Commit Graph

6123 Commits

Author SHA1 Message Date
c298b1ca1a [fix](timezone) fix parse timezone when include GMT or time zone short ids (#25032) 2023-10-03 20:53:16 +08:00
4c94820ff9 [opt](nereids) adjust column stats in filter estimation (#24973)
TPCDS before
query4  9335    8113    8070    8070
query13 3104    1386    1385    1385
query18 1704    1216    1151    1151
query48 840     840     839     839
query61 435     379     383     379
query71 715     570     579     570
query85 2822    2627    2612    2612
query88 1897    1816    1793    1793
Total cold run time: 20852 ms
Total hot run time: 16799 ms

after:
query4  9610    8287    8249    8249
query13 1721    1013    1042    1013
query18 1585    1186    1155    1155
query48 789     777     778     777
query61 384     387     381     381
query71 713     610     584     584
query85 2020    1867    1843    1843
query88 1859    1812    1805    1805
Total cold run time: 18681 ms
Total hot run time: 15807 ms
2023-09-28 21:34:17 +08:00
8eaf0d3a4b [fix](Nereids) ctas varchar length should set to max except column from slot (#25003) 2023-09-28 21:32:33 +08:00
5dd70b8a25 [fix](planner) createColumnAndViewDefs method use wrong analyzer (#25005) 2023-09-28 06:59:04 -05:00
230b7bd15e [test](nereids) Add some tests for PushFilterInsideJoin and FindHashConditionForJoin rule (#24550) 2023-09-28 03:45:05 -05:00
bf4fb32487 [minor](catalog) remove useless compatibilityMatrix in catalog PrimitiveType (#24999) 2023-09-28 02:57:59 -05:00
a574f29d76 [enhancement](Nereids): use enforcer to choose the n-th plan (#22929) 2023-09-28 15:16:24 +08:00
b50c1448df [fix](Nereids) should not replace slot by Alias when do NormalizeSlot (#24928)
when we do NormalizeToSlot, we pushed complex expression and only remain
slot of it. When we do this, we collect alias and their child and
compute its child in bottom project, remain the result slot in current
node. for example

Window(max(...), c1 as a1)

after normalization, we get

Window(max(...), a1)
+-- Project(..., c1 as a1)

But, in some cases, we remove some SlotReference by mistake, for example

Window(max(...), c1, c1 as a1)

after normalization, we get

Window(max(...), a1)
+-- Project(..., c1 as a1)

we lost the SlotReference c1. This PR fix this problem. After this Pr,
we get

Window(max(...), c1, a1)
+-- Project(..., c1, c1 as a1)
2023-09-28 14:51:08 +08:00
377554ee1c [Fix](Job)Job Task does not display error message (#24897) 2023-09-28 14:47:12 +08:00
e863cfe5c7 [fix](nereids) fix multi window projection issue temporarily (#24912)
Current multi-window plan generation has problem on the project sequence, for example:

+--LogicalWindow ( windowExpressions=[avg(sum_sales#115) WindowSpec(...) AS `avg_monthly_sales`#116, rank() WindowSpec(...) AS `rn`#117], ...)
and correspond physical plan is:

+--PhysicalWindow[6572]@16 ( windowFrameGroup=(Funcs=[avg(sum_sales#115) WindowSpec(...) AS `avg_monthly_sales`#116], ... )
    +--PhysicalWindow[6568]@29 ( windowFrameGroup=(Funcs=[rank() WindowSpec(...) AS `rn`#117], ...] )
If the final plan is generated as following:

MultiCastDataSinks
STREAM DATA SINK
  EXCHANGE ID: 20
  HASH_PARTITIONED: rn[#208], i_brand[#202], cc_name[#203], i_category[#201]
Before we eventually resolve the multi-window issue, we add a projection as following and force a mapping but this will not cover all potential problems.

MultiCastDataSinks
STREAM DATA SINK
  EXCHANGE ID: 20
  HASH_PARTITIONED: rn[#219], i_brand[#213], cc_name[#214], i_category[#212]
  PROJECTIONS: i_category[#184], i_brand[#185], cc_name[#186], d_year[#187], d_moy[#188], sum_sales[#189], avg_monthly_sales[#191], rn[#190]
  PROJECTION TUPLE: 20
2023-09-28 14:33:00 +08:00
f5c38b29a5 [Improve](Load)Change the response label prefix of Update and Delete to the corresponding operations (#24996)
Doris Whether it is insert, delete, or update, the label prefix is insert, which may confuse users.

Change
Add Update and Delete label prefix

Test

mysql>  insert into t2 (id,id_str) values (2,'test2');
Query OK, 1 row affected (0.09 sec)
{'label':'insert_b16405a387f14bfa_947dc9b2217ee3df', 'status':'VISIBLE', 'txnId':'17023'}

mysql>  insert into t1 (id,id_str) values (2,'test2');
Query OK, 1 row affected (0.09 sec)
{'label':'insert_c3acdf63bf94e87_ad65a2dca88f5576', 'status':'VISIBLE', 'txnId':'17025'}

mysql> update t2 set id_str='update2';
Query OK, 2 rows affected (5.27 sec)
{'label':'update_903a88c8defe41d5_a7fca85159c84e50', 'status':'VISIBLE', 'txnId':'17026'}


mysql> delete from  t2  where id =2;
Query OK, 0 rows affected (5.56 sec)
{'label':'delete_1ca419aa-b7a2-41f6-9cbd-e14f4c7517f4', 'status':'VISIBLE', 'txnId':'17028'}

mysql> delete from t1 where t1.id in (select id from t2);
Query OK, 1 row affected (4.41 sec)
{'label':'delete_7e2ae75fee9a42b7_9322d4ae8b80a28b', 'status':'VISIBLE', 'txnId':'17034'}
2023-09-28 14:32:35 +08:00
bf808e9aa6 [fix](Nereids): tolerate DateLike overflow in SQL CAST/CONVERT (#24943)
- explicit type cast, we need tolerate overflow and convert it to be NULL
- implicit type cast, throw exception
2023-09-28 12:11:50 +08:00
188d9ab94e [enhancement](statistics) collect table level loaded rows on BE to make RPC light weight (#24609) 2023-09-28 10:51:50 +08:00
42207df89f [refactor](nereids)update NormalizeRepeat comments (#24893)
update NormalizeRepeat comments
2023-09-28 10:42:16 +08:00
584646c054 [improvement](nereids)dphyper GraphSimplifier should consider missed edges when estimating join cost (#21747) 2023-09-28 09:30:57 +08:00
732f821c15 [Fix](inverted index) make parser mode coarse grained by default (#24949) 2023-09-27 21:04:41 +08:00
d4e823950a [bug](json)Fix some problems of json function on Nereids (#24898)
Fix some problems of json_length and json_contains function on Nereids
fix wrong result of json_contains function
Regression test jsonb_p0 to enable Nereids
2023-09-27 21:01:45 +08:00
391a4e29eb [fix](schema) Table column order is changed if add a column and do truncate (#24981) 2023-09-27 20:59:11 +08:00
63b283a848 [fix](Nereids) init Date/DateV2Literal should check non-zero time fields (#24971) 2023-09-27 20:48:36 +08:00
1fb9022d07 [pipelineX](bug) Fix meta scan operator (#24963) 2023-09-27 20:34:47 +08:00
bb7f8d18a8 [fix](nereids) push down filter through partition topn (#24944)
support pushing down filter through partition topn if the filter can pass through window.
fix CreatePartitionTopNFromWindow bug which may generate two partition topn unexpectly.
case:
select * from (select c2, row_number() over (partition by c2) as rn from t1) T where rn<=1 and c2 = 1;
before this pr:
| PhysicalResultSink                       |
| --PhysicalDistribute                     |
| ----filter((rn <= 1))                    |
| ------PhysicalWindow                     |
| --------PhysicalQuickSort                |
| ----------PhysicalDistribute             |
| ------------PhysicalPartitionTopN        |
| --------------filter((T.c2 = 1))         |
| ----------------PhysicalPartitionTopN    |
| ------------------PhysicalProject        |
| --------------------PhysicalOlapScan[t1] |
+------------------------------------------+
after:

| PhysicalResultSink                     |
| --PhysicalDistribute                   |
| ----filter((rn <= 1))                  |
| ------PhysicalWindow                   |
| --------PhysicalQuickSort              |
| ----------PhysicalDistribute           |
| ------------PhysicalPartitionTopN      |
| --------------PhysicalProject          |
| ----------------filter((T.c2 = 1))     |
| ------------------PhysicalOlapScan[t1] |
+----------------------------------------+
2023-09-27 19:38:04 +08:00
00786a3295 [fix](Nereids) could not prune datev1 partition column (#24959)
because storage engine could not process date comparison predicates.
we convert it to datetime comparison predicates.
however, partition prunner could not process cast(slot) cp literal.
so, we convert back in partition pruner to let it work well.

TODO:
move convert date to datetime in translate stage
and only convert predicates for storage engine.
2023-09-27 18:41:56 +08:00
5d138b6928 [remove](function) make execute_impl const and remove running_difference function (#24935) 2023-09-27 18:17:28 +08:00
100d76510c [Fix](HttpServer) Refactor API Endpoints to Only Allow GET Requests for Enhanced Security (#24855) 2023-09-27 17:10:11 +08:00
00e8d1c3b4 [Fix](Planner) disable bitmap type in compare expression (#24792)
Problem:
be core because of bitmap calculation.

Reason:
when be check failed, it would core directly.

Example:
SELECT id_bitmap FROM test_bitmap WHERE id_bitmap IN (NULL) LIMIT 20;

Solved:
Forbidden this kind of expression in fe when analyze. And also forbid bitmap type comparing in other unsupported expressions.
2023-09-27 16:57:06 +08:00
0227292c85 [bug](profile) query profile api of fe cann't get result if non-root user query on the other fe #24858 (#24914)
Issue Number: #24858

If isAllNode is true, the api should only distribute the query to all fe and do not run checkAuthByUserAndQueryId.
If isAllNode is false, the api queries profile on the fe, at this time the api should run checkAuthByUserAndQueryId.
2023-09-27 16:50:41 +08:00
9562e280af [enhancement](Nereids): remove stats derivation in CostAndEnforce job (#24945)
1. remove stats derivation in CostAndEnforce job
2. enforce valid for each stats after estimating
2023-09-27 16:31:03 +08:00
87a30dc41d [feature-wip](arrow-flight)(step3) Support authentication and user session (#24772) 2023-09-27 14:53:58 +08:00
26818de9c8 [feature](jni) support complex types in jni framework (#24810)
Support complex types in jni framework, and successfully run end-to-end on hudi.
### How to Use
Other scanners only need to implement three interfaces in `ColumnValue`:
```
// Get array elements and append into values
void unpackArray(List<ColumnValue> values);

// Get map key array&value array, and append into keys&values
void unpackMap(List<ColumnValue> keys, List<ColumnValue> values);

// Get the struct fields specified by `structFieldIndex`, and append into values
void unpackStruct(List<Integer> structFieldIndex, List<ColumnValue> values);
```
Developers can take `HudiColumnValue` as an example.
2023-09-27 14:47:41 +08:00
a1ab8f96a1 [fix](nereids) mark two phase partition topn global to notice be passthrough logic (#24886)
mark partition topn phase to notice be to handle passthrough logic well, this pr is fe part code.
be side logic: the the phase equals to PTopNPhase.TWO_PAHSE_GLOBAL, it should skip the bypass logic and do the second phase ptopn operation anyway.
2023-09-27 14:08:59 +08:00
1b0e3246ea [pipelineX](fix) Fix exception reporting and Nereids plan (#24936) 2023-09-27 13:15:40 +08:00
452318a9fc [Enhancement](streamload) stream tvf support user specified label (#24219)
stream tvf support user specified label
example:

curl -v --location-trusted -u root: -H "sql: insert into test.t1 WITH LABEL label1 select c1,c2 from http_stream(\"format\" = \"CSV\", \"column_separator\" = \",\")" -T example.csv http://127.0.0.1:8030/api/_http_stream
return:

{
    "TxnId": 2064,
    "Label": "label1",
    "Comment": "",
    "TwoPhaseCommit": "false",
    "Status": "Success",
    "Message": "OK",
    "NumberTotalRows": 2,
    "NumberLoadedRows": 2,
    "NumberFilteredRows": 0,
    "NumberUnselectedRows": 0,
    "LoadBytes": 27,
    "LoadTimeMs": 152,
    "BeginTxnTimeMs": 0,
    "StreamLoadPutTimeMs": 83,
    "ReadDataTimeMs": 92,
    "WriteDataTimeMs": 41,
    "CommitAndPublishTimeMs": 24
}
2023-09-27 12:09:35 +08:00
Pxl
18b5f70a7c [Bug](materialized-view) enable rewrite on select materialized index with aggregate mode (#24691)
enable rewrite on select materialized index with aggregate mode
2023-09-27 11:30:36 +08:00
a8f312794e [feature](nereids)support stats estimation for is-null predicate (#24764)
1. condition order: filter/hashCondition/otherCondition,
2. update regression out
3. remove tpch_sf500 shape case(covered by tpch sf1000)
4. implement is-null stats estimation
5. update ssb shape
2023-09-27 10:04:35 +08:00
b9928a017b [fix](fe)add variant type in decimalv3's compatibilityMatrix table (#24749) 2023-09-27 10:03:13 +08:00
0dd57b1982 [feature](nereids)support subquey in join condition (#24598) 2023-09-27 10:02:56 +08:00
6d27a016b9 [Improvement](regression-test) add http_stream case (#24930) 2023-09-27 09:55:52 +08:00
2e8f9a6429 [improvement](streamload) QeProcessorImpl#registerQuery should be removed (#24836) 2023-09-27 09:48:27 +08:00
f7f359d36c [fix](analysis)fix use regex determine whether time part exists may cause backtracking (#24882) 2023-09-27 09:26:42 +08:00
f59e69b852 [feature](autobucekt) Add support autobucket min buckets config Add support autobucket (#24920)
Signed-off-by: Jack Drogon <jack.xsuperman@gmail.com>
2023-09-27 08:13:53 +08:00
504ab2d726 [fix](es-catalog) If the returned data is incorrect, it will be directly unified as the index was not found (#24831) 2023-09-26 22:59:38 +08:00
841e6188bf [fix](Nereids): fix roundFloor (#24901)
"roundFloor()" does not consider the issue of carry.
2023-09-26 22:37:42 +08:00
90c5461ad2 [fix](Nereids) let dml work well (#24748)
Co-authored-by: sohardforaname <organic_chemistry@foxmail.com>

TODO:
1. support agg_state type
2. support implicit cast literal exception
3. use nereids execute dml for these regression cases:

- test_agg_state_nereids (for TODO 1)
- test_array_insert_overflow (for TODO 2)
- nereids_p0/json_p0/test_json_load_and_function (for TODO 2)
- nereids_p0/json_p0/test_json_unique_load_and_function (for TODO 2)
- nereids_p0/jsonb_p0/test_jsonb_load_and_function (for TODO 2)
- nereids_p0/jsonb_p0/test_jsonb_unique_load_and_function (for TODO 2)
- json_p0/test_json_load_and_function (for TODO 2)
- json_p0/test_json_unique_load_and_function (for TODO 2)
- jsonb_p0/test_jsonb_load_and_function (for TODO 2)
- jsonb_p0/test_jsonb_unique_load_and_function (for TODO 2)
- test_multi_partition_key (for TODO 2)
2023-09-26 21:08:24 +08:00
a6a0e78f32 [Enhancement](streamload) stream tvf support compress (#24303) 2023-09-26 20:58:20 +08:00
55d1090137 [feature](insert) Support group commit stream load (#24304) 2023-09-26 20:57:02 +08:00
e9eaf49db5 [feature](alter colocate group) Support alter colocate group replica allocation (#23320) 2023-09-26 19:55:25 +08:00
77e864df12 [enhancement](delete) use column id in delete push task instead of column name (#24549) 2023-09-26 19:54:55 +08:00
04bf9bce54 [fix](planner)update explode slot's nullable info in analyze phase (#24879) 2023-09-26 18:14:04 +08:00
082bcd820b [feature](insert) Support wal for group commit insert (#23053) 2023-09-26 14:46:24 +08:00
983f948ec4 [fix](Nereids): fix SimplifyComparisonPredicate (#24899)
#16061 exist BUG, it causes wrong cast
2023-09-26 14:28:03 +08:00