Skip to content

Commit

Permalink
[Backport 2.x] Core and Plugin changes for query-level resource usage…
Browse files Browse the repository at this point in the history
…s tracking (#14085)

* Query-level resource usages tracking (#13172)

* Query-level resource usages tracking

Signed-off-by: Chenyang Ji <cyji@amazon.com>

* Moving TaskResourceTrackingService to clusterService

Signed-off-by: Chenyang Ji <cyji@amazon.com>

* use shard response header to piggyback task resource usages

Signed-off-by: Chenyang Ji <cyji@amazon.com>

* split changes for query insights plugin

Signed-off-by: Chenyang Ji <cyji@amazon.com>

* improve the supplier logic and other misc items

Signed-off-by: Chenyang Ji <cyji@amazon.com>

* track resource usage for failed requests

Signed-off-by: Chenyang Ji <cyji@amazon.com>

* move resource usages interactions into TaskResourceTrackingService

Signed-off-by: Chenyang Ji <cyji@amazon.com>

---------

Signed-off-by: Chenyang Ji <cyji@amazon.com>
(cherry picked from commit 3d1fa98)

* fix concurrent modification issue in thread context (#14084)

Signed-off-by: Chenyang Ji <cyji@amazon.com>
(cherry picked from commit c8f0b6d)

* consume query level cpu and memory usage in query insights (#13739)

* consume query level cpu and memory usage in query insights

Signed-off-by: Chenyang Ji <cyji@amazon.com>

* handle failed requests metrics in query insights

Signed-off-by: Chenyang Ji <cyji@amazon.com>

* refactor the code to make it more maintainable

Signed-off-by: Chenyang Ji <cyji@amazon.com>

---------

Signed-off-by: Chenyang Ji <cyji@amazon.com>
(cherry picked from commit 04a417a)

* fix japicmp check for threadContext

Signed-off-by: Chenyang Ji <cyji@amazon.com>
(cherry picked from commit b403fdc)
  • Loading branch information
ansjcy committed Jun 10, 2024
1 parent faa6007 commit 0634ccd
Show file tree
Hide file tree
Showing 44 changed files with 1,200 additions and 164 deletions.
2 changes: 2 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
- Move Remote Store Migration from DocRep to GA and modify remote migration settings name ([#14100](https://github.com/opensearch-project/OpenSearch/pull/14100))
- [Remote State] Add async remote state deletion task running on an interval, configurable by a setting ([#13995](https://github.com/opensearch-project/OpenSearch/pull/13995))
- Add remote routing table for remote state publication with experimental feature flag ([#13304](https://github.com/opensearch-project/OpenSearch/pull/13304))
- Add support for query level resource usage tracking ([#13172](https://github.com/opensearch-project/OpenSearch/pull/13172))
- [Query Insights] Add cpu and memory metrics to top n queries ([#13739](https://github.com/opensearch-project/OpenSearch/pull/13739))

### Dependencies
- Bump `com.github.spullara.mustache.java:compiler` from 0.9.10 to 0.9.13 ([#13329](https://github.com/opensearch-project/OpenSearch/pull/13329), [#13559](https://github.com/opensearch-project/OpenSearch/pull/13559))
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -104,6 +104,10 @@ public long getTotalValue() {
return endValue.get() - startValue;
}

public long getStartValue() {
return startValue;
}

@Override
public String toString() {
return String.valueOf(getTotalValue());
Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,225 @@
/*
* SPDX-License-Identifier: Apache-2.0
*
* The OpenSearch Contributors require contributions made to
* this file be licensed under the Apache-2.0 license or a
* compatible open source license.
*/

package org.opensearch.core.tasks.resourcetracker;

import org.opensearch.common.annotation.PublicApi;
import org.opensearch.core.ParseField;
import org.opensearch.core.common.Strings;
import org.opensearch.core.common.io.stream.StreamInput;
import org.opensearch.core.common.io.stream.StreamOutput;
import org.opensearch.core.common.io.stream.Writeable;
import org.opensearch.core.xcontent.ConstructingObjectParser;
import org.opensearch.core.xcontent.MediaTypeRegistry;
import org.opensearch.core.xcontent.ToXContentObject;
import org.opensearch.core.xcontent.XContentBuilder;

import java.io.IOException;
import java.util.Objects;

import static org.opensearch.core.xcontent.ConstructingObjectParser.constructorArg;

/**
* Task resource usage information with minimal information about the task
* <p>
* Writeable TaskResourceInfo objects are used to represent resource usage
* information of running tasks, which can be propagated to coordinator node
* to infer query-level resource usage
*
* @opensearch.api
*/
@PublicApi(since = "2.15.0")
public class TaskResourceInfo implements Writeable, ToXContentObject {
private final String action;
private final long taskId;
private final long parentTaskId;
private final String nodeId;
private final TaskResourceUsage taskResourceUsage;

private static final ParseField ACTION = new ParseField("action");
private static final ParseField TASK_ID = new ParseField("taskId");
private static final ParseField PARENT_TASK_ID = new ParseField("parentTaskId");
private static final ParseField NODE_ID = new ParseField("nodeId");
private static final ParseField TASK_RESOURCE_USAGE = new ParseField("taskResourceUsage");

public TaskResourceInfo(
final String action,
final long taskId,
final long parentTaskId,
final String nodeId,
final TaskResourceUsage taskResourceUsage
) {
this.action = action;
this.taskId = taskId;
this.parentTaskId = parentTaskId;
this.nodeId = nodeId;
this.taskResourceUsage = taskResourceUsage;
}

public static final ConstructingObjectParser<TaskResourceInfo, Void> PARSER = new ConstructingObjectParser<>(
"task_resource_info",
a -> new Builder().setAction((String) a[0])
.setTaskId((Long) a[1])
.setParentTaskId((Long) a[2])
.setNodeId((String) a[3])
.setTaskResourceUsage((TaskResourceUsage) a[4])
.build()
);

static {
PARSER.declareString(constructorArg(), ACTION);
PARSER.declareLong(constructorArg(), TASK_ID);
PARSER.declareLong(constructorArg(), PARENT_TASK_ID);
PARSER.declareString(constructorArg(), NODE_ID);
PARSER.declareObject(constructorArg(), TaskResourceUsage.PARSER, TASK_RESOURCE_USAGE);
}

@Override
public XContentBuilder toXContent(XContentBuilder builder, Params params) throws IOException {
builder.startObject();
builder.field(ACTION.getPreferredName(), this.action);
builder.field(TASK_ID.getPreferredName(), this.taskId);
builder.field(PARENT_TASK_ID.getPreferredName(), this.parentTaskId);
builder.field(NODE_ID.getPreferredName(), this.nodeId);
builder.startObject(TASK_RESOURCE_USAGE.getPreferredName());
this.taskResourceUsage.toXContent(builder, params);
builder.endObject();
builder.endObject();
return builder;
}

/**
* Builder for {@link TaskResourceInfo}
*/
public static class Builder {
private TaskResourceUsage taskResourceUsage;
private String action;
private long taskId;
private long parentTaskId;
private String nodeId;

public Builder setTaskResourceUsage(final TaskResourceUsage taskResourceUsage) {
this.taskResourceUsage = taskResourceUsage;
return this;
}

public Builder setAction(final String action) {
this.action = action;
return this;
}

public Builder setTaskId(final long taskId) {
this.taskId = taskId;
return this;
}

public Builder setParentTaskId(final long parentTaskId) {
this.parentTaskId = parentTaskId;
return this;
}

public Builder setNodeId(final String nodeId) {
this.nodeId = nodeId;
return this;
}

public TaskResourceInfo build() {
return new TaskResourceInfo(action, taskId, parentTaskId, nodeId, taskResourceUsage);
}
}

/**
* Read task info from a stream.
*
* @param in StreamInput to read
* @return {@link TaskResourceInfo}
* @throws IOException IOException
*/
public static TaskResourceInfo readFromStream(StreamInput in) throws IOException {
return new TaskResourceInfo.Builder().setAction(in.readString())
.setTaskId(in.readLong())
.setParentTaskId(in.readLong())
.setNodeId(in.readString())
.setTaskResourceUsage(TaskResourceUsage.readFromStream(in))
.build();
}

/**
* Get TaskResourceUsage
*
* @return taskResourceUsage
*/
public TaskResourceUsage getTaskResourceUsage() {
return taskResourceUsage;
}

/**
* Get parent task id
*
* @return parent task id
*/
public long getParentTaskId() {
return parentTaskId;
}

/**
* Get task id
* @return task id
*/
public long getTaskId() {
return taskId;
}

/**
* Get node id
* @return node id
*/
public String getNodeId() {
return nodeId;
}

/**
* Get task action
* @return task action
*/
public String getAction() {
return action;
}

@Override
public void writeTo(StreamOutput out) throws IOException {
out.writeString(action);
out.writeLong(taskId);
out.writeLong(parentTaskId);
out.writeString(nodeId);
taskResourceUsage.writeTo(out);
}

@Override
public String toString() {
return Strings.toString(MediaTypeRegistry.JSON, this);
}

@Override
public boolean equals(Object obj) {
if (obj == null || obj.getClass() != TaskResourceInfo.class) {
return false;
}
TaskResourceInfo other = (TaskResourceInfo) obj;
return action.equals(other.action)
&& taskId == other.taskId
&& parentTaskId == other.parentTaskId
&& Objects.equals(nodeId, other.nodeId)
&& taskResourceUsage.equals(other.taskResourceUsage);
}

@Override
public int hashCode() {
return Objects.hash(action, taskId, parentTaskId, nodeId, taskResourceUsage);
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -111,7 +111,15 @@ public List<Setting<?>> getSettings() {
QueryInsightsSettings.TOP_N_LATENCY_QUERIES_ENABLED,
QueryInsightsSettings.TOP_N_LATENCY_QUERIES_SIZE,
QueryInsightsSettings.TOP_N_LATENCY_QUERIES_WINDOW_SIZE,
QueryInsightsSettings.TOP_N_LATENCY_EXPORTER_SETTINGS
QueryInsightsSettings.TOP_N_LATENCY_EXPORTER_SETTINGS,
QueryInsightsSettings.TOP_N_CPU_QUERIES_ENABLED,
QueryInsightsSettings.TOP_N_CPU_QUERIES_SIZE,
QueryInsightsSettings.TOP_N_CPU_QUERIES_WINDOW_SIZE,
QueryInsightsSettings.TOP_N_CPU_EXPORTER_SETTINGS,
QueryInsightsSettings.TOP_N_MEMORY_QUERIES_ENABLED,
QueryInsightsSettings.TOP_N_MEMORY_QUERIES_SIZE,
QueryInsightsSettings.TOP_N_MEMORY_QUERIES_WINDOW_SIZE,
QueryInsightsSettings.TOP_N_MEMORY_EXPORTER_SETTINGS
);
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@
import java.util.Locale;
import java.util.Set;

import static org.opensearch.plugin.insights.settings.QueryInsightsSettings.DEFAULT_TOP_N_LATENCY_QUERIES_INDEX_PATTERN;
import static org.opensearch.plugin.insights.settings.QueryInsightsSettings.DEFAULT_TOP_N_QUERIES_INDEX_PATTERN;
import static org.opensearch.plugin.insights.settings.QueryInsightsSettings.DEFAULT_TOP_QUERIES_EXPORTER_TYPE;
import static org.opensearch.plugin.insights.settings.QueryInsightsSettings.EXPORTER_TYPE;
import static org.opensearch.plugin.insights.settings.QueryInsightsSettings.EXPORT_INDEX;
Expand Down Expand Up @@ -71,7 +71,7 @@ public void validateExporterConfig(final Settings settings) throws IllegalArgume
}
switch (type) {
case LOCAL_INDEX:
final String indexPattern = settings.get(EXPORT_INDEX, DEFAULT_TOP_N_LATENCY_QUERIES_INDEX_PATTERN);
final String indexPattern = settings.get(EXPORT_INDEX, DEFAULT_TOP_N_QUERIES_INDEX_PATTERN);
if (indexPattern.length() == 0) {
throw new IllegalArgumentException("Empty index pattern configured for the exporter");
}
Expand Down
Loading

0 comments on commit 0634ccd

Please sign in to comment.