Merge file storage feature and look-a-like

Author: danolivo

expected/aqo_learn.out

@@ -540,7 +540,7 @@ SELECT * FROM check_estimated_rows('
 20 | 17
 (1 row)
 
-SELECT count(*) FROM -- Learn on the query
+SELECT count(*) FROM
 (SELECT fs FROM aqo_data GROUP BY (fs)) AS q1
 ;
 count

expected/clean_aqo_data.out

@@ -11,10 +11,10 @@ SELECT * FROM a;
 (0 rows)
 
 SELECT 'a'::regclass::oid AS a_oid \gset
-SELECT aqo_cleanup();
-aqo_cleanup
--------------
-(0,0)
+SELECT true FROM aqo_cleanup();
+bool
+------
+t
 (1 row)
 
 /*
@@ -54,10 +54,10 @@ SELECT count(*) FROM aqo_query_stat WHERE
 (1 row)
 
 DROP TABLE a;
-SELECT aqo_cleanup();
-aqo_cleanup
--------------
-(1,1)
+SELECT true FROM aqo_cleanup();
+bool
+------
+t
 (1 row)
 
 /*
@@ -175,10 +175,10 @@ SELECT count(*) FROM aqo_query_stat WHERE
 (1 row)
 
 DROP TABLE a;
-SELECT aqo_cleanup();
-aqo_cleanup
--------------
-(2,3)
+SELECT true FROM aqo_cleanup();
+bool
+------
+t
 (1 row)
 
 /*
@@ -253,10 +253,10 @@ SELECT count(*) FROM aqo_query_stat WHERE
 (1 row)
 
 DROP TABLE b;
-SELECT aqo_cleanup();
-aqo_cleanup
--------------
-(1,1)
+SELECT true FROM aqo_cleanup();
+bool
+------
+t
 (1 row)
 
 -- lines corresponding to b_oid in theese tables deleted

expected/gucs.out

@@ -5,10 +5,10 @@ SET aqo.show_details = true;
 CREATE TABLE t(x int);
 INSERT INTO t (x) (SELECT * FROM generate_series(1, 100) AS gs);
 ANALYZE t;
-SELECT * FROM aqo_reset(); -- Remember! DROP EXTENSION doesn't remove any AQO data gathered.
-aqo_reset
------------
-25
+SELECT true FROM aqo_reset(); -- Remember! DROP EXTENSION doesn't remove any AQO data gathered.
+bool
+------
+t
 (1 row)
 
 -- Check AQO addons to explain (the only stable data)
@@ -123,10 +123,10 @@ SELECT count(*) FROM aqo_query_stat;
 1
 (1 row)
 
-SELECT * FROM aqo_reset(); -- Remove one record from all tables
-aqo_reset
------------
-4
+SELECT true FROM aqo_reset(); -- Remove one record from all tables
+bool
+------
+t
 (1 row)
 
 SELECT count(*) FROM aqo_query_stat;

machine_learning.c

@@ -48,10 +48,14 @@ OkNNr_allocate(int ncols)
 int i;
 
 if (ncols > 0)
-for (i = 0; i < aqo_K; ++i)
-data->matrix[i] = palloc0(sizeof(double) * ncols);
+for (i = 0; i < aqo_K; i++)
+data->matrix[i] = palloc0(ncols * sizeof(double));
+else
+for (i = 0; i < aqo_K; i++)
+data->matrix[i] = NULL;
 
 data->cols = ncols;
+data->rows = -1;
 return data;
 }
 

postprocessing.c

@@ -109,9 +109,8 @@ learn_agg_sample(aqo_obj_stat *ctx, RelSortOut *rels,
 uint64 fs = query_context.fspace_hash;
 int child_fss;
 double target;
-OkNNrdata	data;
+OkNNrdata	*data = OkNNr_allocate(0);
 int fss;
-int i;
 
 /*
 * Learn 'not executed' nodes only once, if no one another knowledge exists
@@ -125,13 +124,10 @@ learn_agg_sample(aqo_obj_stat *ctx, RelSortOut *rels,
 NIL, NULL,NULL);
 fss = get_grouped_exprs_hash(child_fss, aqo_node->grouping_exprs);
 
-memset(&data, 0, sizeof(OkNNrdata));
-for (i = 0; i < aqo_K; i++)
-data.matrix[i] = NULL;
-
 /* Critical section */
-atomic_fss_learn_step(fs, fss, &data, NULL,
+atomic_fss_learn_step(fs, fss, data, NULL,
 target, rfactor, rels->hrels, ctx->isTimedOut);
+OkNNr_free(data);
 /* End of critical section */
 }
 

sql/aqo_learn.sql

@@ -235,8 +235,8 @@ SELECT * FROM check_estimated_rows('
 SELECT *
 FROM aqo_test1 AS t1, aqo_test1 AS t2, aqo_test1 AS t3, aqo_test1 AS t4
 WHERE t1.a = t2.b AND t2.a = t3.b AND t3.a = t4.b;
-');
-SELECT count(*) FROM -- Learn on the query
+'); -- Learn on the query
+SELECT count(*) FROM
 (SELECT fs FROM aqo_data GROUP BY (fs)) AS q1
 ;
 SELECT query_text FROM aqo_query_texts WHERE queryid <> 0; -- Check query

sql/clean_aqo_data.sql

@@ -7,7 +7,7 @@ DROP TABLE IF EXISTS b;
 CREATE TABLE a();
 SELECT * FROM a;
 SELECT 'a'::regclass::oid AS a_oid \gset
-SELECT aqo_cleanup();
+SELECT true FROM aqo_cleanup();
 
 /*
 * lines with a_oid in aqo_data,
@@ -27,7 +27,7 @@ SELECT count(*) FROM aqo_query_stat WHERE
 aqo_queries.fs = ANY(SELECT aqo_data.fs FROM aqo_data WHERE :a_oid=ANY(oids)));
 
 DROP TABLE a;
-SELECT aqo_cleanup();
+SELECT true FROM aqo_cleanup();
 
 /*
 * lines with a_oid in aqo_data,
@@ -79,7 +79,7 @@ SELECT count(*) FROM aqo_query_stat WHERE
 aqo_queries.fs = ANY(SELECT aqo_data.fs FROM aqo_data WHERE :b_oid=ANY(oids)));
 
 DROP TABLE a;
-SELECT aqo_cleanup();
+SELECT true FROM aqo_cleanup();
 
 /*
 * lines corresponding to a_oid and both a_oid's fs deleted in aqo_data,
@@ -115,7 +115,7 @@ SELECT count(*) FROM aqo_query_stat WHERE
 aqo_queries.fs = aqo_queries.queryid);
 
 DROP TABLE b;
-SELECT aqo_cleanup();
+SELECT true FROM aqo_cleanup();
 
 -- lines corresponding to b_oid in theese tables deleted
 SELECT count(*) FROM aqo_data WHERE :b_oid=ANY(oids);

sql/gucs.sql

@@ -8,7 +8,7 @@ CREATE TABLE t(x int);
 INSERT INTO t (x) (SELECT * FROM generate_series(1, 100) AS gs);
 ANALYZE t;
 
-SELECT * FROM aqo_reset(); -- Remember! DROP EXTENSION doesn't remove any AQO data gathered.
+SELECT true FROM aqo_reset(); -- Remember! DROP EXTENSION doesn't remove any AQO data gathered.
 -- Check AQO addons to explain (the only stable data)
 EXPLAIN (ANALYZE, VERBOSE, COSTS OFF, TIMING OFF, SUMMARY OFF)
 SELECT x FROM t;
@@ -33,7 +33,7 @@ SELECT obj_description('aqo_reset'::regproc::oid);
 
 -- Check stat reset
 SELECT count(*) FROM aqo_query_stat;
-SELECT * FROM aqo_reset(); -- Remove one record from all tables
+SELECT true FROM aqo_reset(); -- Remove one record from all tables
 SELECT count(*) FROM aqo_query_stat;
 
 DROP EXTENSION aqo;

storage.c

@@ -983,8 +983,10 @@ aqo_query_texts(PG_FUNCTION_ARGS)
 hash_seq_init(&hash_seq, qtexts_htab);
 while ((entry = hash_seq_search(&hash_seq)) != NULL)
 {
+char *ptr;
+
 Assert(DsaPointerIsValid(entry->qtext_dp));
-char *ptr = dsa_get_address(qtext_dsa, entry->qtext_dp);
+ptr = dsa_get_address(qtext_dsa, entry->qtext_dp);
 values[QT_QUERYID] = Int64GetDatum(entry->queryid);
 values[QT_QUERY_STRING] = CStringGetTextDatum(ptr);
 tuplestore_putvalues(tupstore, tupDesc, values, nulls);
@@ -1170,7 +1172,7 @@ build_knn_matrix(OkNNrdata *data, const OkNNrdata *temp_data)
 {
 Assert(data->cols == temp_data->cols);
 
-if (data->rows >= 0)
+if (data->rows > 0)
 /* trivial strategy - use first suitable record and ignore others */
 return;
 
@@ -1201,8 +1203,7 @@ _fill_knn_data(const DataEntry *entry, List **reloids)
 /* Check invariants */
 Assert(entry->rows < aqo_K);
 Assert(ptr != NULL);
-Assert(entry->key.fs == ((data_key *)ptr)->fs &&
-entry->key.fss == ((data_key *)ptr)->fss);
+Assert(entry->key.fss == ((data_key *)ptr)->fss);
 
 ptr += sizeof(data_key);
 
@@ -1227,6 +1228,7 @@ _fill_knn_data(const DataEntry *entry, List **reloids)
 Assert(offset <= sz);
 
 if (reloids == NULL)
+/* Isn't needed to load reloids list */
 return data;
 
 /* store list of relations. XXX: optimize ? */
@@ -1260,26 +1262,72 @@ load_aqo_data(uint64 fs, int fss, OkNNrdata *data, List **reloids,
 dsa_init();
 
 LWLockAcquire(&aqo_state->data_lock, LW_EXCLUSIVE);
-entry = (DataEntry *) hash_search(data_htab, &key, HASH_FIND, &found);
 
-if (!found)
-goto end;
+if (!wideSearch)
+{
+entry = (DataEntry *) hash_search(data_htab, &key, HASH_FIND, &found);
 
-/* One entry with all correctly filled fields is found */
-Assert(entry);
-Assert(DsaPointerIsValid(entry->data_dp));
+if (!found)
+goto end;
+
+/* One entry with all correctly filled fields is found */
+Assert(entry);
+Assert(DsaPointerIsValid(entry->data_dp));
+
+if (entry->cols != data->cols)
+{
+/* Collision happened? */
+elog(LOG, "[AQO] Does a collision happened? Check it if possible (fs: %lu, fss: %d).",
+fs, fss);
+found = false;
+goto end;
+}
 
-if (entry->cols != data->cols)
+temp_data = _fill_knn_data(entry, reloids);
+build_knn_matrix(data, temp_data);
+}
+else
+/* Iterate across all elements of the table. XXX: Maybe slow. */
 {
-/* Collision happened? */
-elog(LOG, "[AQO] Does a collision happened? Check it if possible (fs: %lu, fss: %d).",
-fs, fss);
+HASH_SEQ_STATUS	hash_seq;
+int noids = -1;
+
 found = false;
-goto end;
+hash_seq_init(&hash_seq, data_htab);
+while ((entry = hash_seq_search(&hash_seq)) != NULL)
+{
+List *tmp_oids = NIL;
+
+if (entry->key.fss != fss || entry->cols != data->cols)
+continue;
+
+temp_data = _fill_knn_data(entry, &tmp_oids);
+
+if (data->rows > 0 && list_length(tmp_oids) != noids)
+{
+/* Dubious case. So log it and skip these data */
+elog(LOG,
+"[AQO] different number depended oids for the same fss %d: "
+"%d and %d correspondingly.",
+fss, list_length(tmp_oids), noids);
+Assert(noids >= 0);
+list_free(tmp_oids);
+continue;
+}
+
+noids = list_length(tmp_oids);
+
+if (reloids != NULL && *reloids == NIL)
+*reloids = tmp_oids;
+else
+list_free(tmp_oids);
+
+build_knn_matrix(data, temp_data);
+found = true;
+}
 }
 
-temp_data = _fill_knn_data(entry, reloids);
-build_knn_matrix(data, temp_data);
+Assert(!found || (data->rows > 0 && data->rows <= aqo_K));
 end:
 LWLockRelease(&aqo_state->data_lock);
 
@@ -1364,7 +1412,10 @@ aqo_data(PG_FUNCTION_ARGS)
 
 elems = palloc(sizeof(*elems) * entry->nrels);
 for(i = 0; i < entry->nrels; i++)
+{
 elems[i] = ObjectIdGetDatum(*(Oid *)ptr);
+ptr += sizeof(Oid);
+}
 
 array = construct_array(elems, entry->nrels, OIDOID,
 sizeof(Oid), true, TYPALIGN_INT);