SQL may forward to master to execute when connecting to follower node, the result should be set to `StmtExecutor#proxyResultSet`
Before this PR, in above scenario , submit analyze sql by mysql client/jdbc whould return get malformed packet/ Communication failed.
set enable function pushdown default to false.
enable it in fuzzy mode to test this feature.
We should remove function pushdown in the future since we already have common expr pushdown.
Co-authored-by: yiguolei <yiguolei@gmail.com>
In this PR, I remove the `makeSureInitialized()` call in `createTable()` method, because it is wrong and useless.
And also rename the methed's name to make it more clear.
1. The class 'ExternalDatabase' has implemented the 'GsonPostProcessable' interface, so
there is redundant codes in some subclass of 'ExternalDatabase'.
2. A LOG object is not used in this file.
Since the polling interval is 0, the CPU will be polled all the time when there is no data
Before and after comparison test, the CPU usage time is reduced by 2000 times
1. The Mysql Go driver has a logic that terminates when it reads an EOF (end-of-file) and expects no data in the buffer. However, the front-end (FE) mistakenly returns an additional OK packet, which causes an exception to be thrown when reading the buffer.
2. Refactor some logic to support full prepared not just in where clause, like
```
select ?, ? from tbl
```
fe foldconstRule make array() function expr with const literal , and would not pass this array literal to be . but we should make fe array string output format is same with be array string output
Add JNI metrics, for example:
```
- HudiJniScanner: 0ns
- FillBlockTime: 31.29ms
- GetRecordReaderTime: 1m5s
- JavaScanTime: 35s991ms
- OpenScannerTime: 1m6s
```
Add three common performance metrics for JNI scanner:
1. `OpenScannerTime`: Time to init and open JNI scanner
2. `JavaScanTime`: Time to scan data and insert into vector table in java side
3. `FillBlockTime`: Time to convert java vector table to c++ block
And support user defined metrics in java side, for example: `OpenScannerTime` is a long time for the open process, we want to determine which sub-process takes too much time, so we add `GetRecordReaderTime` in java side.
The user defined metrics in java side can be attached to BE profile automatically.
Problem:
when use select group_concat(distinct a, 'seg1'), group_concat(distinct b, 'seg2') ... Error would rised
Reason:
Group_concat function regard 'seg' as arguments also, so multi distinct column error would rised
Solved:
let Multi Distinct group_concat function only get first argument as real argument
The current column statistic cache loader is to load data from column_statistics olap table.
This pr is to change the cache loader logic to First load from column_statistics olap table, if no data was loaded, then load from table metadata. This is mainly to support fetch statistics data for external catalog using HMS or Iceberg api.
This is the first PR, next pr will implement the fetch logic for different external catalogs.
Two optimizations:
1. Insert string bytes directly to remove decoding&encoding process.
2. Use native reader to read the hudi base file if it has no log file. Use `explain` to show how many splits are read natively.
When we use a label to load data, this label can not be used twice. But when we execute a sql 'CLEAN LABEL [label] FROM db;', we hope that the same label can be used again.
However, the sql above does not work. This PR is fixing this problem.
if we do type coercion on subquery, it return datatype after type coercion
error info
```
Both side of binary arithmetic is not numeric. left type is DECIMALV3(2, 1) and right type is DECIMAL(27, 9)')
```
in hash join condition, some equals are trustable, some are not.
an equal is trustable if one side is almost unique, like primary key. for such equal condition we could estimate more accurate.
the problem is in rewriten q20, the are 2 equal condition, one is trustable, another is not. But we treat both of them as trustable.
Test result:
on tpch100, from 2.2 sec to 0.44 sec
no impact on tpch other queries
no performance impact on tpcds queries
1. Don't write to table_statistics when create sync analyze job anymore since it's meaningless
2. Capture exceptions when creating each system analyze job to avoid the failure of creation of all automatic collection jobs due to a single job creation failure.
3. Mark auto triggered period job's job type as system
