Miklos Gergely created HIVE-23123:
-------------------------------------
Summary: Export/Import of views and materialized views
Key: HIVE-23123
URL: https://issues.apache.org/jira/browse/HIVE-23123
Project: Hive
Issue Type: Bug
Components: Hive
Reporter: Miklos Gergely
Assignee: Miklos Gergely
According to
https://cwiki.apache.org/confluence/display/Hive/LanguageManual+ImportExport
import and export can be done by using the
{code:java}
export table ...
import table ...
{code}
commands. The document doesn't mention views or materialized views at all, and
in fact we don't support commands like
{code:java}
export view ...
import view ...
export materialized view ...
import materialized view ...
{code}
they can not be parsed at all. The word table is often used though in a broader
sense, when it means all table like entities, including views and materialized
views. For example the various Table classes may represent any of these as well.
If I try to export a view with the export table ... command, it goes fine. A
_metadata file will be created, but no data directory, which is what we'd
expect. If I try to import it back, an exception is thrown due to the lack of
the data dir:
{code:java}
java.lang.AssertionError: null==getPath() for exim_view
at org.apache.hadoop.hive.ql.metadata.Hive.loadTable(Hive.java:3088)
at org.apache.hadoop.hive.ql.exec.MoveTask.execute(MoveTask.java:419)
at org.apache.hadoop.hive.ql.exec.Task.executeTask(Task.java:213)
at org.apache.hadoop.hive.ql.exec.TaskRunner.runSequential(TaskRunner.java:105)
at org.apache.hadoop.hive.ql.Executor.launchTask(Executor.java:364)
at org.apache.hadoop.hive.ql.Executor.launchTasks(Executor.java:335)
at org.apache.hadoop.hive.ql.Executor.runTasks(Executor.java:246)
at org.apache.hadoop.hive.ql.Executor.execute(Executor.java:109)
at org.apache.hadoop.hive.ql.Driver.runInternal(Driver.java:722)
at org.apache.hadoop.hive.ql.Driver.run(Driver.java:491)
at org.apache.hadoop.hive.ql.Driver.run(Driver.java:485)
{code}
Still the view gets imported successfully, as data movement wasn't even
necessary.
If we try to export a materialized view which is transactional, then this
exception occurs:
{code:java}
org.apache.hadoop.hive.ql.parse.SemanticException:
org.apache.hadoop.hive.ql.metadata.InvalidTableException: Table not found
exim_materialized_view_da21d41a_9fe4_4446_9c72_d251496abf9d
at
org.apache.hadoop.hive.ql.parse.AcidExportSemanticAnalyzer.analyzeAcidExport(AcidExportSemanticAnalyzer.java:163)
at
org.apache.hadoop.hive.ql.parse.AcidExportSemanticAnalyzer.analyze(AcidExportSemanticAnalyzer.java:71)
at
org.apache.hadoop.hive.ql.parse.RewriteSemanticAnalyzer.analyzeInternal(RewriteSemanticAnalyzer.java:72)
at
org.apache.hadoop.hive.ql.parse.BaseSemanticAnalyzer.analyze(BaseSemanticAnalyzer.java:289)
at org.apache.hadoop.hive.ql.Compiler.analyze(Compiler.java:220)
at org.apache.hadoop.hive.ql.Compiler.compile(Compiler.java:104)
at org.apache.hadoop.hive.ql.Driver.compile(Driver.java:183)
at org.apache.hadoop.hive.ql.Driver.compileInternal(Driver.java:601)
at org.apache.hadoop.hive.ql.Driver.compileAndRespond(Driver.java:547)
at org.apache.hadoop.hive.ql.Driver.compileAndRespond(Driver.java:541)
{code}
So the export process can not handle it, as the temporary table is not getting
created.
The import command handling have a lot of codes dedicated to importing views
and materialized views, which suggests that we support the importing (and thus
also suggests implicitly that we support the exporting) of views and
materialiezed views.
So the conclusion is that we have to decide if we support exporting/importing
of views and materialized views.
# If we decide not to support them then:
- export process should throw an exception if a view or materialized view is
the subject
- the codes specific to view imports should be removed
# If we decide to support them, then:
- the commands mentioned above above should be introduced
- exception should be thrown if not the proper command used (e.g. export view
on a table)
- the exceptions mentioned above should be fixed
I prefer #1, I don't think we should support the exporting / importing of
views. The point of exporting / importing is the transfer of data, not DDL, it
causes more issues than it solves. Our current documentation also suggests that
it is only supported for tables.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
