godfreyhe commented on a change in pull request #8980: [FLINK-13040] promote
blink table config
URL: https://github.com/apache/flink/pull/8980#discussion_r300522590
##########
File path:
flink-table/flink-table-planner-blink/src/main/java/org/apache/flink/table/api/PlannerConfigOptions.java
##########
@@ -52,95 +54,117 @@
public static final ConfigOption<Boolean>
SQL_OPTIMIZER_SHUFFLE_PARTIAL_KEY_ENABLED =
key("sql.optimizer.shuffle.partial-key.enabled")
.defaultValue(false)
- .withDescription("Enables shuffle by
partial partition keys. " +
+ .withDescription("Enables shuffling by
partial partition keys. " +
"For example, A join
with join condition: L.c1 = R.c1 and L.c2 = R.c2. " +
"If this flag is
enabled, there are 3 shuffle strategy:\n " +
"1. L and R shuffle by
c1 \n" +
"2. L and R shuffle by
c2\n" +
"3. L and R shuffle by
c1 and c2\n" +
"It can reduce some
shuffle cost someTimes.");
+ @Documentation.ExcludeFromDocumentation(value = "Temporary solution to
enable optimization that removes redundant sort for SortMergeJoin, " +
+ "and this config option will be removed later.")
public static final ConfigOption<Boolean>
SQL_OPTIMIZER_SMJ_REMOVE_SORT_ENABLED =
- key("sql.optimizer.smj.remove-sort.enabled")
+ key("sql.optimizer.sort-merge-join.remove-sort.enabled")
.defaultValue(false)
.withDescription("When true, the
optimizer will try to remove redundant sort for SortMergeJoin. " +
"However that will
increase optimization time. Default value is false.");
- public static final ConfigOption<Long>
SQL_OPTIMIZER_HASH_JOIN_BROADCAST_THRESHOLD =
- key("sql.optimizer.hash-join.broadcast.threshold")
+ public static final ConfigOption<Long>
SQL_OPTIMIZER_BROADCAST_JOIN_THRESHOLD =
+ key("sql.optimizer.broadcast.join.threshold")
.defaultValue(1024 * 1024L)
- .withDescription("Maximum size in bytes
for data that could be broadcast to each parallel " +
- "instance that holds a
partition of all data when performing a hash join. " +
- "Broadcast will be
disabled if the value is -1.");
+ .withDescription("Configures the
maximum size in bytes for a table that will be broadcast to all worker " +
+ "nodes when performing
a join. By setting this value to -1 broadcasting can be disabled. ");
+ @Documentation.ExcludeFromDocumentation(value = "Just for corner cases,
when the cbo is totally wrong, " +
+ "we can adjust this value to reduce useless
computation.")
public static final ConfigOption<Double>
SQL_OPTIMIZER_SEMI_JOIN_BUILD_DISTINCT_NDV_RATIO =
key("sql.optimizer.semi-anti-join.build-distinct.ndv-ratio")
.defaultValue(0.8)
- .withDescription("When the semi-side of
semi/anti join can distinct a lot of data in advance," +
- " we will add distinct
node before semi/anti join.");
-
+ .withDescription("In order to reduce
the amount of data on semi/anti join's" +
+ " build side, we will
add distinct node before semi/anti join when" +
+ " the semi-side or
semi/anti join can distinct a lot of data in advance." +
+ " We add this
configuration to help the optimizer to decide whether to" +
+ " add the distinct.");
+
+ @Documentation.ExcludeFromDocumentation(value = "Just for corner cases,
when the cbo is totally wrong, " +
+ "we can adjust this value to reduce useless filter.")
public static final ConfigOption<Long>
SQL_OPTIMIZER_JOIN_NULL_FILTER_THRESHOLD =
key("sql.optimizer.join.null.filter.threshold")
Review comment:
ql.optimizer.join.null-filter.threshold ?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
