colocated join is depended on if the both side of the join conjuncts are simple column with same distribution policy etc. So the key is to figure out the original source column in scan node if there is one. To do that, we should check the slot from both lhs and rhs of outputSmap in join node.
In current implementation, the class Auth is used for:
Manager all authentication and authorization info such as user, role, password, privileges.
Provide an interface for privilege checking
Some user may want to integrate external access management system such as Apache Ranger.
So we should provide a way to let user set their own access controller.
This PR mainly changes:
A new class SystemAccessController
This access controller is used to check the global level privileges and resource privileges.
A new interface CatalogAccessController
This interface is used to check catalog/database/tbl level privileges.
It has a default implements InternalCatalogAccessController.
All privilege checking methods are moved from Auth to either SystemAccessController or
InternalCatalogAccessController
A new class AccessControllerManager
This is the entry point of privilege authentication. All methods previously called from Auth
now are called from AccessControllerManager
Now, user can implement the interface CatalogAccessController to use their own access controller.
And when creating external catalog, user can specified the access controller class name, so that
different external catalog can use different access controller.
Support IPV6 in Apache Doris, the main changes are:
1. enable binding to IPV6 address if network priority in config file contains an IPV6 CIDR string
2. BRPC and HTTP support binding to IPV6 address
3. BRPC and HTTP support visiting IPV6 Services
Hive 1.x may write orc file with internal column name (_col0, _col1, _col2...).
This will cause query result be NULL because column name in orc file doesn't match
with column name in Doris table schema. This pr is to support query Hive orc files with internal column names.
For now, we haven't see any problem in Parquet file, will send new pr to fix parquet if any problem show up in the future.
* [improvement](block exception safe) make block queue exception safe
This is part of exception safe: #16366.
---------
Co-authored-by: yiguolei <yiguolei@gmail.com>
In our current logic, index page will be pre-decoded but it will return OK
as index page use BinaryPlainPageBuilder and first 4 bytes of the page is a offset
so it's high probablility not equal to EncodingTypePB::DICT_ENCODING which
is 5.
Code in bitshuffle_page_pre_decode.h
```
if constexpr (USED_IN_DICT_ENCODING) {
auto type = decode_fixed32_le((const uint8_t*)&data.data[0]);
if (static_cast<EncodingTypePB>(type) != EncodingTypePB::DICT_ENCODING) {
return Status::OK();
}
size_of_dict_header = BINARY_DICT_PAGE_HEADER_SIZE;
data.remove_prefix(4);
}
```
But if type just equal to EncodingTypePB::DICT_ENCODING and then it will use
BitShuffle to decode BinaryPlainPage, which will leads to an fatal error.
In previous version, if the output slot of analyticExpr is not materialized, the analyticExpr is pruned.
But there are some cases that it cannot be pruned.
For example:
SELECT
count(*)
FROM T1,
(SELECT dd
FROM (
SELECT
1.1 as cc,
ROW_NUMBER() OVER() as dd
FROM T2
) V1
ORDER BY cc DESC
limit 1
) V2;
analyticExpr(ROW_NUMBER() OVER() as dd) is not materialized, but we have to generate
WindowGroup for it.
tmp.dd is used by upper count(*), we have to generate data for tmp.dd
In this fix, if an inline view only output one column(in this example, the 'dd'), we materialize this column.
TODO:
In order to prune 'ROW_NUMBER() OVER() as dd', we need to rethink the rule of choosing a column
for count(*). (refer to SingleNodePlanner.materializeTableResultForCrossJoinOrCountStar)
V2 can be transformed to
SELECT cc
FROM (
SELECT
1.1 as cc,
ROW_NUMBER() OVER() as dd
FROM T2
) V1
ORDER BY cc DESC
limit 1
) V2;
Except the byte size of cc and dd, we need to consider the cost to generate cc and dd.
MySql load can load fe server node, but it will cause secure issue that user use it to detect the fe node local file.
For this reason, add a configuration named mysql_load_server_secure_path to set a secure path to load data.
By default, load fe local file feature is disabled by this configuration.
