Add a checkpoint service for data frame transforms (#39836)

Add a checkpoint service for data frame transforms, which allows to ask for a checkpoint of the
source. In future these checkpoints will be stored in the internal index to

 - detect upstream changes
 - updating the data frame without a full re-run
 - allow data frame clients to checkpoint themselves

x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/dataframe/DataFrameMessages.java

@@ -53,6 +53,9 @@ public class DataFrameMessages {
     public static final String LOG_DATA_FRAME_TRANSFORM_CONFIGURATION_BAD_AGGREGATION =
             "Failed to parse aggregation for data frame pivot transform";
 
+    public static final String FAILED_TO_PARSE_TRANSFORM_CHECKPOINTS =
+            "Failed to parse transform checkpoints for [{0}]";
+
     private DataFrameMessages() {
     }
 

x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/dataframe/transforms/DataFrameTransformCheckpoint.java

@@ -0,0 +1,264 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the Elastic License;
+ * you may not use this file except in compliance with the Elastic License.
+ */
+
+package org.elasticsearch.xpack.core.dataframe.transforms;
+
+import org.elasticsearch.ElasticsearchParseException;
+import org.elasticsearch.common.ParseField;
+import org.elasticsearch.common.ParsingException;
+import org.elasticsearch.common.io.stream.StreamInput;
+import org.elasticsearch.common.io.stream.StreamOutput;
+import org.elasticsearch.common.io.stream.Writeable;
+import org.elasticsearch.common.xcontent.ConstructingObjectParser;
+import org.elasticsearch.common.xcontent.ToXContentObject;
+import org.elasticsearch.common.xcontent.XContentBuilder;
+import org.elasticsearch.common.xcontent.XContentParser;
+import org.elasticsearch.xpack.core.dataframe.DataFrameField;
+
+import java.io.IOException;
+import java.util.Arrays;
+import java.util.Collections;
+import java.util.Map;
+import java.util.Map.Entry;
+import java.util.Objects;
+import java.util.TreeMap;
+
+import static org.elasticsearch.common.xcontent.ConstructingObjectParser.constructorArg;
+import static org.elasticsearch.common.xcontent.ConstructingObjectParser.optionalConstructorArg;
+
+/**
+ * Checkpoint document to store the checkpoint of a data frame transform
+ *
+ * The fields:
+ *
+ *  timestamp the timestamp when this document has been created
+ *  checkpoint the checkpoint number, incremented for every checkpoint
+ *  indices a map of the indices from the source including all checkpoints of all indices matching the source pattern, shard level
+ *  time_upper_bound for time-based indices this holds the upper time boundary of this checkpoint
+ *
+ */
+public class DataFrameTransformCheckpoint implements Writeable, ToXContentObject {
+
+    public static DataFrameTransformCheckpoint EMPTY = new DataFrameTransformCheckpoint("empty", 0L, -1L, Collections.emptyMap(), 0L);
+
+    // the timestamp of the checkpoint, mandatory
+    public static final ParseField TIMESTAMP_MILLIS = new ParseField("timestamp_millis");
+    public static final ParseField TIMESTAMP = new ParseField("timestamp");
+
+    // the own checkpoint
+    public static final ParseField CHECKPOINT = new ParseField("checkpoint");
+
+    // checkpoint of the indexes (sequence id's)
+    public static final ParseField INDICES = new ParseField("indices");
+
+    // checkpoint for for time based sync
+    // TODO: consider a lower bound for usecases where you want to transform on a window of a stream
+    public static final ParseField TIME_UPPER_BOUND_MILLIS = new ParseField("time_upper_bound_millis");
+    public static final ParseField TIME_UPPER_BOUND = new ParseField("time_upper_bound");
+
+    private static final String NAME = "data_frame_transform_checkpoint";
+
+    private static final ConstructingObjectParser<DataFrameTransformCheckpoint, Void> STRICT_PARSER = createParser(false);
+    private static final ConstructingObjectParser<DataFrameTransformCheckpoint, Void> LENIENT_PARSER = createParser(true);
+
+    private final String transformId;
+    private final long timestampMillis;
+    private final long checkpoint;
+    private final Map<String, long[]> indicesCheckpoints;
+    private final long timeUpperBoundMillis;
+
+    private static ConstructingObjectParser<DataFrameTransformCheckpoint, Void> createParser(boolean lenient) {
+        ConstructingObjectParser<DataFrameTransformCheckpoint, Void> parser = new ConstructingObjectParser<>(NAME,
+                lenient, args -> {
+                    String id = (String) args[0];
+                    Long timestamp = (Long) args[1];
+                    Long checkpoint = (Long) args[2];
+
+                    @SuppressWarnings("unchecked")
+                    Map<String, long[]> checkpoints = (Map<String, long[]>) args[3];
+
+                    Long timestamp_checkpoint = (Long) args[4];
+
+                    // ignored, only for internal storage: String docType = (String) args[5];
+                    return new DataFrameTransformCheckpoint(id, timestamp, checkpoint, checkpoints, timestamp_checkpoint);
+                });
+
+        parser.declareString(constructorArg(), DataFrameField.ID);
+
+        // note: this is never parsed from the outside where timestamp can be formatted as date time
+        parser.declareLong(constructorArg(), TIMESTAMP_MILLIS);
+        parser.declareLong(constructorArg(), CHECKPOINT);
+
+        parser.declareObject(constructorArg(), (p,c) -> {
+            Map<String, long[]> checkPointsByIndexName = new TreeMap<>();
+            XContentParser.Token token = null;
+            while ((token = p.nextToken()) != XContentParser.Token.END_OBJECT) {
+                if (token != XContentParser.Token.FIELD_NAME) {
+                    throw new ParsingException(p.getTokenLocation(), "Unexpected token " + token + " ");
+                }
+
+                final String indexName = p.currentName();
+                token = p.nextToken();
+                if (token != XContentParser.Token.START_ARRAY) {
+                    throw new ParsingException(p.getTokenLocation(), "Unexpected token " + token + " ");
+                }
+
+                long[] checkpoints = p.listOrderedMap().stream().mapToLong(num -> ((Number) num).longValue()).toArray();
+                checkPointsByIndexName.put(indexName, checkpoints);
+            }
+            return checkPointsByIndexName;
+        }, INDICES);
+        parser.declareLong(optionalConstructorArg(), TIME_UPPER_BOUND_MILLIS);
+        parser.declareString(optionalConstructorArg(), DataFrameField.INDEX_DOC_TYPE);
+
+        return parser;
+    }
+
+    public DataFrameTransformCheckpoint(String transformId, Long timestamp, Long checkpoint, Map<String, long[]> checkpoints,
+            Long timeUpperBound) {
+        this.transformId = transformId;
+        this.timestampMillis = timestamp.longValue();
+        this.checkpoint = checkpoint;
+        this.indicesCheckpoints = Collections.unmodifiableMap(checkpoints);
+        this.timeUpperBoundMillis = timeUpperBound == null ? 0 : timeUpperBound.longValue();
+    }
+
+    public DataFrameTransformCheckpoint(StreamInput in) throws IOException {
+        this.transformId = in.readString();
+        this.timestampMillis = in.readLong();
+        this.checkpoint = in.readLong();
+        this.indicesCheckpoints = readCheckpoints(in.readMap());
+        this.timeUpperBoundMillis = in.readLong();
+    }
+
+    @Override
+    public XContentBuilder toXContent(XContentBuilder builder, Params params) throws IOException {
+        builder.startObject();
+
+        // the id, doc_type and checkpoint is only internally used for storage, the user-facing version gets embedded
+        if (params.paramAsBoolean(DataFrameField.FOR_INTERNAL_STORAGE, false)) {
+            builder.field(DataFrameField.ID.getPreferredName(), transformId);
+            builder.field(CHECKPOINT.getPreferredName(), checkpoint);
+            builder.field(DataFrameField.INDEX_DOC_TYPE.getPreferredName(), NAME);
+        }
+
+        builder.timeField(TIMESTAMP_MILLIS.getPreferredName(), TIMESTAMP.getPreferredName(), timestampMillis);
+
+        if (timeUpperBoundMillis > 0) {
+            builder.timeField(TIME_UPPER_BOUND_MILLIS.getPreferredName(), TIME_UPPER_BOUND.getPreferredName(), timeUpperBoundMillis);
+        }
+
+        builder.startObject(INDICES.getPreferredName());
+        for (Entry<String, long[]> entry : indicesCheckpoints.entrySet()) {
+            builder.array(entry.getKey(), entry.getValue());
+        }
+        builder.endObject();
+
+        builder.endObject();
+        return builder;
+    }
+
+    public String getTransformId() {
+        return transformId;
+    }
+
+    public long getTimestamp() {
+        return timestampMillis;
+    }
+
+    public long getCheckpoint() {
+        return checkpoint;
+    }
+
+    public Map<String, long[]> getIndicesCheckpoints() {
+        return indicesCheckpoints;
+    }
+
+    public long getTimeUpperBound() {
+        return timeUpperBoundMillis;
+    }
+
+    @Override
+    public void writeTo(StreamOutput out) throws IOException {
+        out.writeString(transformId);
+        out.writeLong(timestampMillis);
+        out.writeLong(checkpoint);
+        out.writeGenericValue(indicesCheckpoints);
+        out.writeLong(timeUpperBoundMillis);
+    }
+
+    @Override
+    public boolean equals(Object other) {
+        if (this == other) {
+            return true;
+        }
+
+        if (other == null || getClass() != other.getClass()) {
+            return false;
+        }
+
+        final DataFrameTransformCheckpoint that = (DataFrameTransformCheckpoint) other;
+
+        // compare the timestamp, id, checkpoint and than call matches for the rest
+        return this.timestampMillis == that.timestampMillis && this.checkpoint == that.checkpoint
+                && this.timeUpperBoundMillis == that.timeUpperBoundMillis && matches(that);
+    }
+
+    /**
+     * Compares 2 checkpoints ignoring some inner fields.
+     *
+     * This is for comparing 2 checkpoints to check whether the data frame transform requires an update
+     *
+     * @param that other checkpoint
+     * @return true if checkpoints match
+     */
+    public boolean matches (DataFrameTransformCheckpoint that) {
+        if (this == that) {
+            return true;
+        }
+
+        return Objects.equals(this.transformId, that.transformId)
+                && this.indicesCheckpoints.size() == that.indicesCheckpoints.size() // quick check
+                // do the expensive deep equal operation last
+                && this.indicesCheckpoints.entrySet().stream()
+                        .allMatch(e -> Arrays.equals(e.getValue(), that.indicesCheckpoints.get(e.getKey())));
+    }
+
+    @Override
+    public int hashCode() {
+        int hash = Objects.hash(transformId, timestampMillis, checkpoint, timeUpperBoundMillis);
+
+        for (Entry<String, long[]> e : indicesCheckpoints.entrySet()) {
+            hash = 31 * hash + Objects.hash(e.getKey(), Arrays.hashCode(e.getValue()));
+        }
+        return hash;
+    }
+
+    public static DataFrameTransformCheckpoint fromXContent(final XContentParser parser, boolean lenient) throws IOException {
+        return lenient ? LENIENT_PARSER.apply(parser, null) : STRICT_PARSER.apply(parser, null);
+    }
+
+    public static String documentId(String transformId, long checkpoint) {
+        if (checkpoint < 0) {
+            throw new IllegalArgumentException("checkpoint must be a positive number");
+        }
+
+        return NAME + "-" + transformId + "-" + checkpoint;
+    }
+
+    private static Map<String, long[]> readCheckpoints(Map<String, Object> readMap) {
+        Map<String, long[]> checkpoints = new TreeMap<>();
+        for (Map.Entry<String, Object> e : readMap.entrySet()) {
+            if (e.getValue() instanceof long[]) {
+                checkpoints.put(e.getKey(), (long[]) e.getValue());
+            } else {
+                throw new ElasticsearchParseException("expecting the checkpoints for [{}] to be a long[], but found [{}] instead",
+                        e.getKey(), e.getValue().getClass());
+            }
+        }
+        return checkpoints;
+    }
+}

x-pack/plugin/core/src/test/java/org/elasticsearch/xpack/core/dataframe/transforms/AbstractSerializingDataFrameTestCase.java

@@ -12,13 +12,16 @@ import org.elasticsearch.common.io.stream.Writeable;
 import org.elasticsearch.common.settings.Settings;
 import org.elasticsearch.common.xcontent.NamedXContentRegistry;
 import org.elasticsearch.common.xcontent.ToXContent;
+import org.elasticsearch.common.xcontent.ToXContent.Params;
 import org.elasticsearch.index.query.QueryBuilder;
 import org.elasticsearch.search.SearchModule;
 import org.elasticsearch.search.aggregations.AggregationBuilder;
 import org.elasticsearch.search.aggregations.BaseAggregationBuilder;
 import org.elasticsearch.test.AbstractSerializingTestCase;
+import org.elasticsearch.xpack.core.dataframe.DataFrameField;
 import org.junit.Before;
 
+import java.util.Collections;
 import java.util.List;
 
 import static java.util.Collections.emptyList;
@@ -26,6 +29,9 @@ import static java.util.Collections.emptyList;
 public abstract class AbstractSerializingDataFrameTestCase<T extends ToXContent & Writeable>
         extends AbstractSerializingTestCase<T> {
 
+    protected static Params TO_XCONTENT_PARAMS = new ToXContent.MapParams(
+            Collections.singletonMap(DataFrameField.FOR_INTERNAL_STORAGE, "true"));
+
     /**
      * Test case that ensures aggregation named objects are registered
      */

x-pack/plugin/core/src/test/java/org/elasticsearch/xpack/core/dataframe/transforms/DataFrameTransformCheckpointTests.java

@@ -0,0 +1,133 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the Elastic License;
+ * you may not use this file except in compliance with the Elastic License.
+ */
+
+package org.elasticsearch.xpack.core.dataframe.transforms;
+
+import org.elasticsearch.common.Strings;
+import org.elasticsearch.common.io.stream.Writeable.Reader;
+import org.elasticsearch.common.xcontent.ToXContent;
+import org.elasticsearch.common.xcontent.XContentBuilder;
+import org.elasticsearch.common.xcontent.XContentFactory;
+import org.elasticsearch.common.xcontent.XContentParser;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.List;
+import java.util.Map;
+import java.util.TreeMap;
+
+import static org.elasticsearch.test.TestMatchers.matchesPattern;
+
+public class DataFrameTransformCheckpointTests extends AbstractSerializingDataFrameTestCase<DataFrameTransformCheckpoint> {
+
+    public static DataFrameTransformCheckpoint randomDataFrameTransformCheckpoints() {
+        return new DataFrameTransformCheckpoint(randomAlphaOfLengthBetween(1, 10), randomNonNegativeLong(), randomNonNegativeLong(),
+                randomCheckpointsByIndex(), randomNonNegativeLong());
+    }
+
+    @Override
+    protected DataFrameTransformCheckpoint doParseInstance(XContentParser parser) throws IOException {
+        return DataFrameTransformCheckpoint.fromXContent(parser, false);
+    }
+
+    @Override
+    protected DataFrameTransformCheckpoint createTestInstance() {
+        return randomDataFrameTransformCheckpoints();
+    }
+
+    @Override
+    protected Reader<DataFrameTransformCheckpoint> instanceReader() {
+        return DataFrameTransformCheckpoint::new;
+    }
+
+    @Override
+    protected ToXContent.Params getToXContentParams() {
+        return TO_XCONTENT_PARAMS;
+    }
+
+    public void testXContentForInternalStorage() throws IOException {
+        DataFrameTransformCheckpoint dataFrameTransformCheckpoints = randomDataFrameTransformCheckpoints();
+
+        try (XContentBuilder xContentBuilder = XContentFactory.jsonBuilder()) {
+            XContentBuilder content = dataFrameTransformCheckpoints.toXContent(xContentBuilder, getToXContentParams());
+            String doc = Strings.toString(content);
+
+            assertThat(doc, matchesPattern(".*\"doc_type\"\\s*:\\s*\"data_frame_transform_checkpoint\".*"));
+        }
+
+        try (XContentBuilder xContentBuilder = XContentFactory.jsonBuilder()) {
+            XContentBuilder content = dataFrameTransformCheckpoints.toXContent(xContentBuilder, ToXContent.EMPTY_PARAMS);
+            String doc = Strings.toString(content);
+
+            assertFalse(doc.contains("doc_type"));
+        }
+    }
+
+    public void testXContentForApiUsage() throws IOException {
+        DataFrameTransformCheckpoint dataFrameTransformCheckpoints = new DataFrameTransformCheckpoint(randomAlphaOfLengthBetween(1, 10),
+                1546300800000L, randomNonNegativeLong(), Collections.emptyMap(), 1545609600000L);
+
+        try (XContentBuilder xContentBuilder = XContentFactory.jsonBuilder()) {
+            xContentBuilder.humanReadable(true);
+            XContentBuilder content = dataFrameTransformCheckpoints.toXContent(xContentBuilder, ToXContent.EMPTY_PARAMS);
+            String doc = Strings.toString(content);
+            assertThat(doc, matchesPattern(".*\"timestamp_millis\"\\s*:\\s*1546300800000.*"));
+            assertThat(doc, matchesPattern(".*\"time_upper_bound_millis\"\\s*:\\s*1545609600000.*"));
+            assertThat(doc, matchesPattern(".*\"timestamp\"\\s*:\\s*\"2019-01-01T00:00:00.000Z\".*"));
+            assertThat(doc, matchesPattern(".*\"time_upper_bound\"\\s*:\\s*\"2018-12-24T00:00:00.000Z\".*"));
+        }
+    }
+
+    public void testMatches() throws IOException {
+        String id = randomAlphaOfLengthBetween(1, 10);
+        long timestamp = randomNonNegativeLong();
+        long checkpoint = randomNonNegativeLong();
+        Map<String, long[]> checkpointsByIndex = randomCheckpointsByIndex();
+        Map<String, long[]> otherCheckpointsByIndex = new TreeMap<>(checkpointsByIndex);
+        otherCheckpointsByIndex.put(randomAlphaOfLengthBetween(1, 10), new long[] { 1, 2, 3 });
+        long timeUpperBound = randomNonNegativeLong();
+
+        DataFrameTransformCheckpoint dataFrameTransformCheckpoints = new DataFrameTransformCheckpoint(id, timestamp, checkpoint,
+                checkpointsByIndex, timeUpperBound);
+
+        // same
+        assertTrue(dataFrameTransformCheckpoints.matches(dataFrameTransformCheckpoints));
+        DataFrameTransformCheckpoint dataFrameTransformCheckpointsCopy = copyInstance(dataFrameTransformCheckpoints);
+
+        // with copy
+        assertTrue(dataFrameTransformCheckpoints.matches(dataFrameTransformCheckpointsCopy));
+        assertTrue(dataFrameTransformCheckpointsCopy.matches(dataFrameTransformCheckpoints));
+
+        // other id
+        assertFalse(dataFrameTransformCheckpoints
+                .matches(new DataFrameTransformCheckpoint(id + "-1", timestamp, checkpoint, checkpointsByIndex, timeUpperBound)));
+        // other timestamp
+        assertTrue(dataFrameTransformCheckpoints
+                .matches(new DataFrameTransformCheckpoint(id, (timestamp / 2) + 1, checkpoint, checkpointsByIndex, timeUpperBound)));
+        // other checkpoint
+        assertTrue(dataFrameTransformCheckpoints
+                .matches(new DataFrameTransformCheckpoint(id, timestamp, (checkpoint / 2) + 1, checkpointsByIndex, timeUpperBound)));
+        // other index checkpoints
+        assertFalse(dataFrameTransformCheckpoints
+                .matches(new DataFrameTransformCheckpoint(id, timestamp, checkpoint, otherCheckpointsByIndex, timeUpperBound)));
+        // other time upper bound
+        assertTrue(dataFrameTransformCheckpoints
+                .matches(new DataFrameTransformCheckpoint(id, timestamp, checkpoint, checkpointsByIndex, (timeUpperBound / 2) + 1)));
+    }
+
+    private static Map<String, long[]> randomCheckpointsByIndex() {
+        Map<String, long[]> checkpointsByIndex = new TreeMap<>();
+        for (int i = 0; i < randomIntBetween(1, 10); ++i) {
+            List<Long> checkpoints = new ArrayList<>();
+            for (int j = 0; j < randomIntBetween(1, 20); ++j) {
+                checkpoints.add(randomNonNegativeLong());
+            }
+            checkpointsByIndex.put(randomAlphaOfLengthBetween(1, 10), checkpoints.stream().mapToLong(l -> l).toArray());
+        }
+        return checkpointsByIndex;
+    }
+}

x-pack/plugin/core/src/test/java/org/elasticsearch/xpack/core/dataframe/transforms/DataFrameTransformConfigTests.java

@@ -10,17 +10,14 @@ import org.elasticsearch.common.Strings;
 import org.elasticsearch.common.io.stream.Writeable.Reader;
 import org.elasticsearch.common.xcontent.DeprecationHandler;
 import org.elasticsearch.common.xcontent.ToXContent;
-import org.elasticsearch.common.xcontent.ToXContent.Params;
 import org.elasticsearch.common.xcontent.XContentBuilder;
 import org.elasticsearch.common.xcontent.XContentFactory;
 import org.elasticsearch.common.xcontent.XContentParser;
 import org.elasticsearch.common.xcontent.XContentType;
-import org.elasticsearch.xpack.core.dataframe.DataFrameField;
 import org.elasticsearch.xpack.core.dataframe.transforms.pivot.PivotConfigTests;
 import org.junit.Before;
 
 import java.io.IOException;
-import java.util.Collections;
 import java.util.HashMap;
 import java.util.Map;
 
@@ -28,9 +25,6 @@ import static org.elasticsearch.test.TestMatchers.matchesPattern;
 
 public class DataFrameTransformConfigTests extends AbstractSerializingDataFrameTestCase<DataFrameTransformConfig> {
 
-    private static Params TO_XCONTENT_PARAMS = new ToXContent.MapParams(
-            Collections.singletonMap(DataFrameField.FOR_INTERNAL_STORAGE, "true"));
-
     private String transformId;
     private boolean runWithHeaders;
 

x-pack/plugin/data-frame/qa/single-node-tests/src/test/java/org/elasticsearch/xpack/dataframe/integration/DataFrameConfigurationIndexIT.java

@@ -15,8 +15,8 @@ import org.elasticsearch.common.Strings;
 import org.elasticsearch.common.xcontent.XContentBuilder;
 import org.elasticsearch.common.xcontent.support.XContentMapValues;
 import org.elasticsearch.xpack.core.dataframe.DataFrameField;
-import org.elasticsearch.xpack.dataframe.persistence.DataFrameInternalIndex;
 import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformConfig;
+import org.elasticsearch.xpack.dataframe.persistence.DataFrameInternalIndex;
 
 import java.io.IOException;
 
@@ -47,6 +47,9 @@ public class DataFrameConfigurationIndexIT extends DataFrameRestTestCase {
             client().performRequest(req);
         }
 
+        // refresh the index
+        assertOK(client().performRequest(new Request("POST", DataFrameInternalIndex.INDEX_NAME + "/_refresh")));
+
         Request deleteRequest = new Request("DELETE", DATAFRAME_ENDPOINT + fakeTransformName);
         Response deleteResponse = client().performRequest(deleteRequest);
         assertOK(deleteResponse);

x-pack/plugin/data-frame/src/main/java/org/elasticsearch/xpack/dataframe/DataFrame.java

@@ -57,6 +57,7 @@ import org.elasticsearch.xpack.dataframe.action.TransportPutDataFrameTransformAc
 import org.elasticsearch.xpack.dataframe.action.TransportStartDataFrameTransformAction;
 import org.elasticsearch.xpack.dataframe.action.TransportStartDataFrameTransformTaskAction;
 import org.elasticsearch.xpack.dataframe.action.TransportStopDataFrameTransformAction;
+import org.elasticsearch.xpack.dataframe.checkpoint.DataFrameTransformsCheckpointService;
 import org.elasticsearch.xpack.dataframe.persistence.DataFrameInternalIndex;
 import org.elasticsearch.xpack.dataframe.persistence.DataFrameTransformsConfigManager;
 import org.elasticsearch.xpack.dataframe.rest.action.RestDeleteDataFrameTransformAction;
@@ -98,6 +99,7 @@ public class DataFrame extends Plugin implements ActionPlugin, PersistentTaskPlu
     private final Settings settings;
     private final boolean transportClientMode;
     private final SetOnce<DataFrameTransformsConfigManager> dataFrameTransformsConfigManager = new SetOnce<>();
+    private final SetOnce<DataFrameTransformsCheckpointService> dataFrameTransformsCheckpointService = new SetOnce<>();
     private final SetOnce<SchedulerEngine> schedulerEngine = new SetOnce<>();
 
     public DataFrame(Settings settings) {
@@ -180,8 +182,9 @@ public class DataFrame extends Plugin implements ActionPlugin, PersistentTaskPlu
         }
 
         dataFrameTransformsConfigManager.set(new DataFrameTransformsConfigManager(client, xContentRegistry));
+        dataFrameTransformsCheckpointService.set(new DataFrameTransformsCheckpointService(client));
 
-        return Collections.singletonList(dataFrameTransformsConfigManager.get());
+        return Arrays.asList(dataFrameTransformsConfigManager.get(), dataFrameTransformsCheckpointService.get());
     }
 
     @Override
@@ -207,10 +210,12 @@ public class DataFrame extends Plugin implements ActionPlugin, PersistentTaskPlu
 
         // the transforms config manager should have been created
         assert dataFrameTransformsConfigManager.get() != null;
+        assert dataFrameTransformsCheckpointService.get() != null;
+
         return Collections.singletonList(new DataFrameTransformPersistentTasksExecutor(client, dataFrameTransformsConfigManager.get(),
-            schedulerEngine.get(), threadPool));
+                dataFrameTransformsCheckpointService.get(), schedulerEngine.get(), threadPool));
     }
-    
+
     @Override
     public void close() {
         if (schedulerEngine.get() != null) {

x-pack/plugin/data-frame/src/main/java/org/elasticsearch/xpack/dataframe/action/TransportDeleteDataFrameTransformAction.java

@@ -59,7 +59,7 @@ public class TransportDeleteDataFrameTransformAction extends TransportTasksActio
         IndexerState state = task.getState().getIndexerState();
         if (state.equals(IndexerState.STOPPED)) {
             task.onCancelled();
-            transformsConfigManager.deleteTransformConfiguration(request.getId(), ActionListener.wrap(r -> {
+            transformsConfigManager.deleteTransform(request.getId(), ActionListener.wrap(r -> {
                 listener.onResponse(new Response(true));
             }, listener::onFailure));
         } else {
@@ -80,7 +80,7 @@ public class TransportDeleteDataFrameTransformAction extends TransportTasksActio
                 // we couldn't find the transform in the persistent task CS, but maybe the transform exists in the configuration index,
                 // if so delete the orphaned document and do not throw (for the normal case we want to stop the task first,
                 // than delete the configuration document if and only if the data frame transform is in stopped state)
-                transformsConfigManager.deleteTransformConfiguration(request.getId(), ActionListener.wrap(r -> {
+                transformsConfigManager.deleteTransform(request.getId(), ActionListener.wrap(r -> {
                     listener.onResponse(new Response(true));
                     return;
                 }, listener::onFailure));

x-pack/plugin/data-frame/src/main/java/org/elasticsearch/xpack/dataframe/action/TransportPutDataFrameTransformAction.java

@@ -42,6 +42,7 @@ import org.elasticsearch.xpack.core.dataframe.DataFrameMessages;
 import org.elasticsearch.xpack.core.dataframe.action.PutDataFrameTransformAction;
 import org.elasticsearch.xpack.core.dataframe.action.PutDataFrameTransformAction.Request;
 import org.elasticsearch.xpack.core.dataframe.action.PutDataFrameTransformAction.Response;
+import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformConfig;
 import org.elasticsearch.xpack.core.security.SecurityContext;
 import org.elasticsearch.xpack.core.security.action.user.HasPrivilegesAction;
 import org.elasticsearch.xpack.core.security.action.user.HasPrivilegesRequest;
@@ -51,7 +52,6 @@ import org.elasticsearch.xpack.core.security.authz.permission.ResourcePrivileges
 import org.elasticsearch.xpack.core.security.support.Exceptions;
 import org.elasticsearch.xpack.dataframe.persistence.DataFrameTransformsConfigManager;
 import org.elasticsearch.xpack.dataframe.persistence.DataframeIndex;
-import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformConfig;
 import org.elasticsearch.xpack.dataframe.transforms.pivot.Pivot;
 
 import java.io.IOException;

x-pack/plugin/data-frame/src/main/java/org/elasticsearch/xpack/dataframe/checkpoint/CheckpointException.java

@@ -0,0 +1,26 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the Elastic License;
+ * you may not use this file except in compliance with the Elastic License.
+ */
+
+package org.elasticsearch.xpack.dataframe.checkpoint;
+
+import org.elasticsearch.ElasticsearchException;
+import org.elasticsearch.common.io.stream.StreamInput;
+
+import java.io.IOException;
+
+public class CheckpointException extends ElasticsearchException {
+    public CheckpointException(String msg, Object... params) {
+        super(msg, null, params);
+    }
+
+    public CheckpointException(String msg, Throwable cause, Object... params) {
+        super(msg, cause, params);
+    }
+
+    public CheckpointException(StreamInput in) throws IOException {
+        super(in);
+    }
+}

x-pack/plugin/data-frame/src/main/java/org/elasticsearch/xpack/dataframe/checkpoint/DataFrameTransformsCheckpointService.java

@@ -0,0 +1,136 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the Elastic License;
+ * you may not use this file except in compliance with the Elastic License.
+ */
+
+package org.elasticsearch.xpack.dataframe.checkpoint;
+
+import org.apache.logging.log4j.LogManager;
+import org.apache.logging.log4j.Logger;
+import org.elasticsearch.action.ActionListener;
+import org.elasticsearch.action.admin.indices.get.GetIndexAction;
+import org.elasticsearch.action.admin.indices.get.GetIndexRequest;
+import org.elasticsearch.action.admin.indices.stats.IndicesStatsAction;
+import org.elasticsearch.action.admin.indices.stats.IndicesStatsRequest;
+import org.elasticsearch.action.admin.indices.stats.ShardStats;
+import org.elasticsearch.client.Client;
+import org.elasticsearch.xpack.core.ClientHelper;
+import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformCheckpoint;
+import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformConfig;
+
+import java.util.Arrays;
+import java.util.HashSet;
+import java.util.Map;
+import java.util.Set;
+import java.util.TreeMap;
+
+/**
+ * DataFrameTransform Checkpoint Service
+ *
+ * Allows checkpointing a source of a data frame transform which includes all relevant checkpoints of the source.
+ *
+ * This will be used to checkpoint a transform, detect changes, run the transform in continuous mode.
+ *
+ */
+public class DataFrameTransformsCheckpointService {
+
+    private static final Logger logger = LogManager.getLogger(DataFrameTransformsCheckpointService.class);
+
+    private final Client client;
+
+    public DataFrameTransformsCheckpointService(final Client client) {
+        this.client = client;
+    }
+
+    /**
+     * Get an unnumbered checkpoint. These checkpoints are for persistence but comparing state.
+     *
+     * @param transformConfig the @link{DataFrameTransformConfig}
+     * @param listener listener to call after inner request returned
+     */
+    public void getCheckpoint(DataFrameTransformConfig transformConfig, ActionListener<DataFrameTransformCheckpoint> listener) {
+        getCheckpoint(transformConfig, -1L, listener);
+    }
+
+    /**
+     * Get a checkpoint, used to store a checkpoint.
+     *
+     * @param transformConfig the @link{DataFrameTransformConfig}
+     * @param checkpoint the number of the checkpoint
+     * @param listener listener to call after inner request returned
+     */
+    public void getCheckpoint(DataFrameTransformConfig transformConfig, long checkpoint,
+            ActionListener<DataFrameTransformCheckpoint> listener) {
+        long timestamp = System.currentTimeMillis();
+
+        // placeholder for time based synchronization
+        long timeUpperBound = 0;
+
+        // 1st get index to see the indexes the user has access to
+        GetIndexRequest getIndexRequest = new GetIndexRequest().indices(transformConfig.getSource());
+
+        ClientHelper.executeWithHeadersAsync(transformConfig.getHeaders(), ClientHelper.DATA_FRAME_ORIGIN, client, GetIndexAction.INSTANCE,
+                getIndexRequest, ActionListener.wrap(getIndexResponse -> {
+                    Set<String> userIndices = new HashSet<>(Arrays.asList(getIndexResponse.getIndices()));
+
+                    // 2nd get stats request
+                    ClientHelper.executeAsyncWithOrigin(client, ClientHelper.DATA_FRAME_ORIGIN, IndicesStatsAction.INSTANCE,
+                            new IndicesStatsRequest().indices(transformConfig.getSource()), ActionListener.wrap(response -> {
+                                if (response.getFailedShards() != 0) {
+                                    throw new CheckpointException("Source has [" + response.getFailedShards() + "] failed shards");
+                                }
+
+                                Map<String, long[]> checkpointsByIndex = extractIndexCheckPoints(response.getShards(), userIndices);
+                                DataFrameTransformCheckpoint checkpointDoc = new DataFrameTransformCheckpoint(transformConfig.getId(),
+                                        timestamp, checkpoint, checkpointsByIndex, timeUpperBound);
+                                listener.onResponse(checkpointDoc);
+
+                            }, IndicesStatsRequestException -> {
+                                throw new CheckpointException("Failed to retrieve indices stats", IndicesStatsRequestException);
+                            }));
+
+                }, getIndexException -> {
+                    throw new CheckpointException("Failed to retrieve list of indices", getIndexException);
+                }));
+
+    }
+
+    static Map<String, long[]> extractIndexCheckPoints(ShardStats[] shards, Set<String> userIndices) {
+        Map<String, TreeMap<Integer, Long>> checkpointsByIndex = new TreeMap<>();
+
+        for (ShardStats shard : shards) {
+            String indexName = shard.getShardRouting().getIndexName();
+            if (userIndices.contains(indexName)) {
+                if (checkpointsByIndex.containsKey(indexName)) {
+                    // we have already seen this index, just check/add shards
+                    TreeMap<Integer, Long> checkpoints = checkpointsByIndex.get(indexName);
+                    if (checkpoints.containsKey(shard.getShardRouting().getId())) {
+                        // there is already a checkpoint entry for this index/shard combination, check if they match
+                        if (checkpoints.get(shard.getShardRouting().getId()) != shard.getSeqNoStats().getGlobalCheckpoint()) {
+                            throw new CheckpointException("Global checkpoints mismatch for index [" + indexName + "] between shards of id ["
+                                    + shard.getShardRouting().getId() + "]");
+                        }
+                    } else {
+                        // 1st time we see this shard for this index, add the entry for the shard
+                        checkpoints.put(shard.getShardRouting().getId(), shard.getSeqNoStats().getGlobalCheckpoint());
+                    }
+                } else {
+                    // 1st time we see this index, create an entry for the index and add the shard checkpoint
+                    checkpointsByIndex.put(indexName, new TreeMap<>());
+                    checkpointsByIndex.get(indexName).put(shard.getShardRouting().getId(), shard.getSeqNoStats().getGlobalCheckpoint());
+                }
+            }
+        }
+
+        // create the final structure
+        Map<String, long[]> checkpointsByIndexReduced = new TreeMap<>();
+
+        checkpointsByIndex.forEach((indexName, checkpoints) -> {
+            checkpointsByIndexReduced.put(indexName, checkpoints.values().stream().mapToLong(l -> l).toArray());
+        });
+
+        return checkpointsByIndexReduced;
+    }
+
+}

x-pack/plugin/data-frame/src/main/java/org/elasticsearch/xpack/dataframe/persistence/DataFrameTransformsConfigManager.java

@@ -13,9 +13,6 @@ import org.elasticsearch.ResourceAlreadyExistsException;
 import org.elasticsearch.ResourceNotFoundException;
 import org.elasticsearch.action.ActionListener;
 import org.elasticsearch.action.DocWriteRequest;
-import org.elasticsearch.action.DocWriteResponse;
-import org.elasticsearch.action.delete.DeleteAction;
-import org.elasticsearch.action.delete.DeleteRequest;
 import org.elasticsearch.action.get.GetAction;
 import org.elasticsearch.action.get.GetRequest;
 import org.elasticsearch.action.index.IndexAction;
@@ -32,8 +29,13 @@ import org.elasticsearch.common.xcontent.XContentParser;
 import org.elasticsearch.common.xcontent.XContentType;
 import org.elasticsearch.index.IndexNotFoundException;
 import org.elasticsearch.index.engine.VersionConflictEngineException;
+import org.elasticsearch.index.query.QueryBuilder;
+import org.elasticsearch.index.query.QueryBuilders;
+import org.elasticsearch.index.reindex.DeleteByQueryAction;
+import org.elasticsearch.index.reindex.DeleteByQueryRequest;
 import org.elasticsearch.xpack.core.dataframe.DataFrameField;
 import org.elasticsearch.xpack.core.dataframe.DataFrameMessages;
+import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformCheckpoint;
 import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformConfig;
 
 import java.io.IOException;
@@ -58,6 +60,37 @@ public class DataFrameTransformsConfigManager {
         this.xContentRegistry = xContentRegistry;
     }
 
+    /**
+     * Persist a checkpoint in the internal index
+     *
+     * @param checkpoint the @link{DataFrameTransformCheckpoint}
+     * @param listener listener to call after request has been made
+     */
+    public void putTransformCheckpoint(DataFrameTransformCheckpoint checkpoint, ActionListener<Boolean> listener) {
+        try (XContentBuilder builder = XContentFactory.jsonBuilder()) {
+            XContentBuilder source = checkpoint.toXContent(builder, new ToXContent.MapParams(TO_XCONTENT_PARAMS));
+
+            IndexRequest indexRequest = new IndexRequest(DataFrameInternalIndex.INDEX_NAME)
+                    .opType(DocWriteRequest.OpType.INDEX)
+                    .setRefreshPolicy(WriteRequest.RefreshPolicy.IMMEDIATE)
+                    .id(DataFrameTransformCheckpoint.documentId(checkpoint.getTransformId(), checkpoint.getCheckpoint()))
+                    .source(source);
+
+            executeAsyncWithOrigin(client, DATA_FRAME_ORIGIN, IndexAction.INSTANCE, indexRequest, ActionListener.wrap(r -> {
+                listener.onResponse(true);
+            }, listener::onFailure));
+        } catch (IOException e) {
+            // not expected to happen but for the sake of completeness
+            listener.onFailure(e);
+        }
+    }
+
+    /**
+     * Store the transform configuration in the internal index
+     *
+     * @param transformConfig the @link{DataFrameTransformConfig}
+     * @param listener listener to call after request
+     */
     public void putTransformConfiguration(DataFrameTransformConfig transformConfig, ActionListener<Boolean> listener) {
         try (XContentBuilder builder = XContentFactory.jsonBuilder()) {
             XContentBuilder source = transformConfig.toXContent(builder, new ToXContent.MapParams(TO_XCONTENT_PARAMS));
@@ -89,8 +122,35 @@ public class DataFrameTransformsConfigManager {
         }
     }
 
-    // For transforms returned via GET data_frame/transforms, see the TransportGetDataFrameTransformsAction
-    // This function is only for internal use.
+    /**
+     * Get a stored checkpoint, requires the transform id as well as the checkpoint id
+     *
+     * @param transformId the transform id
+     * @param checkpoint the checkpoint
+     * @param resultListener listener to call after request has been made
+     */
+    public void getTransformCheckpoint(String transformId, long checkpoint, ActionListener<DataFrameTransformCheckpoint> resultListener) {
+        GetRequest getRequest = new GetRequest(DataFrameInternalIndex.INDEX_NAME,
+                DataFrameTransformCheckpoint.documentId(transformId, checkpoint));
+        executeAsyncWithOrigin(client, DATA_FRAME_ORIGIN, GetAction.INSTANCE, getRequest, ActionListener.wrap(getResponse -> {
+
+            if (getResponse.isExists() == false) {
+                // do not fail if checkpoint does not exist but return an empty checkpoint
+                resultListener.onResponse(DataFrameTransformCheckpoint.EMPTY);
+                return;
+            }
+            BytesReference source = getResponse.getSourceAsBytesRef();
+            parseCheckpointsLenientlyFromSource(source, transformId, resultListener);
+        }, resultListener::onFailure));
+    }
+
+    /**
+     * Get the transform configuration for a given transform id. This function is only for internal use. For transforms returned via GET
+     * data_frame/transforms, see the TransportGetDataFrameTransformsAction
+     *
+     * @param transformId the transform id
+     * @param resultListener listener to call after inner request has returned
+     */
     public void getTransformConfiguration(String transformId, ActionListener<DataFrameTransformConfig> resultListener) {
         GetRequest getRequest = new GetRequest(DataFrameInternalIndex.INDEX_NAME, DataFrameTransformConfig.documentId(transformId));
         executeAsyncWithOrigin(client, DATA_FRAME_ORIGIN, GetAction.INSTANCE, getRequest, ActionListener.wrap(getResponse -> {
@@ -112,13 +172,25 @@ public class DataFrameTransformsConfigManager {
         }));
     }
 
-    public void deleteTransformConfiguration(String transformId, ActionListener<Boolean> listener) {
-        DeleteRequest request = new DeleteRequest(DataFrameInternalIndex.INDEX_NAME, DataFrameTransformConfig.documentId(transformId));
-        request.setRefreshPolicy(WriteRequest.RefreshPolicy.IMMEDIATE);
+    /**
+     * This deletes the configuration and all other documents corresponding to the transform id (e.g. checkpoints).
+     *
+     * @param transformId the transform id
+     * @param listener listener to call after inner request returned
+     */
+    public void deleteTransform(String transformId, ActionListener<Boolean> listener) {
+        DeleteByQueryRequest request = new DeleteByQueryRequest()
+                .setAbortOnVersionConflict(false) //since these documents are not updated, a conflict just means it was deleted previously
+                .setSlices(5);
 
-        executeAsyncWithOrigin(client, DATA_FRAME_ORIGIN, DeleteAction.INSTANCE, request, ActionListener.wrap(deleteResponse -> {
+        request.indices(DataFrameInternalIndex.INDEX_NAME);
+        QueryBuilder query = QueryBuilders.termQuery(DataFrameField.ID.getPreferredName(), transformId);
+        request.setQuery(query);
+        request.setRefresh(true);
 
-            if (deleteResponse.getResult() == DocWriteResponse.Result.NOT_FOUND) {
+        executeAsyncWithOrigin(client, DATA_FRAME_ORIGIN, DeleteByQueryAction.INSTANCE, request, ActionListener.wrap(deleteResponse -> {
+
+            if (deleteResponse.getDeleted() == 0) {
                 listener.onFailure(new ResourceNotFoundException(
                         DataFrameMessages.getMessage(DataFrameMessages.REST_DATA_FRAME_UNKNOWN_TRANSFORM, transformId)));
                 return;
@@ -145,4 +217,16 @@ public class DataFrameTransformsConfigManager {
             transformListener.onFailure(e);
         }
     }
+
+    private void parseCheckpointsLenientlyFromSource(BytesReference source, String transformId,
+            ActionListener<DataFrameTransformCheckpoint> transformListener) {
+        try (InputStream stream = source.streamInput();
+                XContentParser parser = XContentFactory.xContent(XContentType.JSON)
+                     .createParser(xContentRegistry, LoggingDeprecationHandler.INSTANCE, stream)) {
+            transformListener.onResponse(DataFrameTransformCheckpoint.fromXContent(parser, true));
+        } catch (Exception e) {
+            logger.error(DataFrameMessages.getMessage(DataFrameMessages.FAILED_TO_PARSE_TRANSFORM_CHECKPOINTS, transformId), e);
+            transformListener.onFailure(e);
+        }
+    }
 }

x-pack/plugin/data-frame/src/main/java/org/elasticsearch/xpack/dataframe/transforms/DataFrameTransformPersistentTasksExecutor.java

@@ -21,6 +21,7 @@ import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransform;
 import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformState;
 import org.elasticsearch.xpack.core.scheduler.SchedulerEngine;
 import org.elasticsearch.xpack.dataframe.DataFrame;
+import org.elasticsearch.xpack.dataframe.checkpoint.DataFrameTransformsCheckpointService;
 import org.elasticsearch.xpack.dataframe.persistence.DataFrameTransformsConfigManager;
 
 import java.util.Map;
@@ -31,14 +32,17 @@ public class DataFrameTransformPersistentTasksExecutor extends PersistentTasksEx
 
     private final Client client;
     private final DataFrameTransformsConfigManager transformsConfigManager;
+    private final DataFrameTransformsCheckpointService dataFrameTransformsCheckpointService;
     private final SchedulerEngine schedulerEngine;
     private final ThreadPool threadPool;
 
     public DataFrameTransformPersistentTasksExecutor(Client client, DataFrameTransformsConfigManager transformsConfigManager,
-            SchedulerEngine schedulerEngine, ThreadPool threadPool) {
+            DataFrameTransformsCheckpointService dataFrameTransformsCheckpointService, SchedulerEngine schedulerEngine,
+            ThreadPool threadPool) {
         super(DataFrameField.TASK_NAME, DataFrame.TASK_THREAD_POOL_NAME);
         this.client = client;
         this.transformsConfigManager = transformsConfigManager;
+        this.dataFrameTransformsCheckpointService = dataFrameTransformsCheckpointService;
         this.schedulerEngine = schedulerEngine;
         this.threadPool = threadPool;
     }
@@ -67,6 +71,7 @@ public class DataFrameTransformPersistentTasksExecutor extends PersistentTasksEx
     protected AllocatedPersistentTask createTask(long id, String type, String action, TaskId parentTaskId,
             PersistentTasksCustomMetaData.PersistentTask<DataFrameTransform> persistentTask, Map<String, String> headers) {
         return new DataFrameTransformTask(id, type, action, parentTaskId, persistentTask.getParams(),
-                (DataFrameTransformState) persistentTask.getState(), client, transformsConfigManager, schedulerEngine, threadPool, headers);
+                (DataFrameTransformState) persistentTask.getState(), client, transformsConfigManager, dataFrameTransformsCheckpointService,
+                schedulerEngine, threadPool, headers);
     }
 }

x-pack/plugin/data-frame/src/main/java/org/elasticsearch/xpack/dataframe/transforms/DataFrameTransformTask.java

@@ -24,16 +24,17 @@ import org.elasticsearch.threadpool.ThreadPool;
 import org.elasticsearch.xpack.core.ClientHelper;
 import org.elasticsearch.xpack.core.dataframe.DataFrameField;
 import org.elasticsearch.xpack.core.dataframe.DataFrameMessages;
-import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransform;
+import org.elasticsearch.xpack.core.dataframe.action.StartDataFrameTransformTaskAction;
+import org.elasticsearch.xpack.core.dataframe.action.StartDataFrameTransformTaskAction.Response;
+import org.elasticsearch.xpack.core.dataframe.action.StopDataFrameTransformAction;
 import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameIndexerTransformStats;
+import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransform;
 import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformConfig;
 import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformState;
 import org.elasticsearch.xpack.core.indexing.IndexerState;
 import org.elasticsearch.xpack.core.scheduler.SchedulerEngine;
 import org.elasticsearch.xpack.core.scheduler.SchedulerEngine.Event;
-import org.elasticsearch.xpack.core.dataframe.action.StartDataFrameTransformTaskAction;
-import org.elasticsearch.xpack.core.dataframe.action.StartDataFrameTransformTaskAction.Response;
-import org.elasticsearch.xpack.core.dataframe.action.StopDataFrameTransformAction;
+import org.elasticsearch.xpack.dataframe.checkpoint.DataFrameTransformsCheckpointService;
 import org.elasticsearch.xpack.dataframe.persistence.DataFrameTransformsConfigManager;
 
 import java.util.Map;
@@ -58,7 +59,8 @@ public class DataFrameTransformTask extends AllocatedPersistentTask implements S
 
     public DataFrameTransformTask(long id, String type, String action, TaskId parentTask, DataFrameTransform transform,
             DataFrameTransformState state, Client client, DataFrameTransformsConfigManager transformsConfigManager,
-            SchedulerEngine schedulerEngine, ThreadPool threadPool, Map<String, String> headers) {
+            DataFrameTransformsCheckpointService transformsCheckpointService, SchedulerEngine schedulerEngine, ThreadPool threadPool,
+            Map<String, String> headers) {
         super(id, type, action, DataFrameField.PERSISTENT_TASK_DESCRIPTION_PREFIX + transform.getId(), parentTask, headers);
         this.transform = transform;
         this.schedulerEngine = schedulerEngine;
@@ -83,8 +85,8 @@ public class DataFrameTransformTask extends AllocatedPersistentTask implements S
             initialGeneration = state.getGeneration();
         }
 
-        this.indexer = new ClientDataFrameIndexer(transform.getId(), transformsConfigManager, new AtomicReference<>(initialState),
-                initialPosition, client);
+        this.indexer = new ClientDataFrameIndexer(transform.getId(), transformsConfigManager, transformsCheckpointService,
+                new AtomicReference<>(initialState), initialPosition, client);
         this.generation = new AtomicReference<Long>(initialGeneration);
     }
 
@@ -226,15 +228,18 @@ public class DataFrameTransformTask extends AllocatedPersistentTask implements S
         private static final int LOAD_TRANSFORM_TIMEOUT_IN_SECONDS = 30;
         private final Client client;
         private final DataFrameTransformsConfigManager transformsConfigManager;
+        private final DataFrameTransformsCheckpointService transformsCheckpointService;
         private final String transformId;
 
         private DataFrameTransformConfig transformConfig = null;
 
         public ClientDataFrameIndexer(String transformId, DataFrameTransformsConfigManager transformsConfigManager,
-                AtomicReference<IndexerState> initialState, Map<String, Object> initialPosition, Client client) {
+                DataFrameTransformsCheckpointService transformsCheckpointService, AtomicReference<IndexerState> initialState,
+                Map<String, Object> initialPosition, Client client) {
             super(threadPool.executor(ThreadPool.Names.GENERIC), initialState, initialPosition);
             this.transformId = transformId;
             this.transformsConfigManager = transformsConfigManager;
+            this.transformsCheckpointService = transformsCheckpointService;
             this.client = client;
         }
 

x-pack/plugin/data-frame/src/test/java/org/elasticsearch/xpack/dataframe/checkpoint/DataFrameTransformsCheckpointServiceTests.java

@@ -0,0 +1,207 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the Elastic License;
+ * you may not use this file except in compliance with the Elastic License.
+ */
+
+package org.elasticsearch.xpack.dataframe.checkpoint;
+
+import org.elasticsearch.action.admin.indices.stats.CommonStats;
+import org.elasticsearch.action.admin.indices.stats.ShardStats;
+import org.elasticsearch.cluster.routing.RecoverySource;
+import org.elasticsearch.cluster.routing.RecoverySource.PeerRecoverySource;
+import org.elasticsearch.cluster.routing.ShardRouting;
+import org.elasticsearch.cluster.routing.UnassignedInfo;
+import org.elasticsearch.common.UUIDs;
+import org.elasticsearch.index.Index;
+import org.elasticsearch.index.cache.query.QueryCacheStats;
+import org.elasticsearch.index.cache.request.RequestCacheStats;
+import org.elasticsearch.index.engine.SegmentsStats;
+import org.elasticsearch.index.fielddata.FieldDataStats;
+import org.elasticsearch.index.flush.FlushStats;
+import org.elasticsearch.index.get.GetStats;
+import org.elasticsearch.index.merge.MergeStats;
+import org.elasticsearch.index.refresh.RefreshStats;
+import org.elasticsearch.index.search.stats.SearchStats;
+import org.elasticsearch.index.seqno.SeqNoStats;
+import org.elasticsearch.index.shard.DocsStats;
+import org.elasticsearch.index.shard.IndexingStats;
+import org.elasticsearch.index.shard.ShardId;
+import org.elasticsearch.index.shard.ShardPath;
+import org.elasticsearch.index.store.StoreStats;
+import org.elasticsearch.index.warmer.WarmerStats;
+import org.elasticsearch.search.suggest.completion.CompletionStats;
+import org.elasticsearch.test.ESTestCase;
+
+import java.nio.file.Path;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Map;
+import java.util.Map.Entry;
+import java.util.Set;
+
+import static org.hamcrest.Matchers.containsString;
+
+public class DataFrameTransformsCheckpointServiceTests extends ESTestCase {
+
+    public void testExtractIndexCheckpoints() {
+        Map<String, long[]> expectedCheckpoints = new HashMap<>();
+        Set<String> indices = randomUserIndices();
+
+        ShardStats[] shardStatsArray = createRandomShardStats(expectedCheckpoints, indices, false, false);
+
+        Map<String, long[]> checkpoints = DataFrameTransformsCheckpointService.extractIndexCheckPoints(shardStatsArray, indices);
+
+        assertEquals(expectedCheckpoints.size(), checkpoints.size());
+        assertEquals(expectedCheckpoints.keySet(), checkpoints.keySet());
+
+        // low-level compare
+        for (Entry<String, long[]> entry : expectedCheckpoints.entrySet()) {
+            assertTrue(Arrays.equals(entry.getValue(), checkpoints.get(entry.getKey())));
+        }
+    }
+
+    public void testExtractIndexCheckpointsLostPrimaries() {
+        Map<String, long[]> expectedCheckpoints = new HashMap<>();
+        Set<String> indices = randomUserIndices();
+
+        ShardStats[] shardStatsArray = createRandomShardStats(expectedCheckpoints, indices, true, false);
+
+        Map<String, long[]> checkpoints = DataFrameTransformsCheckpointService.extractIndexCheckPoints(shardStatsArray, indices);
+
+        assertEquals(expectedCheckpoints.size(), checkpoints.size());
+        assertEquals(expectedCheckpoints.keySet(), checkpoints.keySet());
+
+        // low-level compare
+        for (Entry<String, long[]> entry : expectedCheckpoints.entrySet()) {
+            assertTrue(Arrays.equals(entry.getValue(), checkpoints.get(entry.getKey())));
+        }
+    }
+
+    public void testExtractIndexCheckpointsInconsistentGlobalCheckpoints() {
+        Map<String, long[]> expectedCheckpoints = new HashMap<>();
+        Set<String> indices = randomUserIndices();
+
+        ShardStats[] shardStatsArray = createRandomShardStats(expectedCheckpoints, indices, randomBoolean(), true);
+
+        // fail
+        CheckpointException e = expectThrows(CheckpointException.class,
+                () -> DataFrameTransformsCheckpointService.extractIndexCheckPoints(shardStatsArray, indices));
+
+        assertThat(e.getMessage(), containsString("Global checkpoints mismatch"));    }
+
+    /**
+     * Create a random set of 3 index names
+     * @return set of indices a simulated user has access to
+     */
+    private static Set<String> randomUserIndices() {
+        Set<String> indices = new HashSet<>();
+
+        // never create an empty set
+        if (randomBoolean()) {
+            indices.add("index-1");
+        } else {
+            indices.add("index-2");
+        }
+        if (randomBoolean()) {
+            indices.add("index-3");
+        }
+        return indices;
+    }
+
+    /**
+     * create a ShardStats for testing with random fuzzing
+     *
+     * @param expectedCheckpoints output parameter to return the checkpoints to expect
+     * @param userIndices set of indices that are visible
+     * @param skipPrimaries whether some shards do not have a primary shard at random
+     * @param inconsistentGlobalCheckpoints whether to introduce inconsistent global checkpoints
+     * @return array of ShardStats
+     */
+    private static ShardStats[] createRandomShardStats(Map<String, long[]> expectedCheckpoints, Set<String> userIndices,
+            boolean skipPrimaries, boolean inconsistentGlobalCheckpoints) {
+
+        // always create the full list
+        List<Index> indices = new ArrayList<>();
+        indices.add(new Index("index-1", UUIDs.randomBase64UUID(random())));
+        indices.add(new Index("index-2", UUIDs.randomBase64UUID(random())));
+        indices.add(new Index("index-3", UUIDs.randomBase64UUID(random())));
+
+        List<ShardStats> shardStats = new ArrayList<>();
+        for (final Index index : indices) {
+            int numShards = randomIntBetween(1, 5);
+
+            List<Long> checkpoints = new ArrayList<>();
+            for (int shardIndex = 0; shardIndex < numShards; shardIndex++) {
+                // we need at least one replica for testing
+                int numShardCopies = randomIntBetween(2, 4);
+
+                int inconsistentReplica = -1;
+                if (inconsistentGlobalCheckpoints) {
+                    inconsistentReplica = randomIntBetween(0, numShardCopies - 1);
+                }
+
+                // SeqNoStats asserts that checkpoints are logical
+                long localCheckpoint = randomLongBetween(0L, 100000000L);
+                long globalCheckpoint = randomBoolean() ? localCheckpoint : randomLongBetween(0L, 100000000L);
+                long maxSeqNo = Math.max(localCheckpoint, globalCheckpoint);
+
+                SeqNoStats seqNoStats = new SeqNoStats(maxSeqNo, localCheckpoint, globalCheckpoint);
+                checkpoints.add(globalCheckpoint);
+
+                for (int replica = 0;  replica < numShardCopies; replica++) {
+                    ShardId shardId = new ShardId(index, shardIndex);
+                    boolean primary = (replica == 0);
+
+                    if (skipPrimaries) {
+                        primary = randomBoolean();
+                    }
+
+                    Path path = createTempDir().resolve("indices").resolve(index.getUUID()).resolve(String.valueOf(shardIndex));
+                    ShardRouting shardRouting = ShardRouting.newUnassigned(shardId, primary,
+                        primary ? RecoverySource.EmptyStoreRecoverySource.INSTANCE : PeerRecoverySource.INSTANCE,
+                        new UnassignedInfo(UnassignedInfo.Reason.INDEX_CREATED, null)
+                        );
+                    shardRouting = shardRouting.initialize("node-0", null, ShardRouting.UNAVAILABLE_EXPECTED_SHARD_SIZE);
+                    shardRouting = shardRouting.moveToStarted();
+
+                    CommonStats stats = new CommonStats();
+                    stats.fieldData = new FieldDataStats();
+                    stats.queryCache = new QueryCacheStats();
+                    stats.docs = new DocsStats();
+                    stats.store = new StoreStats();
+                    stats.indexing = new IndexingStats();
+                    stats.search = new SearchStats();
+                    stats.segments = new SegmentsStats();
+                    stats.merge = new MergeStats();
+                    stats.refresh = new RefreshStats();
+                    stats.completion = new CompletionStats();
+                    stats.requestCache = new RequestCacheStats();
+                    stats.get = new GetStats();
+                    stats.flush = new FlushStats();
+                    stats.warmer = new WarmerStats();
+
+                    if (inconsistentReplica == replica) {
+                        // overwrite
+                        seqNoStats = new SeqNoStats(maxSeqNo, localCheckpoint, globalCheckpoint + randomLongBetween(10L, 100L));
+                    }
+
+                    shardStats.add(new ShardStats(shardRouting, new ShardPath(false, path, path, shardId), stats, null, seqNoStats, null));
+                }
+            }
+
+            if (userIndices.contains(index.getName())) {
+                expectedCheckpoints.put(index.getName(), checkpoints.stream().mapToLong(l -> l).toArray());
+            }
+        }
+        // shuffle the shard stats
+        Collections.shuffle(shardStats, random());
+        ShardStats[] shardStatsArray = shardStats.toArray(new ShardStats[0]);
+        return shardStatsArray;
+    }
+
+}

x-pack/plugin/data-frame/src/test/java/org/elasticsearch/xpack/dataframe/persistence/DataFrameSingleNodeTestCase.java

@@ -11,6 +11,7 @@ import org.elasticsearch.action.LatchedActionListener;
 import org.elasticsearch.cluster.ClusterState;
 import org.elasticsearch.common.CheckedConsumer;
 import org.elasticsearch.common.settings.Settings;
+import org.elasticsearch.index.reindex.ReindexPlugin;
 import org.elasticsearch.plugins.Plugin;
 import org.elasticsearch.test.ESSingleNodeTestCase;
 import org.elasticsearch.xpack.core.template.TemplateUtils;
@@ -43,7 +44,7 @@ public abstract class DataFrameSingleNodeTestCase extends ESSingleNodeTestCase {
 
     @Override
     protected Collection<Class<? extends Plugin>> getPlugins() {
-        return pluginList(LocalStateDataFrame.class);
+        return pluginList(LocalStateDataFrame.class, ReindexPlugin.class);
     }
 
     protected <T> void assertAsync(Consumer<ActionListener<T>> function, T expected, CheckedConsumer<T, ? extends Exception> onAnswer,

x-pack/plugin/data-frame/src/test/java/org/elasticsearch/xpack/dataframe/persistence/DataFrameTransformsConfigManagerTests.java

@@ -9,6 +9,8 @@ package org.elasticsearch.xpack.dataframe.persistence;
 import org.elasticsearch.ResourceAlreadyExistsException;
 import org.elasticsearch.ResourceNotFoundException;
 import org.elasticsearch.xpack.core.dataframe.DataFrameMessages;
+import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformCheckpoint;
+import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformCheckpointTests;
 import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformConfig;
 import org.elasticsearch.xpack.core.dataframe.transforms.DataFrameTransformConfigTests;
 import org.junit.Before;
@@ -48,7 +50,7 @@ public class DataFrameTransformsConfigManagerTests extends DataFrameSingleNodeTe
 
     public void testDeleteMissingTransform() throws InterruptedException {
         // the index does not exist yet
-        assertAsync(listener -> transformsConfigManager.deleteTransformConfiguration("not_there", listener), (Boolean) null, null, e -> {
+        assertAsync(listener -> transformsConfigManager.deleteTransform("not_there", listener), (Boolean) null, null, e -> {
             assertEquals(ResourceNotFoundException.class, e.getClass());
             assertEquals(DataFrameMessages.getMessage(DataFrameMessages.REST_DATA_FRAME_UNKNOWN_TRANSFORM, "not_there"), e.getMessage());
         });
@@ -60,13 +62,13 @@ public class DataFrameTransformsConfigManagerTests extends DataFrameSingleNodeTe
                 true, null, null);
 
         // same test, but different code path
-        assertAsync(listener -> transformsConfigManager.deleteTransformConfiguration("not_there", listener), (Boolean) null, null, e -> {
+        assertAsync(listener -> transformsConfigManager.deleteTransform("not_there", listener), (Boolean) null, null, e -> {
             assertEquals(ResourceNotFoundException.class, e.getClass());
             assertEquals(DataFrameMessages.getMessage(DataFrameMessages.REST_DATA_FRAME_UNKNOWN_TRANSFORM, "not_there"), e.getMessage());
         });
     }
 
-    public void testCreateReadDelete() throws InterruptedException {
+    public void testCreateReadDeleteTransform() throws InterruptedException {
         DataFrameTransformConfig transformConfig = DataFrameTransformConfigTests.randomDataFrameTransformConfig();
 
         // create transform
@@ -84,15 +86,14 @@ public class DataFrameTransformsConfigManagerTests extends DataFrameSingleNodeTe
         });
 
         // delete transform
-        assertAsync(listener -> transformsConfigManager.deleteTransformConfiguration(transformConfig.getId(), listener), true, null, null);
+        assertAsync(listener -> transformsConfigManager.deleteTransform(transformConfig.getId(), listener), true, null, null);
 
         // delete again
-        assertAsync(listener -> transformsConfigManager.deleteTransformConfiguration(transformConfig.getId(), listener), (Boolean) null,
-                null, e -> {
-                    assertEquals(ResourceNotFoundException.class, e.getClass());
-                    assertEquals(DataFrameMessages.getMessage(DataFrameMessages.REST_DATA_FRAME_UNKNOWN_TRANSFORM, transformConfig.getId()),
-                            e.getMessage());
-                });
+        assertAsync(listener -> transformsConfigManager.deleteTransform(transformConfig.getId(), listener), (Boolean) null, null, e -> {
+            assertEquals(ResourceNotFoundException.class, e.getClass());
+            assertEquals(DataFrameMessages.getMessage(DataFrameMessages.REST_DATA_FRAME_UNKNOWN_TRANSFORM, transformConfig.getId()),
+                    e.getMessage());
+        });
 
         // try to get deleted transform
         assertAsync(listener -> transformsConfigManager.getTransformConfiguration(transformConfig.getId(), listener),
@@ -102,4 +103,29 @@ public class DataFrameTransformsConfigManagerTests extends DataFrameSingleNodeTe
                             e.getMessage());
                 });
     }
+
+    public void testCreateReadDeleteCheckPoint() throws InterruptedException {
+        DataFrameTransformCheckpoint checkpoint = DataFrameTransformCheckpointTests.randomDataFrameTransformCheckpoints();
+
+        // create
+        assertAsync(listener -> transformsConfigManager.putTransformCheckpoint(checkpoint, listener), true, null, null);
+
+        // read
+        assertAsync(listener -> transformsConfigManager.getTransformCheckpoint(checkpoint.getTransformId(), checkpoint.getCheckpoint(),
+                listener), checkpoint, null, null);
+
+        // delete
+        assertAsync(listener -> transformsConfigManager.deleteTransform(checkpoint.getTransformId(), listener), true, null, null);
+
+        // delete again
+        assertAsync(listener -> transformsConfigManager.deleteTransform(checkpoint.getTransformId(), listener), (Boolean) null, null, e -> {
+            assertEquals(ResourceNotFoundException.class, e.getClass());
+            assertEquals(DataFrameMessages.getMessage(DataFrameMessages.REST_DATA_FRAME_UNKNOWN_TRANSFORM, checkpoint.getTransformId()),
+                    e.getMessage());
+        });
+
+        // getting a non-existing checkpoint returns null
+        assertAsync(listener -> transformsConfigManager.getTransformCheckpoint(checkpoint.getTransformId(), checkpoint.getCheckpoint(),
+                listener), DataFrameTransformCheckpoint.EMPTY, null, null);
+    }
 }