use of org.apache.falcon.entity.v0.feed.Feed in project atlas by apache.
the class FalconBridge method createProcessEntity.
/**
* Creates process entity
*
* @param process process entity
* @param falconStore config store
* @return process instance reference
*
* @throws FalconException if retrieving from the configuration store fail
*/
public static List<Referenceable> createProcessEntity(org.apache.falcon.entity.v0.process.Process process, ConfigurationStore falconStore) throws FalconException {
LOG.info("Creating process Entity : {}", process.getName());
// The requirement is for each cluster, create a process entity with name
// clustername.processname
List<Referenceable> entities = new ArrayList<>();
if (process.getClusters() != null) {
for (Cluster processCluster : process.getClusters().getClusters()) {
org.apache.falcon.entity.v0.cluster.Cluster cluster = falconStore.get(EntityType.CLUSTER, processCluster.getName());
Referenceable clusterReferenceable = getClusterEntityReference(cluster.getName(), cluster.getColo());
entities.add(clusterReferenceable);
List<Referenceable> inputs = new ArrayList<>();
if (process.getInputs() != null) {
for (Input input : process.getInputs().getInputs()) {
Feed feed = falconStore.get(EntityType.FEED, input.getFeed());
Referenceable inputReferenceable = getFeedDataSetReference(feed, clusterReferenceable);
entities.add(inputReferenceable);
inputs.add(inputReferenceable);
}
}
List<Referenceable> outputs = new ArrayList<>();
if (process.getOutputs() != null) {
for (Output output : process.getOutputs().getOutputs()) {
Feed feed = falconStore.get(EntityType.FEED, output.getFeed());
Referenceable outputReferenceable = getFeedDataSetReference(feed, clusterReferenceable);
entities.add(outputReferenceable);
outputs.add(outputReferenceable);
}
}
if (!inputs.isEmpty() || !outputs.isEmpty()) {
Referenceable processEntity = new Referenceable(FalconDataTypes.FALCON_PROCESS.getName());
processEntity.set(AtlasClient.NAME, process.getName());
processEntity.set(AtlasClient.REFERENCEABLE_ATTRIBUTE_NAME, getProcessQualifiedName(process.getName(), cluster.getName()));
processEntity.set(FalconBridge.FREQUENCY, process.getFrequency().toString());
if (!inputs.isEmpty()) {
processEntity.set(AtlasClient.PROCESS_ATTRIBUTE_INPUTS, inputs);
}
if (!outputs.isEmpty()) {
processEntity.set(AtlasClient.PROCESS_ATTRIBUTE_OUTPUTS, outputs);
}
// set cluster
processEntity.set(FalconBridge.RUNSON, clusterReferenceable);
// Set user
if (process.getACL() != null) {
processEntity.set(AtlasClient.OWNER, process.getACL().getOwner());
}
if (StringUtils.isNotEmpty(process.getTags())) {
processEntity.set(FalconBridge.TAGS, EventUtil.convertKeyValueStringToMap(process.getTags()));
}
if (process.getPipelines() != null) {
processEntity.set(FalconBridge.PIPELINES, process.getPipelines());
}
processEntity.set(FalconBridge.WFPROPERTIES, getProcessEntityWFProperties(process.getWorkflow(), process.getName()));
entities.add(processEntity);
}
}
}
return entities;
}
use of org.apache.falcon.entity.v0.feed.Feed in project atlas by apache.
the class FalconHookIT method getTableFeed.
private Feed getTableFeed(String feedResource, String clusterName, String secondClusterName) throws Exception {
Feed feed = loadEntity(EntityType.FEED, feedResource, "feed" + random());
org.apache.falcon.entity.v0.feed.Cluster feedCluster = feed.getClusters().getClusters().get(0);
feedCluster.setName(clusterName);
String dbName = "db" + random();
String tableName = "table" + random();
feedCluster.getTable().setUri(getTableUri(dbName, tableName));
String dbName2 = "db" + random();
String tableName2 = "table" + random();
if (secondClusterName != null) {
org.apache.falcon.entity.v0.feed.Cluster feedCluster2 = feed.getClusters().getClusters().get(1);
feedCluster2.setName(secondClusterName);
feedCluster2.getTable().setUri(getTableUri(dbName2, tableName2));
}
STORE.publish(EntityType.FEED, feed);
String feedId = assertFeedIsRegistered(feed, clusterName);
assertFeedAttributes(feedId);
verifyFeedLineage(feed.getName(), clusterName, feedId, dbName, tableName);
if (secondClusterName != null) {
String feedId2 = assertFeedIsRegistered(feed, secondClusterName);
assertFeedAttributes(feedId2);
verifyFeedLineage(feed.getName(), secondClusterName, feedId2, dbName2, tableName2);
}
return feed;
}
use of org.apache.falcon.entity.v0.feed.Feed in project atlas by apache.
the class FalconHookIT method testCreateProcess.
@Test
public void testCreateProcess() throws Exception {
Cluster cluster = loadEntity(EntityType.CLUSTER, CLUSTER_RESOURCE, "cluster" + random());
STORE.publish(EntityType.CLUSTER, cluster);
assertClusterIsRegistered(cluster);
Feed infeed = getTableFeed(FEED_RESOURCE, cluster.getName(), null);
String infeedId = atlasClient.getEntity(FalconDataTypes.FALCON_FEED.getName(), AtlasClient.REFERENCEABLE_ATTRIBUTE_NAME, FalconBridge.getFeedQualifiedName(infeed.getName(), cluster.getName())).getId()._getId();
Feed outfeed = getTableFeed(FEED_RESOURCE, cluster.getName());
String outFeedId = atlasClient.getEntity(FalconDataTypes.FALCON_FEED.getName(), AtlasClient.REFERENCEABLE_ATTRIBUTE_NAME, FalconBridge.getFeedQualifiedName(outfeed.getName(), cluster.getName())).getId()._getId();
Process process = loadEntity(EntityType.PROCESS, PROCESS_RESOURCE, "process" + random());
process.getClusters().getClusters().get(0).setName(cluster.getName());
process.getInputs().getInputs().get(0).setFeed(infeed.getName());
process.getOutputs().getOutputs().get(0).setFeed(outfeed.getName());
STORE.publish(EntityType.PROCESS, process);
String pid = assertProcessIsRegistered(process, cluster.getName());
Referenceable processEntity = atlasClient.getEntity(pid);
assertNotNull(processEntity);
assertEquals(processEntity.get(AtlasClient.NAME), process.getName());
assertEquals(((List<Id>) processEntity.get("inputs")).get(0)._getId(), infeedId);
assertEquals(((List<Id>) processEntity.get("outputs")).get(0)._getId(), outFeedId);
}
use of org.apache.falcon.entity.v0.feed.Feed in project incubator-atlas by apache.
the class FalconHookIT method getHDFSFeed.
private TypeUtils.Pair<String, Feed> getHDFSFeed(String feedResource, String clusterName) throws Exception {
Feed feed = loadEntity(EntityType.FEED, feedResource, "feed" + random());
org.apache.falcon.entity.v0.feed.Cluster feedCluster = feed.getClusters().getClusters().get(0);
feedCluster.setName(clusterName);
STORE.publish(EntityType.FEED, feed);
String feedId = assertFeedIsRegistered(feed, clusterName);
assertFeedAttributes(feedId);
String processId = assertEntityIsRegistered(FalconDataTypes.FALCON_FEED_CREATION.getName(), AtlasClient.REFERENCEABLE_ATTRIBUTE_NAME, FalconBridge.getFeedQualifiedName(feed.getName(), clusterName));
Referenceable processEntity = atlasClient.getEntity(processId);
assertEquals(((List<Id>) processEntity.get("outputs")).get(0).getId()._getId(), feedId);
String inputId = ((List<Id>) processEntity.get("inputs")).get(0).getId()._getId();
Referenceable pathEntity = atlasClient.getEntity(inputId);
assertEquals(pathEntity.getTypeName(), HiveMetaStoreBridge.HDFS_PATH);
List<Location> locations = FeedHelper.getLocations(feedCluster, feed);
Location dataLocation = FileSystemStorage.getLocation(locations, LocationType.DATA);
assertEquals(pathEntity.get(AtlasClient.REFERENCEABLE_ATTRIBUTE_NAME), FalconBridge.normalize(dataLocation.getPath()));
return TypeUtils.Pair.of(feedId, feed);
}
use of org.apache.falcon.entity.v0.feed.Feed in project incubator-atlas by apache.
the class FalconHookIT method testCreateProcess.
@Test
public void testCreateProcess() throws Exception {
Cluster cluster = loadEntity(EntityType.CLUSTER, CLUSTER_RESOURCE, "cluster" + random());
STORE.publish(EntityType.CLUSTER, cluster);
assertClusterIsRegistered(cluster);
Feed infeed = getTableFeed(FEED_RESOURCE, cluster.getName(), null);
String infeedId = atlasClient.getEntity(FalconDataTypes.FALCON_FEED.getName(), AtlasClient.REFERENCEABLE_ATTRIBUTE_NAME, FalconBridge.getFeedQualifiedName(infeed.getName(), cluster.getName())).getId()._getId();
Feed outfeed = getTableFeed(FEED_RESOURCE, cluster.getName());
String outFeedId = atlasClient.getEntity(FalconDataTypes.FALCON_FEED.getName(), AtlasClient.REFERENCEABLE_ATTRIBUTE_NAME, FalconBridge.getFeedQualifiedName(outfeed.getName(), cluster.getName())).getId()._getId();
Process process = loadEntity(EntityType.PROCESS, PROCESS_RESOURCE, "process" + random());
process.getClusters().getClusters().get(0).setName(cluster.getName());
process.getInputs().getInputs().get(0).setFeed(infeed.getName());
process.getOutputs().getOutputs().get(0).setFeed(outfeed.getName());
STORE.publish(EntityType.PROCESS, process);
String pid = assertProcessIsRegistered(process, cluster.getName());
Referenceable processEntity = atlasClient.getEntity(pid);
assertNotNull(processEntity);
assertEquals(processEntity.get(AtlasClient.NAME), process.getName());
assertEquals(((List<Id>) processEntity.get("inputs")).get(0)._getId(), infeedId);
assertEquals(((List<Id>) processEntity.get("outputs")).get(0)._getId(), outFeedId);
}
Aggregations