Optimize insert path in the DB by avoiding ON CONFLICT

Turns out there is a horrible CPU perf penalty on the DB for ON CONFLICT DO NOTHING.
yet in our data, conflicts are rare. So we first try inserting without ON CONFLICT
and fall back if there is a unique constraint violation.

Unlike retriying with COPY this does not require sending the data again
and thus saves latency and bandwidth.

Author: cevian

CHANGELOG.md

@@ -23,6 +23,7 @@ We use the following categories for changes:
 ### Fixed
 - Fix broken `promscale_packager` telemetry field for docker envs [#1077]
 - Fix compression of old chunks thus reducing storage requirements [#1081]
+- Improved INSERT performance by avoidng ON CONFLICT [#1090]
 
 ## [0.8.0] - 2022-01-18
 

pkg/migrations/migration_files_generated.go

@@ -3067,13 +3067,24 @@ $$
 DECLARE
   num_rows BIGINT;
 BEGIN
+    --turns out there is a horrible CPU perf penalty on the DB for ON CONFLICT DO NOTHING.
+    --yet in our data, conflicts are rare. So we first try inserting without ON CONFLICT
+    --and fall back if there is a unique constraint violation.
     EXECUTE FORMAT(
      'INSERT INTO  prom_data.%1$I (time, value, series_id)
-          SELECT * FROM unnest($1, $2, $3) a(t,v,s) ORDER BY s,t ON CONFLICT DO NOTHING',
+          SELECT * FROM unnest($1, $2, $3) a(t,v,s) ORDER BY s,t',
         metric_table
     ) USING time_array, value_array, series_id_array;
     GET DIAGNOSTICS num_rows = ROW_COUNT;
     RETURN num_rows;
+EXCEPTION WHEN unique_violation THEN 
+	EXECUTE FORMAT(
+	'INSERT INTO  prom_data.%1$I (time, value, series_id)
+		 SELECT * FROM unnest($1, $2, $3) a(t,v,s) ORDER BY s,t ON CONFLICT DO NOTHING',
+	   metric_table
+	) USING time_array, value_array, series_id_array;
+	GET DIAGNOSTICS num_rows = ROW_COUNT;
+	RETURN num_rows;
 END;
 $$
 LANGUAGE PLPGSQL;