[Doris On ES] Skip function_call expr when process predicate
Fixed#3801
Do not push-down function_call such as split_xxx when process predicate, Doris BE is responsible for processing these predicate
All rows in table:
```
+------+------+------+------------+------------+
| k1 | k2 | k3 | UpdateTime | ArriveTime |
+------+------+------+------------+------------+
| NULL | NULL | kkk1 | 123456789 | NULL |
| kkk1 | NULL | NULL | 123456789 | NULL |
| NULL | kkk2 | NULL | 123456789 | NULL |
+------+------+------+------------+------------+
```
The following predicate could not push down to ES.
```
SQL 1:
mysql> select * from (select split_part(k1, "1", 1) as kk from case_replay_for_milimin) t where t.kk is not null;
+------+
| kk |
+------+
| kkk |
+------+
1 row in set (0.02 sec)
SQL 2:
mysql> select * from (select split_part(k1, "1", 1) as kk from case_replay_for_milimin) t where t.kk > 'a';
+------+
| kk |
+------+
| kkk |
+------+
SQL 3:
mysql> select * from (select split_part(k1, "1", 1) as kk from case_replay_for_milimin) t where t.kk > '2';
+------+
| kk |
+------+
| kkk |
+------+
1 row in set (0.03 sec)
```
The other PR : https://github.com/apache/incubator-doris/pull/3513 (https://github.com/apache/incubator-doris/issues/3479) try to resolved the `inner hits node is not an array` because when a query( batch-size) run against new segment without this field, as-well the filter_path just only take `hits.hits.fields` 、`hits.hits._source` into account, this would appear an null inner hits node:
```
{
"_scroll_id": "DXF1ZXJ5QW5kRmV0Y2gBAAAAAAAAHaUWY1ExUVd0ZWlRY2",
"hits": {
"total": 1
}
}
```
Unfortunately this PR introduce another serious inconsistent result with different batch_size because of misusing the `total`.
To avoid this two problem, we just add `hits.hits._score` to filter_path when `docvalue_mode` is true, `_score` would always `null` , and populate the inner hits node:
```
{
"_scroll_id": "DXF1ZXJ5QW5kRmV0Y2gBAAAAAAAAHaUWY1ExUVd0ZWlRY2",
"hits": {
"total": 1,
"hits": [
{
"_score": null
}
]
}
}
```
related issue: https://github.com/apache/incubator-doris/issues/3752
This CL mainly changes:
1. Add a new BE config `max_pushdown_conditions_per_column` to limit the number of conditions of a single column that can be pushed down to storage engine.
2. Add 2 new session variables `max_scan_key_num` and `doris_max_scan_key_num` which can set in session level and overwrite the config value in BE.
Add a JSON format for existing metrics like this.
```
{
"tags":
{
"metric":"thread_pool",
"name":"thrift-server-pool",
"type":"active_thread_num"
},
"unit":"number",
"value":3
}
```
I add a new JsonMetricVisitor to handle the transformation.
It's not to modify existing PrometheusMetricVisitor and SimpleCoreMetricVisitor.
Also I add
1. A unit item to indicate the metric better
2. Cloning tablet statistics divided by database.
3. Use white space to replace newline in audit.log
This CL mainly changes:
1. Support `SELECT INTO OUTFILE` command.
2. Support export query result to a file via Broker.
3. Support CSV export format with specified column separator and line delimiter.
Why this case happened
In current implement, translation into dsl only if it is not the first charactor.
Thus, when sql is write like '%abc', translation would not run.
How fixed
Now, translation will trigger with charactor '?' or '*'
if it is the first charactor, translate directly
else, check the preceding char is escaped or not to determin translation or not
1. Delete Invalid Counter In Data_Stream_Sender. (#3598)
2. Add Counter For PartitionHashTable of PartitionAggregationNode:
* Hash Probe Method
* Row processed by Aggregation
* HashFilledBuckets: Counter How Many FilledBuckets in Aggragation
* HTResize: Counter How Many Resize of HashTable
* HashProbe: Counter Probe of HashTable
* HashFailedProbe: Counter Failed Probe of HashTable
* HashTravelLength: Total TravelLength for Probe
* HashCollisions: Counter of HashCollision
3. Del some unecessary code in PartitionHashTable by template
#3479
Here I try to explain the cause of the problem and how to fix it.
**The Cause of The problem**
Take the case in issue(#3479 ) as an example:
The general results are as follows:
```
GET table/_doc/_search
{"query":{"match_all":{}},"stored_fields":"_none_","docvalue_fields":["k1"],"sort":["_doc"],"size":100}
{
"took": 6,
"timed_out": false,
"_shards": {
……
},
"hits": {
"total": 3,
"max_score": null,
"hits": [
{
"_index": "table",
"_score": null,
"sort": [
0
]
},
{
"_index": "table",
"_score": null,
"fields": {
"k1": [
"kkk1"
]
},
"sort": [
0
]
},
{
"_index": "table",
"_score": null,
"sort": [
0
]
}
]
}
}
```
But in Doris on ES，Be fetched data parallelly on all shards, and use `filter_path` to reduce the network cost. The process will be as follows:
```
GET table/_doc/_search?preference=_shards:1&filter_path=_scroll_id,hits.hits._source,hits.total,_id,hits.hits._source.fields,hits.hits.fields
{"query":{"match_all":{}},"stored_fields":"_none_","docvalue_fields":["k1"],"sort":["_doc"],"size":100}
{
"hits": {
"total": 0
}
}
GET table/_doc/_search?preference=_shards:2&filter_path=_scroll_id,hits.hits._source,hits.total,_id,hits.hits._source.fields,hits.hits.fields
{"query":{"match_all":{}},"stored_fields":"_none_","docvalue_fields":["k1"],"sort":["_doc"],"size":100}
{
"hits": {
"total": 1
}
}
GET table/_doc/_search?preference=_shards:3&filter_path=_scroll_id,hits.hits._source,hits.total,_id,hits.hits._source.fields,hits.hits.fields
{"query":{"match_all":{}},"stored_fields":"_none_","docvalue_fields":["k1"],"sort":["_doc"],"size":100}
{
"hits": {
"total": 1,
"hits": [
{
"fields": {
"k1": [
"kkk1"
]
}
}
]
}
}
```
*Scan-Worker On BE which processed result of shard2 will failed.*
**The reasons are as follows:**
1. "filter_path" causes the hits.hits object not exist.
2. In the current implementation, if there are some data rows（total > 0）, the hits.hits. object must be an array
**How To Fix it**
Two Method:
1. modify "filter_path" to contain the hits.
Pros: Fixed Code is very simple
Cons: More network cost
2. Deal with the case where fields are missing in a batch.
Pros: No loss of performance
Cons: Code is more complex
Performance first, I use Method2.
**Design**
1. Add a variable "_doc_value_mode" into Class "EsScrollParser" to =indicate whether the data processed by this parser is doc_value_mode or not.
2. "_doc_value_mode" is passed from ESScollReader <- ESScanner <- ScrollQueryBuilder::build() that determines whether DSL is enable doc_value_mode
3. When hits.hits of response from ES is empty and total > 0. We know there are data lines, but the corresponding fields do not exist. EsScrollParser will use "_doc_value_mode" and _total to construct _total lines which fields are assigned with 'NULL'
ImplementaItion Notes
NodeChannel
_cur_batch -> _pending_batches: when _cur_batch is filled up, move it to _pending_batches.
add_row() just produce batches.
try_send_and_fetch_status() tries to consume one pending batch. If has in flight packet, skip send in this round.
So we can add one sender thread to be in charge of all node channels try_send.
IndexChannel
init(), open() stay the same.
Use for_each_node_channel() to expose the detailed changes of NodeChannel.(It's more easy to read & modify)
Sender thread
See func OlapTableSink::_send_batch_process()
Why use polling？
If we use wait/notify, it will notify when generate a new batch. We can't skip sending this batch, coz it won't notify the same batch again. So wait/notify can't avoid blocking simply.
So I choose polling.
It's wasting to continuously try_send(), but it's difficult to set the suitable polling interval. Thus, I add std::this_thread::yield() to give up the time slice, give priority to other process/threads (if there are other process/threads waiting in the queue).
This CL mainly made the following modifications:
1. Delete Invalid MemoryUsed Counter and Add PeakMemUsage in each exec node and datastreamsender
2. Add intent in child execnode profile，make it is easily to know the relationship between execnode
3. Del _is_result_order we not support any more in olap_scan_node.h and olap_scan_node.cpp
4. Add scan_disk method to olap_scanner to fix the counter _num_disks_accessed_counter
5. Now we do not use buffer pool to read and write disk, so annotation eadio counter and
6. Delete the MemUsed counter in exec node.
The child.open() function is not called before this commit.
If the assert num rows node has child which process data in open function, the assert num rows node will fetch no data from child. So the result will be empty(incorrect).
This error only appear in inner subquery which has a aggregation function.
For example:
`select * from table where k1=(select k1 from (select avg(k1) from table) a);`
The first level of subquery returns a non-scalar value, so the assert num rows node is needed.
The second level of subquery has a aggregation function, so the child of assert node is aggregate node.
However, if the open stage of the aggregate node is not called, the get next state of aggregate node will return empty set.
So the result is wrong.
Fixed#3435.
This CL mainly made the following modifications:
1. Delete Invalid method in Running Profile Class.
2. Move Memlimit Counter from blockmgr to fragment and add PeakMemUsage Counter
3. Fix the bug of buffer pool memlimit counter
4. Call compute_time_in_profile() before pretty_print() to show the _local_time_percent without child running profile
5. Add TransferThread ThreadToken count in AveThreadToken Counter
Process castexpr, such as: k (float) > 2.0, k(int) > 3.2, Doris On Es should ignore this doris native cast transformation for every row's col value, we push down this `cast semantic` to Elasticsearch.
I believe in this `predicate` situation, would decrease the mount of data for transmission。
k1 is float:
````
k1 >= 5
````
push-down filter:
```
{"range":{"k1":{"gte":"5.000000"}}}
```
k2 is int :
```
k2 > 3.2
```
push-down filter:
```
{"range":{"k2":{"gte":"3.2"}}}
```
This PR is just a transitional way，but it is better to move the predicates transformation from Doris BE to Doris BE, in this way, Doris BE is responsible for fetching data from ES.
Add a `enable_keyword_sniff ` configuration item in creating External Elasticsearch Table ，it default to true , would to sniff the `keyword` type on the `text analyzed` Field and return the `json_path` which substitute the origin col name.
```
CREATE EXTERNAL TABLE `test` (
`k1` varchar(20) COMMENT "",
`create_time` datetime COMMENT ""
) ENGINE=ELASTICSEARCH
PROPERTIES (
"hosts" = "http://10.74.167.16:8200",
"user" = "root",
"password" = "root",
"index" = "test",
"type" = "doc",
"enable_keyword_sniff" = "true"
);
```
note: `enable_keyword_sniff` default to "true"
run this SQL：
```
select * from test where k1 = "wu yun feng"
```
Output predicate DSL：
```
{"term":{"k1.keyword":"wu yun feng"}}
```
and in this PR, I remove the elasticsearch version detected logic for now this is useless, maybe future is needed.
Relate Issue: https://github.com/apache/incubator-doris/issues/3248
SQL:
```
select * from test where (k2 = 6 and k3 = 1) or (k2 = 2 and k3 =3 and k4 = 'beijing');
```
Output filter:
```
((#k2:[6 TO 6] #k3:[1 TO 1]) (#(#k2:[2 TO 2] #k3:[3 TO 3]) #k4:beijing))~1
```
SQL:
```
select * from test where (k2 = 6 or k3 = 7) or (k2 = 2 and k3 =3 and (k4 = 'beijing' or k4 = 'zhaochun'));
```
Output filter:
```
(k2:[6 TO 6] k3:[7 TO 7] (#(#k2:[2 TO 2] #k3:[3 TO 3]) #((k4:beijing k4:zhaochun)~1)))~1
```
SQL:
```
select * from test where (k2 = 6 or k3 = 7) or (k2 = 2 and abs(k3) =3 and (k4 = 'beijing' or k4 = 'zhaochun'));
```
Output filter (`abs` can not be pushed down to es, so doris on es would not process this scenario ):
```
match_all
```
select date_format(k10, '%Y%m%d') as myk10 from baseall group by myk10;
The date_format function in query above will be stored in MemPool during
the query execution. If the query handles millions of rows, it will
consume much memory. Should clear the MemPool at interval.
improve performent of hash join when build table has to many duplicated rows, this will cause hash table collisions and slow down the probe performence.
In this pr when join type is semi join or anti join, we will build a hash table without duplicated rows.
benchmark:
dataset: tpcds dataset `store_sales` and `catalog_sales`
```
mysql> select count(*) from catalog_sales;
+----------+
| count(*) |
+----------+
| 14401261 |
+----------+
1 row in set (0.44 sec)
mysql> select count(distinct cs_bill_cdemo_sk) from catalog_sales;
+------------------------------------+
| count(DISTINCT `cs_bill_cdemo_sk`) |
+------------------------------------+
| 1085080 |
+------------------------------------+
1 row in set (2.46 sec)
mysql> select count(*) from store_sales;
+----------+
| count(*) |
+----------+
| 28800991 |
+----------+
1 row in set (0.84 sec)
mysql> select count(distinct ss_addr_sk) from store_sales;
+------------------------------+
| count(DISTINCT `ss_addr_sk`) |
+------------------------------+
| 249978 |
+------------------------------+
1 row in set (2.57 sec)
```
test querys:
query1: `select count(*) from (select store_sales.ss_addr_sk from store_sales left semi join catalog_sales on catalog_sales.cs_bill_cdemo_sk = store_sales.ss_addr_sk) a;`
query2: `select count(*) from (select catalog_sales.cs_bill_cdemo_sk from catalog_sales left semi join store_sales on catalog_sales.cs_bill_cdemo_sk = store_sales.ss_addr_sk) a;`
benchmark result:
||query1|query2|
|:--:|:--:|:--:|
|before|14.76 sec|3 min 16.52 sec|
|after|12.64 sec|10.34 sec|
