximum Storage", "kB", spaceUsedKB, es);
}
else
{
ExplainIndentText(es);
appendStringInfo(es->str,
"Storage: %s Maximum Storage: " INT64_FORMAT "kB\n",
storageType,
spaceUsedKB);
}
}
/*
* Show information on memoize hits/misses/evictions and memory usage.
*/

View File

@ -109,6 +109,7 @@ struct Tuplestorestate
bool truncated; /* tuplestore_trim has removed tuples? */
int64 availMem; /* remaining memory available, in bytes */
int64 allowedMem; /* total memory allowed, in bytes */
int64 maxSpace; /* maximum space used in memory */
int64 tuples; /* number of tuples added */
BufFile *myfile; /* underlying file, or NULL if none */
MemoryContext context; /* memory context for holding tuples */
@ -238,6 +239,7 @@ static Tuplestorestate *tuplestore_begin_common(int eflags,
int maxKBytes);
static void tuplestore_puttuple_common(Tuplestorestate *state, void *tuple);
static void dumptuples(Tuplestorestate *state);
static void tuplestore_updatemax(Tuplestorestate *state);
static unsigned int getlen(Tuplestorestate *state, bool eofOK);
static void *copytup_heap(Tuplestorestate *state, void *tup);
static void writetup_heap(Tuplestorestate *state, void *tup);
@ -262,6 +264,7 @@ tuplestore_begin_common(int eflags, bool interXact, int maxKBytes)
state->truncated = false;
state->allowedMem = maxKBytes * 1024L;
state->availMem = state->allowedMem;
state->maxSpace = 0;
state->myfile = NULL;
state->context = CurrentMemoryContext;
state->resowner = CurrentResourceOwner;
@ -420,6 +423,9 @@ tuplestore_clear(Tuplestorestate *state)
int i;
TSReadPointer *readptr;
/* update the maxSpace before doing any USEMEM/FREEMEM adjustments */
tuplestore_updatemax(state);
if (state->myfile)
BufFileClose(state->myfile);
state->myfile = NULL;
@ -1402,6 +1408,9 @@ tuplestore_trim(Tuplestorestate *state)
Assert(nremove >= state->memtupdeleted);
Assert(nremove <= state->memtupcount);
/* before freeing any memory, update maxSpace */
tuplestore_updatemax(state);
/* Release no-longer-needed tuples */
for (i = state->memtupdeleted; i < nremove; i++)
{
@ -1444,6 +1453,49 @@ tuplestore_trim(Tuplestorestate *state)
}
}
/*
* tuplestore_updatemax
* Update maxSpace field
*/
static void
tuplestore_updatemax(Tuplestorestate *state)
{
if (state->status == TSS_INMEM)
state->maxSpace = Max(state->maxSpace,
state->allowedMem - state->availMem);
}
/*
* tuplestore_storage_type_name
* Return a string description of the storage method used to store the
* tuples.
*/
const char *
tuplestore_storage_type_name(Tuplestorestate *state)
{
if (state->status == TSS_INMEM)
return "Memory";
else
return "Disk";
}
/*
* tuplestore_space_used
* Return the maximum space used in memory unless the tuplestore has spilled
* to disk, in which case, return the disk space used.
*/
int64
tuplestore_space_used(Tuplestorestate *state)
{
/* First, update the maxSpace field */
tuplestore_updatemax(state);
if (state->status == TSS_INMEM)
return state->maxSpace;
else
return BufFileSize(state->myfile);
}
/*
* tuplestore_in_memory
*
@ -1513,6 +1565,7 @@ writetup_heap(Tuplestorestate *state, void *tup)
if (state->backward) /* need trailing length word? */
BufFileWrite(state->myfile, &tuplen, sizeof(tuplen));
/* no need to call tuplestore_updatemax() when not in TSS_INMEM */
FREEMEM(state, GetMemoryChunkSpace(tuple));
heap_free_minimal_tuple(tuple);
}

View File

@ -65,6 +65,10 @@ extern void tuplestore_copy_read_pointer(Tuplestorestate *state,
extern void tuplestore_trim(Tuplestorestate *state);
extern const char *tuplestore_storage_type_name(Tuplestorestate *state);
extern int64 tuplestore_space_used(Tuplestorestate *state);
extern bool tuplestore_in_memory(Tuplestorestate *state);
extern bool tuplestore_gettupleslot(Tuplestorestate *state, bool forward,

View File

@ -1,6 +1,24 @@
--
-- Test partitioning planner code
--
-- Helper function which can be used for masking out portions of EXPLAIN
-- ANALYZE which could contain information that's not consistent on all
-- platforms.
create function explain_analyze(query text) returns setof text
language plpgsql as
$$
declare
ln text;
begin
for ln in
execute format('explain (analyze, costs off, summary off, timing off) %s',
query)
loop
ln := regexp_replace(ln, 'Maximum Storage: \d+', 'Maximum Storage: N');
return next ln;
end loop;
end;
$$;
-- Force generic plans to be used for all prepared statements in this file.
set plan_cache_mode = force_generic_plan;
create table lp (a char) partition by list (a);
@ -2826,9 +2844,9 @@ deallocate ab_q5;
deallocate ab_q6;
-- UPDATE on a partition subtree has been seen to have problems.
insert into ab values (1,2);
explain (analyze, costs off, summary off, timing off)
update ab_a1 set b = 3 from ab where ab.a = 1 and ab.a = ab_a1.a;
QUERY PLAN
select explain_analyze('
update ab_a1 set b = 3 from ab where ab.a = 1 and ab.a = ab_a1.a;');
explain_analyze
-------------------------------------------------------------------------------------------
Update on ab_a1 (actual rows=0 loops=1)
Update on ab_a1_b1 ab_a1_1
@ -2851,6 +2869,7 @@ update ab_a1 set b = 3 from ab where ab.a = 1 and ab.a = ab_a1.a;
-> Bitmap Index Scan on ab_a1_b3_a_idx (actual rows=1 loops=1)
Index Cond: (a = 1)
-> Materialize (actual rows=1 loops=1)
Storage: Memory Maximum Storage: NkB
-> Append (actual rows=1 loops=1)
-> Bitmap Heap Scan on ab_a1_b1 ab_1 (actual rows=0 loops=1)
Recheck Cond: (a = 1)
@ -2866,7 +2885,7 @@ update ab_a1 set b = 3 from ab where ab.a = 1 and ab.a = ab_a1.a;
Heap Blocks: exact=1
-> Bitmap Index Scan on ab_a1_b3_a_idx (actual rows=1 loops=1)
Index Cond: (a = 1)
(36 rows)
(37 rows)
table ab;
a | b
@ -2877,9 +2896,9 @@ table ab;
-- Test UPDATE where source relation has run-time pruning enabled
truncate ab;
insert into ab values (1, 1), (1, 2), (1, 3), (2, 1);
explain (analyze, costs off, summary off, timing off)
update ab_a1 set b = 3 from ab_a2 where ab_a2.b = (select 1);
QUERY PLAN
select explain_analyze('
update ab_a1 set b = 3 from ab_a2 where ab_a2.b = (select 1);');
explain_analyze
------------------------------------------------------------------------------
Update on ab_a1 (actual rows=0 loops=1)
Update on ab_a1_b1 ab_a1_1
@ -2893,6 +2912,7 @@ update ab_a1 set b = 3 from ab_a2 where ab_a2.b = (select 1);
-> Seq Scan on ab_a1_b2 ab_a1_2 (actual rows=1 loops=1)
-> Seq Scan on ab_a1_b3 ab_a1_3 (actual rows=1 loops=1)
-> Materialize (actual rows=1 loops=3)
Storage: Memory Maximum Storage: NkB
-> Append (actual rows=1 loops=1)
-> Seq Scan on ab_a2_b1 ab_a2_1 (actual rows=1 loops=1)
Filter: (b = (InitPlan 1).col1)
@ -2900,7 +2920,7 @@ update ab_a1 set b = 3 from ab_a2 where ab_a2.b = (select 1);
Filter: (b = (InitPlan 1).col1)
-> Seq Scan on ab_a2_b3 ab_a2_3 (never executed)
Filter: (b = (InitPlan 1).col1)
(19 rows)
(20 rows)
select tableoid::regclass, * from ab;
tableoid | a | b
@ -4419,3 +4439,4 @@ explain (costs off) select * from hp_contradict_test where a === 1 and b === 1 a
drop table hp_contradict_test;
drop operator class part_test_int4_ops2 using hash;
drop operator ===(int4, int4);
drop function explain_analyze(text);

View File

@ -2,6 +2,25 @@
-- Test partitioning planner code
--
-- Helper function which can be used for masking out portions of EXPLAIN
-- ANALYZE which could contain information that's not consistent on all
-- platforms.
create function explain_analyze(query text) returns setof text
language plpgsql as
$$
declare
ln text;
begin
for ln in
execute format('explain (analyze, costs off, summary off, timing off) %s',
query)
loop
ln := regexp_replace(ln, 'Maximum Storage: \d+', 'Maximum Storage: N');
return next ln;
end loop;
end;
$$;
-- Force generic plans to be used for all prepared statements in this file.
set plan_cache_mode = force_generic_plan;
@ -676,15 +695,15 @@ deallocate ab_q6;
-- UPDATE on a partition subtree has been seen to have problems.
insert into ab values (1,2);
explain (analyze, costs off, summary off, timing off)
update ab_a1 set b = 3 from ab where ab.a = 1 and ab.a = ab_a1.a;
select explain_analyze('
update ab_a1 set b = 3 from ab where ab.a = 1 and ab.a = ab_a1.a;');
table ab;
-- Test UPDATE where source relation has run-time pruning enabled
truncate ab;
insert into ab values (1, 1), (1, 2), (1, 3), (2, 1);
explain (analyze, costs off, summary off, timing off)
update ab_a1 set b = 3 from ab_a2 where ab_a2.b = (select 1);
select explain_analyze('
update ab_a1 set b = 3 from ab_a2 where ab_a2.b = (select 1);');
select tableoid::regclass, * from ab;
drop table ab, lprt_a;
@ -1318,3 +1337,5 @@ explain (costs off) select * from hp_contradict_test where a === 1 and b === 1 a
drop table hp_contradict_test;
drop operator class part_test_int4_ops2 using hash;
drop operator ===(int4, int4);
drop function explain_analyze(text);