This PR mainly adds a rewrite rule 'ExtractCommonFactorsRule'
used to extract wide common factors in the planning stage for 'Expr'.
The main purpose of this rule is to extract (Range or In) expressions
that can be combined from each or clause.
E.g:
Origin expr: (1<a<3 and b in ('a') ) or (2<a<4 and b in ('b'))
Rewritten expr: (1<a<4 ) and (b in ('a', 'b')) and ((1<a<3 and b in ('a') ) or (2<a<4 and b in ('b')))
Although the range of the wide common factors is larger than the real range,
the wide common factors only involve a single column, so it can be pushed down to the scan node,
thereby reducing the amount of scanned data in advance and improving the query speed.
It should be noted that this optimization strategy is not for all scenarios.
When filter rate of the wide common factor is too low,
the query will consume an extra time to calculate the wide common factors.
So this strategy can be switched by configuring session vairables 'extract_wide_range_expr'.
The default policy is enabled which means this strategy takes effect.
If you encounter unsatisfactory filtering rate, you can set the variable to false.
It will turn off the strategy.
Fixed#6082
* [Enhance] Support show unrecoverable tablets
The unrecoverable tablets are tablets which non of their replicas are healthy.
We should be able to find out these tablets then manual intervention.
And these tablets should not be added to the tablet scheduler.
random_shuflle will generate same random sequence when call multiple times,
although we use twice random, but when there is no change in the size relationship
between the adjacent numbers, the result of the second shuffle will not change either
* Organize FE configuration file description
Organize FE configuration file description
* Delete redundant numbers
Delete redundant numbers
* Add two configuration parameters of spring boot upload file
Add two configuration parameters of spring boot upload file
* Add configuration instructions
Add configuration instructions
* Fix typos
Fix typos
* Add English documentation of BE configuration
Add English documentation of BE configuration
* Modify style
Modify style
* Modify punctuation
Modify punctuation
* Correct the errors in the text
Correct the errors in the text
* Modify some ads and content issues
Modify some ads and content issues
* [Bug-fix] Fix wrong data distribution judgment
The Fragment where OlapScanNode is located has three data distribution possibilities.
1. UNPARTITIONED: The scan range of OlapScanNode contains only one instance(BE)
2. RANDOM: Involving multi-partitioned tables in OlapScanNode.
3. HASH_PARTITIONED: The involving table is in the colocate group.
For a multi-partition table, although the data in each individual partition is distributed according to the bucketing column,
the same bucketing column between different partitions is not necessarily in the same be.
So the data distribution is RANDOM.
If Doris wrongly plan RANDOM as HASH_PARTITIONED, it will lead to the wrong colocate agg node.
The result of query is incorrect.
To avoid showing too many memtracker on BE web pages.
The MemTracker level now has 3 levels: OVERVIEW, TASK and VERBOSE.
OVERVIEW Mainly used for main memory consumption module such as Query/Load/Metadata.
TASK is mainly used to record the memory overhead of a single task such as a single query, load, and compaction task.
VERBOSE is used for other more detailed memtrackers.
QueryDetail is used to statistic the current query details.
This property will only be set when the query starts to execute.
So in the query planning stage, using this attribute in the first query will cause 'NullPointerException'.
After that, this attribute retains the value of the previous query
until it is updated by the subsequent process.
Because code of 'colocateagg' uses this attribute incorrectly in its planning,
it causes 'NullPointerException' when clients like pymysql connect to doris and send the first query.
Fixed#6017
We created multiple rowset readers to read data of one tablet,
after one rowset reader has reached EOF, it can be released to
reduce resource (typically memory) consumption.
As the same, we can release segment reader when it reach EOF.
SelectStmt use new TableName(null, tableRef.getAlias()) to expand star expression. tableRef.getAlias() is full name include database name and table name.
Using it as table name will generate wrong sql in CreateViewStmt.
This patch fix this problem and use correct database name and table name in expand star method.
