This is an automated email from the ASF dual-hosted git repository.

wzhou pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/impala.git

commit 27b45ad32ab72af266be8b7d28aa9761c85e5219
Author: Eyizoha <[email protected]>
AuthorDate: Thu Oct 26 16:11:52 2023 +0800

    IMPALA-12508: Update query option "enabled_runtime_filter_types" comments
    
    The query option "enabled_runtime_filter_types" currently allows
    specifying a comma-separated list of runtime filter types (including
    BLOOM, MIN_MAX and IN_LIST) to enable or setting it to "all" to use all
    types. But the comment for this query option does not mention IN_LIST
    and states that it only applies to Kudu, which is outdated, as it also
    affects HDFS scans after IMPALA-10325.
    
    Therefore, this patch update the relevant comments to avoid any
    misleading caused by the outdated comments.
    
    Change-Id: Iecff7c655e273712d851d9ce94ef5713e579aa72
    Reviewed-on: http://gerrit.cloudera.org:8080/20627
    Reviewed-by: Impala Public Jenkins <[email protected]>
    Tested-by: Impala Public Jenkins <[email protected]>
---
 common/thrift/ImpalaService.thrift                         | 14 +++++++++++---
 .../org/apache/impala/planner/RuntimeFilterGenerator.java  |  3 ++-
 2 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/common/thrift/ImpalaService.thrift 
b/common/thrift/ImpalaService.thrift
index 83e37f629..cb1a3f1b3 100644
--- a/common/thrift/ImpalaService.thrift
+++ b/common/thrift/ImpalaService.thrift
@@ -532,11 +532,19 @@ enum TImpalaQueryOptions {
   RETRY_FAILED_QUERIES = 102
 
   // Enabled runtime filter types to be applied to scanner.
-  // This option only apply to Kudu now, will apply to HDFS once we support
-  // min-max filter for HDFS.
+  // This option only apply to Hdfs scan node and Kudu scan node.
+  // Specify the enabled types by a comma-separated list or enable all types 
by "ALL".
   //     BLOOM   - apply bloom filter only,
   //     MIN_MAX - apply min-max filter only.
-  //     ALL     - apply both bloom filter and min-max filter (default).
+  //     IN_LIST - apply in-list filter only.
+  //     ALL     - apply all types of runtime filters.
+  // Default is [BLOOM, MIN_MAX].
+  // Depending on the scan node type, Planner can schedule compatible runtime 
filter type
+  // as follows:
+  // Kudu scan: BLOOM, MIN_MAX
+  // Hdfs scan on Parquet file: BLOOM, MIN_MAX
+  // Hdfs scan on ORC file: BLOOM, IN_LIST
+  // Hdfs scan on other kind of file: BLOOM
   ENABLED_RUNTIME_FILTER_TYPES = 103
 
   // Enable asynchronous codegen.
diff --git 
a/fe/src/main/java/org/apache/impala/planner/RuntimeFilterGenerator.java 
b/fe/src/main/java/org/apache/impala/planner/RuntimeFilterGenerator.java
index 5034a61ad..fc69cdc06 100644
--- a/fe/src/main/java/org/apache/impala/planner/RuntimeFilterGenerator.java
+++ b/fe/src/main/java/org/apache/impala/planner/RuntimeFilterGenerator.java
@@ -1087,7 +1087,8 @@ public final class RuntimeFilterGenerator {
    *    scan node.
    * 3. Only Hdfs and Kudu scan nodes are supported:
    *     a. If the target is an HdfsScanNode, the filter must be type 
BLOOM/IN_LIST for
-   *        non Parquet tables, or type BLOOM/MIN_MAX/IN_LIST for Parquet 
tables.
+   *        ORC tables, or type BLOOM/MIN_MAX for Parquet tables, or type 
BLOOM for other
+   *        kind of tables.
    *     b. If the target is a KuduScanNode, the filter could be type MIN_MAX, 
and/or
    *        BLOOM, the target must be a slot ref on a column, and the comp op 
cannot
    *        be 'not distinct'.

Reply via email to