TabletSchema and Schema no longer track memory, only track columns count. because cannot accurately track memory size.
TabletMeta MemTracker changed to track TabletSchema columns count.
Segment::_meta_mem_usage Unknown value overflow, causes the value of SegmentMeta MemTracker is similar to -2912341218700198079. So, temporarily put it in experimental type tracker.
now in pipeline, when result block queue is empty, will be reschedule, and then choose a batch of scanner,
but sometimes, get_available_thread_slot_num() will return thread_slot_num <= 0, so it's will do nothing,
and then block queue will always empty.
have no chance to reschedule again until query timeout.
1. create aggregation table
2. insert some data
3. drop the table and create again
4. modify some parameters for some branch
5. insert some data
6. change the parameters back to its default
For left outer join or full outer join, when build side data is empty, null data is output for build side, but nested column data of nullable column is not properly initialized, which may cause decimal arithmetic overflow
Currently, _flush_active_memtables() is using stale memtracker data, especially when some other thread has just it.
Refresh memtrackers before flush to avoid this problem.
_child_block in nest loop join , table value function, repeat node will be shared between ExecNode and related operator, but it should not be a unique ptr in operator, it belongs to exec node.
It will double free the block, if operator's close method is not called correctly.
It should be a shared ptr, then it will not core even if the opeartor's close method is not called.
Refactor write path code by abstract base class. Whether to use `StorageEngine` or `CloudStorageEngine` will be determined during compilation instead of runtime `config::cloud_mode` to avoid unexpected null pointer or undefined behavior issues caused by merging code.
Class that depend on `StorageEngine` but are shared by the cloud mode need to have an abstract base class. Common code should be extracted into the base class, while the code that depends on `StorageEngine` should be implemented in a `StorageEngine` mix-in class of the base class.
BE will core dump if result block is invalid when we doing result serialization.
An existing bug case is described in #28030, so we add check branch to avoid BE core dump due to out of range related problem.
* [regression-test](Variant) Add more cases related to schema changes
And fix bugs about schema change for variant:
fix bug schema change crash on doing schema change with tablet schema that contains extracted columns
