rust/greptimedb
1
0
Fork 0
mirror of https://github.com/GreptimeTeam/greptimedb.git synced 2025-12-22 22:20:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: promql histogram with aggregation (#7393)

Browse Source 
* fix: promql histogram with aggregation

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>

* update test constructors

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>

* sqlness tests

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>

* update sqlness result

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>

* redact partition number

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>

---------

Signed-off-by: Ruihang Xia <waynestxia@gmail.com>
This commit is contained in:
Ruihang Xia
2025-12-12 15:32:04 +08:00
committed by GitHub
parent baffed8c6a
commit cbfdeca64c
5 changed files with 250 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

91
tests/cases/standalone/common/promql/histogram_multi_partition.result Normal file
View File

@@ -0,0 +1,91 @@
-- Minimal repro for histogram quantile over multi-partition input.
create table histogram_gap_bucket (
ts timestamp time index,
le string,
shard string,
val double,
primary key (shard, le)
) partition on columns (shard) (
shard < 'n',
shard >= 'n'
);
Affected Rows: 0
insert into histogram_gap_bucket values
(0, '0.5', 'a', 1),
(0, '1', 'a', 2),
(0, '+Inf', 'a', 2),
(0, '0.5', 'z', 2),
(0, '1', 'z', 4),
(0, '+Inf', 'z', 4),
(10000, '0.5', 'a', 1),
(10000, '1', 'a', 2),
(10000, '+Inf', 'a', 2),
(10000, '0.5', 'z', 1),
(10000, '1', 'z', 3),
(10000, '+Inf', 'z', 3);
Affected Rows: 12
-- Ensure the physical plan keeps the required repartition/order before folding buckets.
-- SQLNESS REPLACE (metrics.*) REDACTED
-- SQLNESS REPLACE (peers.*) REDACTED
-- SQLNESS REPLACE region=\d+\(\d+,\s+\d+\) region=REDACTED
-- SQLNESS REPLACE (-+) -
-- SQLNESS REPLACE (\s\s+) _
-- SQLNESS REPLACE Hash\(\[ts@1\],.* Hash([ts@1],REDACTED
-- SQLNESS REPLACE Hash\(\[le@0,\sts@1\],.* Hash([le@0, ts@1],REDACTED
tql analyze (0, 10, '10s') histogram_quantile(0.5, sum by (le) (histogram_gap_bucket));
+-+-+-+
| stage | node | plan_|
+-+-+-+
| 0_| 0_|_HistogramFoldExec: le=@0, field=@2, quantile=0.5 REDACTED
|_|_|_SortExec: expr=[ts@1 ASC NULLS LAST, CAST(le@0 AS Float64) ASC NULLS LAST], preserve_partitioning=[true] REDACTED
|_|_|_CoalesceBatchesExec: target_batch_size=8192 REDACTED
|_|_|_RepartitionExec: partitioning=Hash([ts@1],REDACTED
|_|_|_AggregateExec: mode=FinalPartitioned, gby=[le@0 as le, ts@1 as ts], aggr=[sum(histogram_gap_bucket.val)] REDACTED
|_|_|_CoalesceBatchesExec: target_batch_size=8192 REDACTED
|_|_|_RepartitionExec: partitioning=Hash([le@0, ts@1],REDACTED
|_|_|_AggregateExec: mode=Partial, gby=[le@0 as le, ts@1 as ts], aggr=[sum(histogram_gap_bucket.val)] REDACTED
|_|_|_CooperativeExec REDACTED
|_|_|_MergeScanExec: REDACTED
|_|_|_|
| 1_| 0_|_AggregateExec: mode=FinalPartitioned, gby=[le@0 as le, ts@1 as ts], aggr=[__sum_state(histogram_gap_bucket.val)] REDACTED
|_|_|_CoalesceBatchesExec: target_batch_size=8192 REDACTED
|_|_|_RepartitionExec: partitioning=Hash([le@0, ts@1],REDACTED
|_|_|_AggregateExec: mode=Partial, gby=[le@1 as le, ts@0 as ts], aggr=[__sum_state(histogram_gap_bucket.val)] REDACTED
|_|_|_ProjectionExec: expr=[ts@0 as ts, le@1 as le, val@3 as val] REDACTED
|_|_|_PromInstantManipulateExec: range=[0..10000], lookback=[300000], interval=[10000], time index=[ts] REDACTED
|_|_|_PromSeriesDivideExec: tags=["shard", "le"] REDACTED
|_|_|_CooperativeExec REDACTED
|_|_|_SeriesScan: region=REDACTED, "partition_count":{"count":1, "mem_ranges":1, "files":0, "file_ranges":0}, "distribution":"PerSeries" REDACTED
|_|_|_|
| 1_| 1_|_AggregateExec: mode=FinalPartitioned, gby=[le@0 as le, ts@1 as ts], aggr=[__sum_state(histogram_gap_bucket.val)] REDACTED
|_|_|_CoalesceBatchesExec: target_batch_size=8192 REDACTED
|_|_|_RepartitionExec: partitioning=Hash([le@0, ts@1],REDACTED
|_|_|_AggregateExec: mode=Partial, gby=[le@1 as le, ts@0 as ts], aggr=[__sum_state(histogram_gap_bucket.val)] REDACTED
|_|_|_ProjectionExec: expr=[ts@0 as ts, le@1 as le, val@3 as val] REDACTED
|_|_|_PromInstantManipulateExec: range=[0..10000], lookback=[300000], interval=[10000], time index=[ts] REDACTED
|_|_|_PromSeriesDivideExec: tags=["shard", "le"] REDACTED
|_|_|_CooperativeExec REDACTED
|_|_|_SeriesScan: region=REDACTED, "partition_count":{"count":1, "mem_ranges":1, "files":0, "file_ranges":0}, "distribution":"PerSeries" REDACTED
|_|_|_|
|_|_| Total rows: 2_|
+-+-+-+
-- SQLNESS SORT_RESULT 2 1
tql eval (0, 10, '10s') histogram_quantile(0.5, sum by (le) (histogram_gap_bucket));
+---------------------+-------------------------------+
| ts | sum(histogram_gap_bucket.val) |
+---------------------+-------------------------------+
| 1970-01-01T00:00:00 | 0.5 |
| 1970-01-01T00:00:10 | 0.5833333333333334 |
+---------------------+-------------------------------+
drop table histogram_gap_bucket;
Affected Rows: 0

40
tests/cases/standalone/common/promql/histogram_multi_partition.sql Normal file
View File

@@ -0,0 +1,40 @@
-- Minimal repro for histogram quantile over multi-partition input.
create table histogram_gap_bucket (
ts timestamp time index,
le string,
shard string,
val double,
primary key (shard, le)
) partition on columns (shard) (
shard < 'n',
shard >= 'n'
);
insert into histogram_gap_bucket values
(0, '0.5', 'a', 1),
(0, '1', 'a', 2),
(0, '+Inf', 'a', 2),
(0, '0.5', 'z', 2),
(0, '1', 'z', 4),
(0, '+Inf', 'z', 4),
(10000, '0.5', 'a', 1),
(10000, '1', 'a', 2),
(10000, '+Inf', 'a', 2),
(10000, '0.5', 'z', 1),
(10000, '1', 'z', 3),
(10000, '+Inf', 'z', 3);
-- Ensure the physical plan keeps the required repartition/order before folding buckets.
-- SQLNESS REPLACE (metrics.*) REDACTED
-- SQLNESS REPLACE (peers.*) REDACTED
-- SQLNESS REPLACE region=\d+\(\d+,\s+\d+\) region=REDACTED
-- SQLNESS REPLACE (-+) -
-- SQLNESS REPLACE (\s\s+) _
-- SQLNESS REPLACE Hash\(\[ts@1\],.* Hash([ts@1],REDACTED
-- SQLNESS REPLACE Hash\(\[le@0,\sts@1\],.* Hash([le@0, ts@1],REDACTED
tql analyze (0, 10, '10s') histogram_quantile(0.5, sum by (le) (histogram_gap_bucket));
-- SQLNESS SORT_RESULT 2 1
tql eval (0, 10, '10s') histogram_quantile(0.5, sum by (le) (histogram_gap_bucket));
drop table histogram_gap_bucket;

3
tests/cases/standalone/common/promql/simple_histogram.result
View File

@@ -319,6 +319,7 @@ insert into histogram4_bucket values
Affected Rows: 7
-- SQLNESS SORT_RESULT 3 1
tql eval (2900, 3000, '100s') histogram_quantile(0.9, histogram4_bucket);
+---------------------+---+-----+
@@ -332,6 +333,7 @@ drop table histogram4_bucket;
Affected Rows: 0
-- SQLNESS SORT_RESULT 3 1
tql eval(0, 10, '10s') histogram_quantile(0.99, sum by(pod,instance, fff) (rate(greptime_servers_postgres_query_elapsed_bucket{instance=~"xxx"}[1m])));
++
@@ -395,6 +397,7 @@ insert into histogram5_bucket values
Affected Rows: 12
-- SQLNESS SORT_RESULT 3 1
tql eval (3000, 3015, '3s') histogram_quantile(0.5, histogram5_bucket);
+---------------------+---+--------------------+

4
tests/cases/standalone/common/promql/simple_histogram.sql
View File

@@ -184,10 +184,12 @@ insert into histogram4_bucket values
-- INF here is missing
;
-- SQLNESS SORT_RESULT 3 1
tql eval (2900, 3000, '100s') histogram_quantile(0.9, histogram4_bucket);
drop table histogram4_bucket;
-- SQLNESS SORT_RESULT 3 1
tql eval(0, 10, '10s') histogram_quantile(0.99, sum by(pod,instance, fff) (rate(greptime_servers_postgres_query_elapsed_bucket{instance=~"xxx"}[1m])));
-- test case where table exists but doesn't have 'le' column should raise error
@@ -233,7 +235,7 @@ insert into histogram5_bucket values
(3015000, "5", "a", 30),
(3015000, "+Inf", "a", 50);
-- SQLNESS SORT_RESULT 3 1
tql eval (3000, 3015, '3s') histogram_quantile(0.5, histogram5_bucket);
drop table histogram5_bucket;