Fix three bugs:
1. `repeated_parent_def_level ` should be the definition of its repeated parent.
2. Failed to parse schema like `decimal(p, s)`
3. Fill wrong offsets for array type
This pr is a temporary circumvention fix.
In regression case `inverted_index_p0/test_add_drop_index.groovy`, both bitmap index and inverted index are created
on the same table, when create or drop bitmap index will change table's state to `SCHEMA_CHANGE`, create or drop
inverted index not change the table's state.
Before do create or drop inverted index check the table's state whether is `NORMAL` or not, Because of replay log for
'bitmap index' has change table state, and it didn't finish soon lead to table's state not change back to `NORMAL`, then
replay log for 'inverted index' failed, FE start failed.
remove duplicate type definition in function context
remove unused method in function context
not need stale state in vexpr context because vexpr is stateless and function context saves state and they are cloned.
remove useless slot_size in all tuple or slot descriptor.
remove doris_udf namespace, it is useless.
remove some unused macro definitions.
init v_conjuncts in vscanner, not need write the same code in every scanner.
using unique ptr to manage function context since it could only belong to a single expr context.
Issue Number: close #xxx
---------
Co-authored-by: yiguolei <yiguolei@gmail.com>
This is the first step to introduce official hadoop libhdfs to Doris.
Because the current hdfs client libhdfs3 lacks some important feature and is hard to maintain.
Download the hadoop 3.3.4 binary from hadoop website: https://hadoop.apache.org/releases.html
Extract libs and headers which are used for libhdfs, and pack them into hadoop_lib_3.3.4-x86.tar.gz
Upload it to https://github.com/apache/doris-thirdparty/releases/tag/hadoop-libs-3.3.4
TODO:
The hadoop libs for arm is missing, we need to find a way to build it
* [enhancement](transaction) Reduce hold writeLock time for DatabaseTransactionMgr to clear transaction
* fix ut
* remove unnessary field for remove txn bdbje log
---------
Co-authored-by: caiconghui1 <caiconghui1@jd.com>
Modify the default value of mem_limit to auto. auto means process mem limit is equal to max(physical mem * 0.9, 6.4G).
6.4G is the maximum memory reserved for the system.
when a restore job which has a plenty of replicas, it may fail due to timeout. The error message is：
[RestoreJob.checkAndPrepareMeta():782] begin to send create replica tasks to BE for restore. total 381344 tasks. timeout: 600000
Currently, the max value of timeout is fixed, it's not suitable for such cases.
fix heap-use-after-free
The OrcReader has a internal FileInputStream, If the file is empty, the memory of FileInputStream will leak.
Besides, there is a Statistics instance in FileInputStream. FileInputStream maybe delete if the orc reader
is inited failed, but Statistics maybe used when orc reader is closed, causing heap-use-after-free error.
Potential memory leak
When init file scanner in file scan node, the file scanner prepare failed, the memory of file scanner will leak.
segcompaction is async and in parallel with load job. If the load job is
canncelling, memory structures will be destroyed and cause segcompaction
crash. This commit will wait segcompaction finished before destruction.
* [enhancement](stream load pipe) using queryid or load id to identify stream load pipe instead of fragment instance id
NewLoadStreamMgr already has pipe and other info. Do not need save the pipe into fragment state. and FragmentState should be more clear.
But this pr will change the behaviour of BE.
I will pick the pr to doris 1.2.3 and add the load id to FE support. The user could upgrade from 1.2.3 to 2.x
Co-authored-by: yiguolei <yiguolei@gmail.com>
turn on all test case in scalar function W except width_bucket(fix be bug in next PR)
turn off all test case for group_concat(distinct order by)
fix return nullable in TimestampArithmetic
Clear transaction state log occupies too much time, so we change clear transaction log level from info to debug
Co-authored-by: caiconghui1 <caiconghui1@jd.com>
In #16343, we split the timeout variable into two ones (one is for query and another is for insertion).
The function `ConnectProcessor::handleQuery` uses the corresponding session variable to change the timeout for the queries requested by MySQL client. However, the function `StmtExecutor::handleInsert` doesn't use the session variable to change the timeout, so we can't change the timeout for the CTAS and MTMV insertion job.
Reason: column_name[k5], decimal value is not valid for definition, value=123.123, precision=9, scale=3, min=-999999.999, max=-999999.999; . src line [];
#17273
