Multi-stage query task runtime
Fault tolerance is partially implemented. Workers get relaunched when they are killed unexpectedly. The controller does not get relaunched if it is killed unexpectedly.
Worker task stage outputs are stored in the working directory given by
druid.indexer.task.baseDir. Stages that generate a large amount of output data may exhaust all available disk space. In this case, the query fails with an UnknownError with a message including "No space left on device".
GROUPING SETSare not implemented. Queries using these features return a QueryNotSupported error.
The numeric varieties of the
LATESTaggregators do not work properly. Attempting to use the numeric varieties of these aggregators lead to an error like
java.lang.ClassCastException: class java.lang.Double cannot be cast to class org.apache.druid.collections.SerializablePair. The string varieties, however, do work properly.
REPLACEstatements with column lists, like
INSERT INTO tbl (a, b, c) SELECT ..., is not implemented.
INSERT ... SELECTand
REPLACE ... SELECTinsert columns from the
SELECTstatement based on column name. This differs from SQL standard behavior, where columns are inserted based on position.
EXTERNwith input sources that match large numbers of files may exhaust available memory on the controller task.
EXTERNrefers to external files. Use