perf(Persons): Use optimize_aggregation_in_order for person queries (#…

…17817) * perf(Persons): Use optimize_aggregation_in_order for person queries * Update query snapshots * Update query snapshots * Update query snapshots * Update query snapshots * Update query snapshots * Update query snapshots * Update query snapshots * Update query snapshots --------- Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
PostHog · Oct 5, 2023 · 476d102 · 476d102
1 parent 0508730
commit 476d102
Show file tree

Hide file tree

Showing 22 changed files with 184 additions and 169 deletions.
diff --git a/ee/clickhouse/models/test/__snapshots__/test_cohort.ambr b/ee/clickhouse/models/test/__snapshots__/test_cohort.ambr
@@ -20,7 +20,7 @@
      GROUP BY id
      HAVING max(is_deleted) = 0
      AND ((((has(['something'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$some_prop'), '^"|"$', ''))))
-           AND ((has(['something'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$another_prop'), '^"|"$', ''))))))) as person
+           AND ((has(['something'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$another_prop'), '^"|"$', '')))))) SETTINGS optimize_aggregation_in_order = 1) as person
   UNION ALL
   SELECT person_id,
          cohort_id,
@@ -54,7 +54,7 @@
             AND ((has(['something1'], replaceRegexpAll(JSONExtractRaw(properties, '$some_prop'), '^"|"$', '')))) )
      GROUP BY id
      HAVING max(is_deleted) = 0
-     AND ((has(['something1'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$some_prop'), '^"|"$', ''))))) as person
+     AND ((has(['something1'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$some_prop'), '^"|"$', '')))) SETTINGS optimize_aggregation_in_order = 1) as person
   UNION ALL
   SELECT person_id,
          cohort_id,
@@ -111,7 +111,7 @@
                   AND (((((NOT has(['something1'], replaceRegexpAll(JSONExtractRaw(properties, '$some_prop'), '^"|"$', ''))))))) )
            GROUP BY id
            HAVING max(is_deleted) = 0
-           AND (((((NOT has(['something1'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$some_prop'), '^"|"$', ''))))))))) person ON person.person_id = behavior_query.person_id
+           AND (((((NOT has(['something1'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$some_prop'), '^"|"$', ''))))))) SETTINGS optimize_aggregation_in_order = 1)) person ON person.person_id = behavior_query.person_id
      WHERE 1 = 1
        AND ((((performed_event_condition_15_level_level_0_level_0_level_0_0)))) ) as person
   UNION ALL
@@ -176,7 +176,7 @@
                               AND (((((NOT has(['something1'], replaceRegexpAll(JSONExtractRaw(properties, '$some_prop'), '^"|"$', ''))))))) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (((((NOT has(['something1'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$some_prop'), '^"|"$', ''))))))))) person ON person.person_id = behavior_query.person_id
+                       AND (((((NOT has(['something1'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$some_prop'), '^"|"$', ''))))))) SETTINGS optimize_aggregation_in_order = 1)) person ON person.person_id = behavior_query.person_id
                  WHERE 1 = 1
                    AND ((((performed_event_condition_17_level_level_0_level_0_level_0_0)))) ) ))
   '

diff --git a/ee/clickhouse/queries/funnels/test/__snapshots__/test_funnel_correlation.ambr b/ee/clickhouse/queries/funnels/test/__snapshots__/test_funnel_correlation.ambr
@@ -170,7 +170,7 @@
         FROM person
         WHERE team_id = 2
         GROUP BY id
-        HAVING max(is_deleted) = 0) person ON person.id = funnel_actors.actor_id) aggregation_target_with_props
+        HAVING max(is_deleted) = 0 SETTINGS optimize_aggregation_in_order = 1) person ON person.id = funnel_actors.actor_id) aggregation_target_with_props
   GROUP BY prop.1,
                prop.2
   HAVING prop.1 NOT IN []
@@ -242,7 +242,7 @@
                               AND (has(['Positive'], replaceRegexpAll(JSONExtractRaw(properties, '$browser'), '^"|"$', ''))) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['Positive'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', '')))) person ON person.id = pdi.person_id
+                       AND (has(['Positive'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', ''))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND event IN ['paid', 'user signed up']
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2020-01-01 00:00:00', 'UTC')
@@ -326,7 +326,7 @@
                               AND (has(['Positive'], replaceRegexpAll(JSONExtractRaw(properties, '$browser'), '^"|"$', ''))) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['Positive'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', '')))) person ON person.id = pdi.person_id
+                       AND (has(['Positive'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', ''))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND event IN ['paid', 'user signed up']
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2020-01-01 00:00:00', 'UTC')
@@ -410,7 +410,7 @@
                               AND (has(['Negative'], replaceRegexpAll(JSONExtractRaw(properties, '$browser'), '^"|"$', ''))) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['Negative'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', '')))) person ON person.id = pdi.person_id
+                       AND (has(['Negative'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', ''))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND event IN ['paid', 'user signed up']
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2020-01-01 00:00:00', 'UTC')
@@ -494,7 +494,7 @@
                               AND (has(['Negative'], replaceRegexpAll(JSONExtractRaw(properties, '$browser'), '^"|"$', ''))) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['Negative'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', '')))) person ON person.id = pdi.person_id
+                       AND (has(['Negative'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', ''))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND event IN ['paid', 'user signed up']
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2020-01-01 00:00:00', 'UTC')
@@ -594,7 +594,7 @@
         FROM person
         WHERE team_id = 2
         GROUP BY id
-        HAVING max(is_deleted) = 0) person ON person.id = funnel_actors.actor_id) aggregation_target_with_props
+        HAVING max(is_deleted) = 0 SETTINGS optimize_aggregation_in_order = 1) person ON person.id = funnel_actors.actor_id) aggregation_target_with_props
   GROUP BY prop.1,
                prop.2
   HAVING prop.1 NOT IN []
@@ -666,7 +666,7 @@
                               AND (has(['Positive'], "pmat_$browser")) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['Positive'], argMax(person."pmat_$browser", version)))) person ON person.id = pdi.person_id
+                       AND (has(['Positive'], argMax(person."pmat_$browser", version))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND event IN ['paid', 'user signed up']
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2020-01-01 00:00:00', 'UTC')
@@ -750,7 +750,7 @@
                               AND (has(['Positive'], "pmat_$browser")) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['Positive'], argMax(person."pmat_$browser", version)))) person ON person.id = pdi.person_id
+                       AND (has(['Positive'], argMax(person."pmat_$browser", version))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND event IN ['paid', 'user signed up']
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2020-01-01 00:00:00', 'UTC')
@@ -834,7 +834,7 @@
                               AND (has(['Negative'], "pmat_$browser")) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['Negative'], argMax(person."pmat_$browser", version)))) person ON person.id = pdi.person_id
+                       AND (has(['Negative'], argMax(person."pmat_$browser", version))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND event IN ['paid', 'user signed up']
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2020-01-01 00:00:00', 'UTC')
@@ -918,7 +918,7 @@
                               AND (has(['Negative'], "pmat_$browser")) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['Negative'], argMax(person."pmat_$browser", version)))) person ON person.id = pdi.person_id
+                       AND (has(['Negative'], argMax(person."pmat_$browser", version))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND event IN ['paid', 'user signed up']
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2020-01-01 00:00:00', 'UTC')

diff --git a/ee/clickhouse/queries/funnels/test/__snapshots__/test_funnel_correlations_persons.ambr b/ee/clickhouse/queries/funnels/test/__snapshots__/test_funnel_correlations_persons.ambr
@@ -445,7 +445,7 @@
                               AND (has(['bar'], replaceRegexpAll(JSONExtractRaw(properties, 'foo'), '^"|"$', ''))) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['bar'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), 'foo'), '^"|"$', '')))) person ON person.id = pdi.person_id
+                       AND (has(['bar'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), 'foo'), '^"|"$', ''))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND event IN ['$pageview', 'insight analyzed']
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2021-01-01 00:00:00', 'UTC')
@@ -579,7 +579,7 @@
                               AND (has(['bar'], replaceRegexpAll(JSONExtractRaw(properties, 'foo'), '^"|"$', ''))) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['bar'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), 'foo'), '^"|"$', '')))) person ON person.id = pdi.person_id
+                       AND (has(['bar'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), 'foo'), '^"|"$', ''))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2021-01-01 00:00:00', 'UTC')
                       AND toTimeZone(timestamp, 'UTC') <= toDateTime('2021-01-08 23:59:59', 'UTC')
@@ -711,7 +711,7 @@
                               AND (has(['bar'], replaceRegexpAll(JSONExtractRaw(properties, 'foo'), '^"|"$', ''))) )
                        GROUP BY id
                        HAVING max(is_deleted) = 0
-                       AND (has(['bar'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), 'foo'), '^"|"$', '')))) person ON person.id = pdi.person_id
+                       AND (has(['bar'], replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), 'foo'), '^"|"$', ''))) SETTINGS optimize_aggregation_in_order = 1) person ON person.id = pdi.person_id
                     WHERE team_id = 2
                       AND toTimeZone(timestamp, 'UTC') >= toDateTime('2021-01-01 00:00:00', 'UTC')
                       AND toTimeZone(timestamp, 'UTC') <= toDateTime('2021-01-08 23:59:59', 'UTC')

diff --git a/ee/clickhouse/queries/test/__snapshots__/test_breakdown_props.ambr b/ee/clickhouse/queries/test/__snapshots__/test_breakdown_props.ambr
@@ -73,7 +73,7 @@
         FROM person
         WHERE team_id = 2
         GROUP BY id
-        HAVING max(is_deleted) = 0) person ON pdi.person_id = person.id
+        HAVING max(is_deleted) = 0 SETTINGS optimize_aggregation_in_order = 1) person ON pdi.person_id = person.id
      WHERE team_id = 2
        AND event = '$pageview'
        AND toTimeZone(timestamp, 'UTC') >= toDateTime('2019-12-21 00:00:00', 'UTC')
@@ -105,7 +105,7 @@
         FROM person
         WHERE team_id = 2
         GROUP BY id
-        HAVING max(is_deleted) = 0) person ON pdi.person_id = person.id
+        HAVING max(is_deleted) = 0 SETTINGS optimize_aggregation_in_order = 1) person ON pdi.person_id = person.id
      WHERE team_id = 2
        AND event = '$pageview'
        AND toTimeZone(timestamp, 'UTC') >= toDateTime('2019-12-21 00:00:00', 'UTC')
@@ -143,7 +143,7 @@
                AND ((replaceRegexpAll(JSONExtractRaw(properties, '$browser'), '^"|"$', '') ILIKE '%test%')) )
         GROUP BY id
         HAVING max(is_deleted) = 0
-        AND ((replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', '') ILIKE '%test%'))) person ON pdi.person_id = person.id
+        AND ((replaceRegexpAll(JSONExtractRaw(argMax(person.properties, version), '$browser'), '^"|"$', '') ILIKE '%test%')) SETTINGS optimize_aggregation_in_order = 1) person ON pdi.person_id = person.id
      WHERE team_id = 2
        AND event = '$pageview'
        AND toTimeZone(timestamp, 'UTC') >= toDateTime('2019-12-21 00:00:00', 'UTC')
@@ -204,7 +204,7 @@
         FROM person
         WHERE team_id = 2
         GROUP BY id
-        HAVING max(is_deleted) = 0) person ON pdi.person_id = person.id
+        HAVING max(is_deleted) = 0 SETTINGS optimize_aggregation_in_order = 1) person ON pdi.person_id = person.id
      INNER JOIN
        (SELECT "$session_id",
                dateDiff('second', min(timestamp), max(timestamp)) as session_duration
@@ -245,7 +245,7 @@
         FROM person
         WHERE team_id = 2
         GROUP BY id
-        HAVING max(is_deleted) = 0) person ON pdi.person_id = person.id
+        HAVING max(is_deleted) = 0 SETTINGS optimize_aggregation_in_order = 1) person ON pdi.person_id = person.id
      INNER JOIN
        (SELECT "$session_id",
                dateDiff('second', min(timestamp), max(timestamp)) as session_duration