1. add support `CAST AS Struct` from Struct type;
2. fix crash while `CAST('{}' AS Struct)`;
3. `CAST('' AS complext_type)` should return NULL instead of empty object;
---------
Co-authored-by: cambyzju <zhuxiaoli01@baidu.com>
* add clickbench and arm pipeline trigger；test merge check
* set compile required; add clickbench and arm pipeline trigger to buildall;
* avoid duplicate builds
* simplify auto trigger code and avoid repeated triggers 0310
for some reasons, transaction pushlish succeed replica num less than quorum,
this transaction's status can not to be VISIBLE, and this publish task of this
replica of this tablet on this backend need retry publish success to
make transaction VISIBLE when last publish failed.
Signed-off-by: nextdreamblue <zxw520blue1@163.com>
the input replicate_offsets should be the same size as ColumnArray's offset.
```
IColumn::Offsets replicate_offsets(get_offsets().size(), 0);
// |---------------------|-------------------------|-------------------------|
// [0, begin) [begin, begin + count_sz) [begin + count_sz, size())
// do not need to copy copy counts[n] times do not need to copy
```
we should
The current distribution model for Doris is as follows:
OlapTableSink seperate the original Block into serveral subblocks of each node(BE) by tablets distribution and distributes subblocks to storage engine of backends, then the storage engine will seperate the subblock into multiple tablets channel and each delta writer will handle partial of the block.
This model causes blocks to be split according to tablets, and the splitting process can be a relatively heavy operation. After splitting, the blocks are distributed to different DeltaWriters (Memtables) through RPCs to TabletChannels. The distribution operation on TabletChannels is also a relatively heavy operation. If the distribution property of the table is RANDOM distribution, then we have the opportunity to distribute the blocks according to the complete block during distribution. The advantage of doing so is to reduce memory copying and improve write locality, similar to appending the entire block to the memtable.
This optimze could save 10% ~ 20% CPU cost of RANDOM distribution table load when enable load_to_single_tablet
If table has no partition, backup will report error:
2023-03-06 17:35:32,971 ERROR (backupHandler|24) [Daemon.run():118] daemon thread got exception. name: backupHandler
java.util.NoSuchElementException: No value present
at java.util.Optional.get(Optional.java:135) ~[?:1.8.0_152]
at org.apache.doris.catalog.OlapTable.selectiveCopy(OlapTable.java:1259) ~[doris-fe.jar:1.0-SNAPSHOT]
at org.apache.doris.backup.BackupJob.prepareBackupMeta(BackupJob.java:505) ~[doris-fe.jar:1.0-SNAPSHOT]
at org.apache.doris.backup.BackupJob.prepareAndSendSnapshotTask(BackupJob.java:398) ~[doris-fe.jar:1.0-SNAPSHOT]
at org.apache.doris.backup.BackupJob.run(BackupJob.java:301) ~[doris-fe.jar:1.0-SNAPSHOT]
at org.apache.doris.backup.BackupHandler.runAfterCatalogReady(BackupHandler.java:188) ~[doris-fe.jar:1.0-SNAPSHOT]
at org.apache.doris.common.util.MasterDaemon.runOneCycle(MasterDaemon.java:58) ~[doris-fe.jar:1.0-SNAPSHOT]
at org.apache.doris.common.util.Daemon.run(Daemon.java:116) ~[doris-fe.jar:1.0-SNAPSHOT]
Example:
SELECT ROUTINE_SCHEMA AS PROCEDURE_CAT, NULL AS PROCEDURE_SCHEM,ROUTINE_NAME AS PROCEDURE_NAME,NULL AS NUM_INPUT_PARAMS,NULL AS NUM_OUTPUT_PARAMS,NULL AS NUM_RESULT_SETS,ROUTINE_COMMENT AS REMARKS,IF(ROUTINE_TYPE = 'FUNCTION', 2,IF(ROUTINE_TYPE= 'PROCEDURE', 1, 0)) AS PROCEDURE_TYPE FROM INFORMATION_SCHEMA.ROUTINES WHERE ROUTINE_SCHEMA = DATABASE();
ERROR 1105 (HY000): errCode = 2, detailMessage = invalid parameter
This wrong and some BI tools could not work correctly.
In the past, only simple predicates (slot=const), and, like, or (only bitmap index) could be pushed down to the storage layer. scan process:
Read part of the column first, and calculate the row ids with a simple push-down predicate.
Use row ids to read the remaining columns and pass them to the scanner, and the scanner filters the remaining predicates.
This pr will also push-down the remaining predicates (functions, nested predicates...) in the scanner to the storage layer for filtering. scan process:
Read part of the column first, and use the push-down simple predicate to calculate the row ids, (same as above)
Use row ids to read the columns needed for the remaining predicates, and use the pushed-down remaining predicates to reduce the number of row ids again.
Use row ids to read the remaining columns and pass them to the scanner.
* (merge-on-write) when if publish and be down, need recalc delete bitmap for MoW
Signed-off-by: nextdreamblue <zxw520blue1@163.com>
* fix code
Signed-off-by: nextdreamblue <zxw520blue1@163.com>
---------
Signed-off-by: nextdreamblue <zxw520blue1@163.com>
Notice some changes:
1. Support cancel query for mysql load
2. Change the thread pool for mysql load manager.
3. Fix sucret path check logic
4. Fix some doc error
before this pr, add or drop inverted index not change table state, maybe multiple alter jobs executed at the same time, that may lead to some unexpected problems.
This pull request for bellow problem :
when fe config set sys_log_verbos_modules = org.apache.doris, which will make fe get writeLock longer. In this config, make a stream load, that stream load will failed with this message ([ANALYSIS_ERROR]errCode = 2, detailMessage = get tableList write lock timeout, tableList=(Table [id=86135, name=flink_connector, type=OLAP]))
refactor cost-model frameWork:
1. Use Cost class to encapsulate double cost
2. Use the `addChildCost` function to calculate the cost with children rather than add directly
Note we use the `Cost` class because we hope to customize the operator of adding a child host. Therefore, only when the cost would add the child Cost or be added by the parent we use `Cost`. Otherwise, we use double such as `upperbound`
1. 'insert into' profile has 'insert' type, can not query by 'load' type
2. 'insert into' profile does not have job_id, can not query by job_id. so put all profiles key with query_id
3. 'broker load' profile does not have some infos, npe
ECB algorithm, block_encryption_mode does not take effect, it only takes effect when init vector is provided.
Solved: 192/256 supports calculation without init vector
For other algorithms, an error should be reported when there is no init vector
Initialization Vector. The default value for the block_encryption_mode system variable is aes-128-ecb, or ECB mode, which does not require an initialization vector. The alternative permitted block encryption modes CBC, CFB1, CFB8, CFB128, and OFB all require an initialization vector.
Reference: https://dev.mysql.com/doc/refman/8.0/en/encryption-functions.html#function_aes-decrypt
Note: This fix does not support smooth upgrades. during upgrade process, query may report error: funciton not found
