Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[MINOR] improvement(spark-client): put sparkConf as extra properties while client request accessCluster #2254

Open
wants to merge 6 commits into
base: master
Choose a base branch
from
+41 −15
Open

[MINOR] improvement(spark-client): put sparkConf as extra properties while client request accessCluster #2254

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
827ee04
Put sparkConf as extra properties while client request accessCluster
maobaolong Nov 18, 2024
f538b56
Address suggestion, remove the spark prefix to be common
maobaolong Nov 20, 2024
9b01609
Add a exclude properties config
maobaolong Nov 25, 2024
1c93abf
Set default value
maobaolong Nov 25, 2024
9a3b25d
Remove weired config value to keep empty list by default
maobaolong Nov 25, 2024
8d3f2ac
Remove larger than 100 filter
maobaolong Nov 25, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 13 additions & 0 deletions client-spark/common/src/main/java/org/apache/spark/shuffle/RssSparkConfig.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,8 @@

package org.apache.spark.shuffle;

import java.util.HashMap;
import java.util.Map;
import java.util.Set;

import scala.Tuple2;
Expand Down Expand Up @@ -512,4 +514,15 @@ public static RssConf toRssConf(SparkConf sparkConf) {
}
return rssConf;
}

public static Map<String, String> sparkConfToMap(SparkConf sparkConf) {
Map<String, String> map = new HashMap<>();

for (Tuple2<String, String> tuple : sparkConf.getAll()) {
String key = tuple._1;
map.put(key, tuple._2);
}

return map;
}
}
17 changes: 2 additions & 15 deletions ...-spark/common/src/main/java/org/apache/uniffle/shuffle/manager/RssShuffleManagerBase.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,8 +34,6 @@
import java.util.function.Supplier;
import java.util.stream.Collectors;

import scala.Tuple2;

import com.google.common.annotations.VisibleForTesting;
import com.google.common.collect.Maps;
import com.google.common.collect.Sets;
Expand Down Expand Up @@ -1064,7 +1062,7 @@ protected void registerShuffleServers(
}
LOG.info("Start to register shuffleId {}", shuffleId);
long start = System.currentTimeMillis();
Map<String, String> sparkConfMap = sparkConfToMap(getSparkConf());
Map<String, String> sparkConfMap = RssSparkConfig.sparkConfToMap(getSparkConf());
serverToPartitionRanges.entrySet().stream()
.forEach(
entry -> {
Expand Down Expand Up @@ -1095,7 +1093,7 @@ protected void registerShuffleServers(
}
LOG.info("Start to register shuffleId[{}]", shuffleId);
long start = System.currentTimeMillis();
Map<String, String> sparkConfMap = sparkConfToMap(getSparkConf());
Map<String, String> sparkConfMap = RssSparkConfig.sparkConfToMap(getSparkConf());
Set<Map.Entry<ShuffleServerInfo, List<PartitionRange>>> entries =
serverToPartitionRanges.entrySet();
entries.stream()
Expand Down Expand Up @@ -1141,15 +1139,4 @@ public boolean isRssStageRetryForFetchFailureEnabled() {
public SparkConf getSparkConf() {
return sparkConf;
}

public Map<String, String> sparkConfToMap(SparkConf sparkConf) {
Map<String, String> map = new HashMap<>();

for (Tuple2<String, String> tuple : sparkConf.getAll()) {
String key = tuple._1;
map.put(key, tuple._2);
}

return map;
}
}
13 changes: 13 additions & 0 deletions client-spark/spark2/src/main/java/org/apache/spark/shuffle/DelegationRssShuffleManager.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,7 @@

package org.apache.spark.shuffle;

import java.util.List;
import java.util.Map;
import java.util.Set;

Expand All @@ -32,6 +33,8 @@
import org.apache.uniffle.client.impl.grpc.CoordinatorGrpcRetryableClient;
import org.apache.uniffle.client.request.RssAccessClusterRequest;
import org.apache.uniffle.client.response.RssAccessClusterResponse;
import org.apache.uniffle.common.config.RssClientConf;
import org.apache.uniffle.common.config.RssConf;
import org.apache.uniffle.common.exception.RssException;
import org.apache.uniffle.common.rpc.StatusCode;
import org.apache.uniffle.common.util.Constants;
Expand Down Expand Up @@ -124,6 +127,16 @@ private boolean tryAccessCluster() {
extraProperties.put(
ACCESS_INFO_REQUIRED_SHUFFLE_NODES_NUM, String.valueOf(assignmentShuffleNodesNum));

RssConf rssConf = RssSparkConfig.toRssConf(sparkConf);
List<String> excludeProperties =
rssConf.get(RssClientConf.RSS_CLIENT_REPORT_EXCLUDE_PROPERTIES);
// Put all spark conf into extra properties, except which length is longer than 100
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should you remove the comment?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done

// to avoid extra properties too long.
rssConf.getAll().stream()
.filter(entry -> StringUtils.length((String) entry.getValue()) < 100)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should you remove <100?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done

.filter(entry -> !excludeProperties.contains(entry.getKey()))
.forEach(entry -> extraProperties.put(entry.getKey(), (String) entry.getValue()));

Set<String> assignmentTags = RssSparkShuffleUtils.getAssignmentTags(sparkConf);
try {
if (coordinatorClient != null) {
Expand Down
13 changes: 13 additions & 0 deletions client-spark/spark3/src/main/java/org/apache/spark/shuffle/DelegationRssShuffleManager.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,7 @@

package org.apache.spark.shuffle;

import java.util.List;
import java.util.Map;
import java.util.Set;

Expand All @@ -32,6 +33,8 @@
import org.apache.uniffle.client.impl.grpc.CoordinatorGrpcRetryableClient;
import org.apache.uniffle.client.request.RssAccessClusterRequest;
import org.apache.uniffle.client.response.RssAccessClusterResponse;
import org.apache.uniffle.common.config.RssClientConf;
import org.apache.uniffle.common.config.RssConf;
import org.apache.uniffle.common.exception.RssException;
import org.apache.uniffle.common.rpc.StatusCode;
import org.apache.uniffle.common.util.Constants;
Expand Down Expand Up @@ -124,6 +127,16 @@ private boolean tryAccessCluster() {
extraProperties.put(
ACCESS_INFO_REQUIRED_SHUFFLE_NODES_NUM, String.valueOf(assignmentShuffleNodesNum));

RssConf rssConf = RssSparkConfig.toRssConf(sparkConf);
List<String> excludeProperties =
rssConf.get(RssClientConf.RSS_CLIENT_REPORT_EXCLUDE_PROPERTIES);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The default is null? Will it cause an error?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for your tip, I set a default value.

// Put all spark conf into extra properties, except which length is longer than 100
// to avoid extra properties too long.
rssConf.getAll().stream()
.filter(entry -> StringUtils.length((String) entry.getValue()) < 100)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is too tricky....

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Em... Thanks for your review! I'm sorry for this.

I would greatly appreciate it if you could give me some advice.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can implement a white list or configuration filter. The white list can be configured in the configuration. The configuration filter could have length configuration filter. The length could be configured in the configuration.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do you think use black list instead of white list can be possible?

Copy link
Contributor

@jerqi jerqi Nov 25, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok for me, too. black list would better to be replaced by excludeList.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

done

.filter(entry -> !excludeProperties.contains(entry.getKey()))
.forEach(entry -> extraProperties.put(entry.getKey(), (String) entry.getValue()));

Set<String> assignmentTags = RssSparkShuffleUtils.getAssignmentTags(sparkConf);
try {
if (coordinatorClient != null) {
Expand Down
6 changes: 6 additions & 0 deletions common/src/main/java/org/apache/uniffle/common/config/RssClientConf.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -303,4 +303,10 @@ public class RssClientConf {
.withDescription(
"The block id manager class of server for this application, "
+ "the implementation of this interface to manage the shuffle block ids");
public static final ConfigOption<List<String>> RSS_CLIENT_REPORT_EXCLUDE_PROPERTIES =
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Add a blank line.

ConfigOptions.key("rss.client.reportExcludeProperties")
.stringType()
.asList()
.defaultValues()
.withDescription("the report exclude properties could be configured by this option");
}
Loading