[SPARK-20194] Add support for partition pruning to in-memory catalog #17510
Conversation
|
ok to test |
| _.references.map(_.name).toSet.subsetOf(partitionColumnNames) | ||
| } | ||
| if (nonPartitionPruningPredicates.nonEmpty) { | ||
| sys.error("Expected only partition pruning predicates: " + nonPartitionPruningPredicates) |
There was a problem hiding this comment.
Nit: Throwing an AnalysisException is preferred.
There was a problem hiding this comment.
Could you add the negative test cases in your newly added test cases?
| } | ||
|
|
||
| test("list partitions by filter") { | ||
| val tz = TimeZone.getDefault().getID() |
There was a problem hiding this comment.
Nit: val tz = TimeZone.getDefault.getID
|
Not related to this PR. It sounds like we have a bug in |
|
Test build #75461 has finished for PR 17510 at commit
|
|
Test build #75464 has finished for PR 17510 at commit
|
|
LGTM cc @cloud-fan @hvanhovell @ericl |
| if (predicates.nonEmpty) { | ||
| val clientPrunedPartitions = client.getPartitionsByFilter(rawTable, predicates).map { part => | ||
| val clientPrunedPartitions = | ||
| client.getPartitionsByFilter(rawTable, predicates).map { part => |
There was a problem hiding this comment.
if predicates.isEmpty, the previous code will run client.getPartitions. Can you double check there is no performance regression?
There was a problem hiding this comment.
A similar optimization is done in the function itself: client.getPartitionsByFilter(), while Hive Shim_v0_12 delegates to getAllPartitions anyway.
I've now made the only piece of non-trivial code along that path lazy, so I think we're good.
| val catalog = newBasicCatalog() | ||
|
|
||
| def checkAnswer(table: CatalogTable, filters: Seq[Expression], | ||
| expected: Set[CatalogTablePartition]): Unit = { |
There was a problem hiding this comment.
nit: code style
def checkAnswer(
param1: XX, param2: XX, ...
|
LGTM except some minor comments |
|
Test build #75479 has finished for PR 17510 at commit
|
|
Thanks! Merging to master. |
|
Thanks for moving so fast! |
4 participants
What changes were proposed in this pull request?
This patch implements
listPartitionsByFilter()forInMemoryCatalogand thus resolves an outstanding TODO causing thePruneFileSourcePartitionsoptimizer rule not to apply when "spark.sql.catalogImplementation" is set to "in-memory" (which is the default).The change is straightforward: it extracts the code for further filtering of the list of partitions returned by the metastore's
getPartitionsByFilter()out fromHiveExternalCatalogintoExternalCatalogUtilsand calls this new function fromInMemoryCatalogon the whole list of partitions.Now that this method is implemented we can always pass the
CatalogTableto theDataSourceinFindDataSourceTable, so that the latter is resolved to a relation with aCatalogFileIndex, which is what thePruneFileSourcePartitionsrule matches for.How was this patch tested?
Ran existing tests and added new test for
listPartitionsByFilterinExternalCatalogSuite, which is subclassed by bothInMemoryCatalogSuiteandHiveExternalCatalogSuite.