/**
* Initializes the storm instance
*
* @return a real storm controller if possible, else a no controller
*/
@SuppressWarnings("unchecked")
public static IStormController getController() {
final GlobalPropertiesBean globals =
Lambdas.get(
() -> {
try {
return BeanTemplateUtils.from(
PropertiesUtils.getSubConfig(
ModuleUtils.getStaticConfig(), GlobalPropertiesBean.PROPERTIES_ROOT)
.orElse(null),
GlobalPropertiesBean.class);
} catch (IOException e) {
_logger.error(
ErrorUtils.getLongForm(
"Couldn't set globals property bean in storm harvest tech onInit: {0}", e));
return null;
}
});
if (null == globals) {
return new NoStormController();
}
_logger.info(
"Loading storm config from: "
+ globals.local_yarn_config_dir()
+ File.separator
+ "storm.yaml");
Yaml yaml = new Yaml();
InputStream input;
Map<String, Object> object;
try {
input =
new FileInputStream(
new File(globals.local_yarn_config_dir() + File.separator + "storm.yaml"));
object = (Map<String, Object>) yaml.load(input);
} catch (FileNotFoundException e) {
_logger.error(
ErrorUtils.getLongForm("Error reading storm.yaml in storm harvest tech onInit: {0}", e));
object = new HashMap<String, Object>();
}
if (object.containsKey(backtype.storm.Config.NIMBUS_HOST)) {
_logger.info("starting in remote mode v5");
_logger.info(object.get(backtype.storm.Config.NIMBUS_HOST));
// run in distributed mode
IStormController storm_controller =
StormControllerUtil.getRemoteStormController(
(String) object.get(backtype.storm.Config.NIMBUS_HOST),
(int) object.get(backtype.storm.Config.NIMBUS_THRIFT_PORT),
(String) object.get(backtype.storm.Config.STORM_THRIFT_TRANSPORT_PLUGIN));
return storm_controller;
} else {
return new NoStormController();
}
}
@Override
protected void configure() {
final Config config = ModuleUtils.getStaticConfig();
try {
final DataImportConfigurationBean bean =
BeanTemplateUtils.from(
PropertiesUtils.getSubConfig(config, DataImportConfigurationBean.PROPERTIES_ROOT)
.orElse(null),
DataImportConfigurationBean.class);
this.bind(DataImportConfigurationBean.class).toInstance(bean);
this.bind(AnalyticStateTriggerCheckFactory.class).in(Scopes.SINGLETON);
} catch (Exception e) {
throw new RuntimeException(
ErrorUtils.get(
ErrorUtils.INVALID_CONFIG_ERROR,
DataImportConfigurationBean.class.toString(),
config.getConfig(DataImportConfigurationBean.PROPERTIES_ROOT)),
e);
}
}
/**
* Entry point
*
* @param args - config_file source_key harvest_tech_id
* @throws Exception
*/
public static void main(final String[] args) {
try {
if (args.length < 1) {
System.out.println("CLI: config_file");
System.exit(-1);
}
System.out.println("Running with command line: " + Arrays.toString(args));
final Config config = ConfigFactory.parseFile(new File(args[0]));
final DataImportManagerModule app =
ModuleUtils.initializeApplication(
Arrays.asList(new Module()),
Optional.of(config),
Either.left(DataImportManagerModule.class));
app.start();
} catch (Throwable e) {
_logger.error(ErrorUtils.getLongForm("Exception reached main(): {0}", e));
try {
e.printStackTrace();
} catch (Exception e2) { // the exception failed!
}
System.exit(-1);
}
}
@Test
public void test_ageOut() throws IOException, InterruptedException, ExecutionException {
// Call test_endToEnd_autoTime to create 5 time based indexes
// 2015-01-01 -> 2015-05-01
// How far is now from 2015-05-03
final Date d = TimeUtils.getDateFromSuffix("2015-03-02").success();
final long total_time_ms = new Date().getTime() - d.getTime();
final long total_days = total_time_ms / (1000L * 3600L * 24L);
final String age_out = ErrorUtils.get("{0} days", total_days);
final DataBucketBean bucket =
BeanTemplateUtils.build(DataBucketBean.class)
.with("full_name", "/test/end-end/auto-time")
.with(
DataBucketBean::data_schema,
BeanTemplateUtils.build(DataSchemaBean.class)
.with(
DataSchemaBean::temporal_schema,
BeanTemplateUtils.build(TemporalSchemaBean.class)
.with(TemporalSchemaBean::exist_age_max, age_out)
.done()
.get())
.done()
.get())
.done()
.get();
final String template_name = ElasticsearchIndexUtils.getBaseIndexName(bucket);
test_endToEnd_autoTime(false);
_index_service
._crud_factory
.getClient()
.admin()
.indices()
.prepareCreate(template_name + "_2015-03-01_1")
.execute()
.actionGet();
final GetMappingsResponse gmr =
_index_service
._crud_factory
.getClient()
.admin()
.indices()
.prepareGetMappings(template_name + "*")
.execute()
.actionGet();
assertEquals(6, gmr.getMappings().keys().size());
CompletableFuture<BasicMessageBean> cf =
_index_service.getDataService().get().handleAgeOutRequest(bucket);
BasicMessageBean res = cf.get();
assertEquals(true, res.success());
assertTrue("sensible message: " + res.message(), res.message().contains(" 2 "));
assertTrue(
"Message marked as loggable: " + res.details(),
Optional.ofNullable(res.details()).filter(m -> m.containsKey("loggable")).isPresent());
System.out.println("Return from to delete: " + res.message());
Thread.sleep(5000L); // give the indexes time to delete
final GetMappingsResponse gmr2 =
_index_service
._crud_factory
.getClient()
.admin()
.indices()
.prepareGetMappings(template_name + "*")
.execute()
.actionGet();
assertEquals(3, gmr2.getMappings().keys().size());
// Check some edge cases:
// 1) Run it again, returns success but not loggable:
CompletableFuture<BasicMessageBean> cf2 =
_index_service.getDataService().get().handleAgeOutRequest(bucket);
BasicMessageBean res2 = cf2.get();
assertEquals(true, res2.success());
assertTrue("sensible message: " + res2.message(), res2.message().contains(" 0 "));
assertTrue(
"Message _not_ marked as loggable: " + res2.details(),
!Optional.ofNullable(res2.details()).map(m -> m.get("loggable")).isPresent());
// 2) No temporal settings
final DataBucketBean bucket3 =
BeanTemplateUtils.build(DataBucketBean.class)
.with("full_name", "/test/handle/age/out/delete/not/temporal")
.with(
DataBucketBean::data_schema,
BeanTemplateUtils.build(DataSchemaBean.class).done().get())
.done()
.get();
CompletableFuture<BasicMessageBean> cf3 =
_index_service.getDataService().get().handleAgeOutRequest(bucket3);
BasicMessageBean res3 = cf3.get();
// no temporal settings => returns success
assertEquals(true, res3.success());
// 3) Unparseable temporal settings (in theory won't validate but we can test here)
final DataBucketBean bucket4 =
BeanTemplateUtils.build(DataBucketBean.class)
.with("full_name", "/test/handle/age/out/delete/temporal/malformed")
.with(
DataBucketBean::data_schema,
BeanTemplateUtils.build(DataSchemaBean.class)
.with(
DataSchemaBean::temporal_schema,
BeanTemplateUtils.build(TemporalSchemaBean.class)
.with(TemporalSchemaBean::exist_age_max, "bananas")
.done()
.get())
.done()
.get())
.done()
.get();
CompletableFuture<BasicMessageBean> cf4 =
_index_service.getDataService().get().handleAgeOutRequest(bucket4);
BasicMessageBean res4 = cf4.get();
// no temporal settings => returns success
assertEquals(false, res4.success());
}
@Test
public void test_getStreamingTopology()
throws UnsupportedFileSystemException, InterruptedException, ExecutionException {
final DataBucketBean bucket = createBucket("test_tech_id_stream");
final String pathname1 =
System.getProperty("user.dir") + "/misc_test_assets/simple-topology-example.jar";
final Path path1 = FileContext.getLocalFSFileContext().makeQualified(new Path(pathname1));
final String pathname2 =
System.getProperty("user.dir") + "/misc_test_assets/simple-harvest-example2.jar";
final Path path2 = FileContext.getLocalFSFileContext().makeQualified(new Path(pathname2));
List<SharedLibraryBean> lib_elements = createSharedLibraryBeans(path1, path2);
//////////////////////////////////////////////////////
// 1) Check - if called with an error, then just passes that error along
final BasicMessageBean error =
SharedErrorUtils.buildErrorMessage("test_source", "test_message", "test_error");
final Validation<BasicMessageBean, IEnrichmentStreamingTopology> test1 =
DataBucketChangeActor.getStreamingTopology(
bucket,
new BucketActionMessage.BucketActionOfferMessage(bucket),
"test_source2",
Validation.fail(error));
assertTrue("Got error back", test1.isFail());
assertEquals("test_source", test1.fail().source());
assertEquals("test_message", test1.fail().command());
assertEquals("test_error", test1.fail().message());
//////////////////////////////////////////////////////
// 2) Check the error handling inside getStreamingTopology
final ImmutableMap<String, Tuple2<SharedLibraryBean, String>> test2_input =
ImmutableMap.<String, Tuple2<SharedLibraryBean, String>>builder()
.put("test_tech_id_stream_2b", Tuples._2T(null, null))
.build();
final Validation<BasicMessageBean, IEnrichmentStreamingTopology> test2a =
DataBucketChangeActor.getStreamingTopology(
BeanTemplateUtils.clone(bucket)
.with(
DataBucketBean::streaming_enrichment_topology,
BeanTemplateUtils.build(EnrichmentControlMetadataBean.class)
.with(
EnrichmentControlMetadataBean::library_names_or_ids,
Arrays.asList("test_tech_id_stream_2a"))
.done()
.get())
.done(),
new BucketActionMessage.BucketActionOfferMessage(bucket),
"test_source2a",
Validation.success(test2_input));
assertTrue("Got error back", test2a.isFail());
assertEquals("test_source2a", test2a.fail().source());
assertEquals("BucketActionOfferMessage", test2a.fail().command());
assertEquals(
ErrorUtils.get(
SharedErrorUtils.SHARED_LIBRARY_NAME_NOT_FOUND,
bucket.full_name(),
"(unknown)"), // (cloned bucket above)
test2a.fail().message());
final Validation<BasicMessageBean, IEnrichmentStreamingTopology> test2b =
DataBucketChangeActor.getStreamingTopology(
BeanTemplateUtils.clone(bucket)
.with(
DataBucketBean::streaming_enrichment_topology,
BeanTemplateUtils.build(EnrichmentControlMetadataBean.class)
.with(
EnrichmentControlMetadataBean::library_names_or_ids,
Arrays.asList("test_tech_id_stream_2b"))
.done()
.get())
.done(),
new BucketActionMessage.BucketActionOfferMessage(bucket),
"test_source2b",
Validation.success(test2_input));
assertTrue("Got error back", test2b.isFail());
assertEquals("test_source2b", test2b.fail().source());
assertEquals("BucketActionOfferMessage", test2b.fail().command());
assertEquals(
ErrorUtils.get(
SharedErrorUtils.SHARED_LIBRARY_NAME_NOT_FOUND,
bucket.full_name(),
"(unknown)"), // (cloned bucket above)
test2a.fail().message());
//////////////////////////////////////////////////////
// 3) OK now it will actually do something
final String java_name =
_service_context.getGlobalProperties().local_cached_jar_dir()
+ File.separator
+ "test_tech_id_stream.cache.jar";
_logger.info(
"Needed to delete locally cached file? " + java_name + ": " + new File(java_name).delete());
// Requires that the file has already been cached:
final Validation<BasicMessageBean, String> cached_file =
JarCacheUtils.getCachedJar(
_service_context.getGlobalProperties().local_cached_jar_dir(),
lib_elements.get(0),
_service_context.getStorageService(),
"test3",
"test3")
.get();
if (cached_file.isFail()) {
fail("About to crash with: " + cached_file.fail().message());
}
assertTrue("The cached file exists: " + java_name, new File(java_name).exists());
// OK the setup is done and validated now actually test the underlying call:
final ImmutableMap<String, Tuple2<SharedLibraryBean, String>> test3_input =
ImmutableMap.<String, Tuple2<SharedLibraryBean, String>>builder()
.put("test_tech_id_stream", Tuples._2T(lib_elements.get(0), cached_file.success()))
.build();
final Validation<BasicMessageBean, IEnrichmentStreamingTopology> test3 =
DataBucketChangeActor.getStreamingTopology(
BeanTemplateUtils.clone(bucket)
.with(
DataBucketBean::streaming_enrichment_topology,
BeanTemplateUtils.build(EnrichmentControlMetadataBean.class)
.with(
EnrichmentControlMetadataBean::library_names_or_ids,
Arrays.asList("test_tech_id_stream"))
.done()
.get())
.done(),
new BucketActionMessage.BucketActionOfferMessage(bucket),
"test_source3",
Validation.success(test3_input));
if (test3.isFail()) {
fail("About to crash with: " + test3.fail().message());
}
assertTrue("getStreamingTopology call succeeded", test3.isSuccess());
assertTrue("topology created: ", test3.success() != null);
assertEquals(lib_elements.get(0).misc_entry_point(), test3.success().getClass().getName());
// (Try again but with failing version, due to class not found)
final ImmutableMap<String, Tuple2<SharedLibraryBean, String>> test3a_input =
ImmutableMap.<String, Tuple2<SharedLibraryBean, String>>builder()
.put("test_tech_id_stream_fail", Tuples._2T(lib_elements.get(3), cached_file.success()))
.build();
final Validation<BasicMessageBean, IEnrichmentStreamingTopology> test3a =
DataBucketChangeActor.getStreamingTopology(
BeanTemplateUtils.clone(bucket)
.with(
DataBucketBean::streaming_enrichment_topology,
BeanTemplateUtils.build(EnrichmentControlMetadataBean.class)
.with(
EnrichmentControlMetadataBean::library_names_or_ids,
Arrays.asList("test_tech_id_stream_fail"))
.done()
.get())
.done(),
new BucketActionMessage.BucketActionOfferMessage(bucket),
"test_source3",
Validation.success(test3a_input));
assertTrue("Got error back", test3a.isFail());
assertTrue(
"Right error: " + test3a.fail().message(),
test3a.fail().message().contains("com.ikanow.aleph2.test.example.ExampleStreamTopology"));
// Now check with the "not just the harvest tech" flag set
final String java_name2 =
_service_context.getGlobalProperties().local_cached_jar_dir()
+ File.separator
+ "test_module_id.cache.jar";
_logger.info(
"Needed to delete locally cached file? "
+ java_name2
+ ": "
+ new File(java_name2).delete());
// Requires that the file has already been cached:
final Validation<BasicMessageBean, String> cached_file2 =
JarCacheUtils.getCachedJar(
_service_context.getGlobalProperties().local_cached_jar_dir(),
lib_elements.get(1),
_service_context.getStorageService(),
"test3b",
"test3b")
.get();
if (cached_file2.isFail()) {
fail("About to crash with: " + cached_file2.fail().message());
}
assertTrue("The cached file exists: " + java_name, new File(java_name2).exists());
final ImmutableMap<String, Tuple2<SharedLibraryBean, String>> test3b_input =
ImmutableMap.<String, Tuple2<SharedLibraryBean, String>>builder()
.put("test_tech_id_stream", Tuples._2T(lib_elements.get(0), cached_file.success()))
.put("test_module_id", Tuples._2T(lib_elements.get(1), cached_file.success()))
.build();
final EnrichmentControlMetadataBean enrichment_module =
new EnrichmentControlMetadataBean(
"test_tech_name",
Collections.emptyList(),
true,
null,
Arrays.asList("test_tech_id_stream", "test_module_id"),
null,
null);
final Validation<BasicMessageBean, IEnrichmentStreamingTopology> test3b =
DataBucketChangeActor.getStreamingTopology(
BeanTemplateUtils.clone(bucket)
.with(DataBucketBean::streaming_enrichment_topology, enrichment_module)
.done(),
new BucketActionMessage.BucketActionOfferMessage(bucket),
"test_source3b",
Validation.success(test3b_input));
if (test3b.isFail()) {
fail("About to crash with: " + test3b.fail().message());
}
assertTrue("getStreamingTopology call succeeded", test3b.isSuccess());
assertTrue("topology created: ", test3b.success() != null);
assertEquals(lib_elements.get(0).misc_entry_point(), test3b.success().getClass().getName());
// TODO add a test for disabled streaming but config given (should default to passthrough top
// and
// ignore given topology
}
@Test
public void test_cacheJars()
throws UnsupportedFileSystemException, InterruptedException, ExecutionException {
try {
// Preamble:
// 0) Insert 2 library beans into the management db
final DataBucketBean bucket = createBucket("test_tech_id_stream");
final String pathname1 =
System.getProperty("user.dir") + "/misc_test_assets/simple-harvest-example.jar";
final Path path1 = FileContext.getLocalFSFileContext().makeQualified(new Path(pathname1));
final String pathname2 =
System.getProperty("user.dir") + "/misc_test_assets/simple-harvest-example2.jar";
final Path path2 = FileContext.getLocalFSFileContext().makeQualified(new Path(pathname2));
List<SharedLibraryBean> lib_elements = createSharedLibraryBeans(path1, path2);
final IManagementDbService underlying_db =
_service_context.getService(IManagementDbService.class, Optional.empty()).get();
final IManagementCrudService<SharedLibraryBean> library_crud =
underlying_db.getSharedLibraryStore();
library_crud.deleteDatastore();
assertEquals("Cleansed library store", 0L, (long) library_crud.countObjects().get());
library_crud.storeObjects(lib_elements).get();
assertEquals("Should have 4 library beans", 4L, (long) library_crud.countObjects().get());
// 0a) Check with no streaming, gets nothing
{
CompletableFuture<
Validation<BasicMessageBean, Map<String, Tuple2<SharedLibraryBean, String>>>>
reply_structure =
DataBucketChangeActor.cacheJars(
bucket,
_service_context.getCoreManagementDbService(),
_service_context.getGlobalProperties(),
_service_context.getStorageService(),
_service_context,
"test1_source",
"test1_command");
if (reply_structure.get().isFail()) {
fail("About to crash with: " + reply_structure.get().fail().message());
}
assertTrue("cacheJars should return valid reply", reply_structure.get().isSuccess());
final Map<String, Tuple2<SharedLibraryBean, String>> reply_map =
reply_structure.get().success();
assertEquals(0L, reply_map.size()); // (both modules, 1x for _id and 1x for name)
}
// 0b) Create the more complex bucket
final EnrichmentControlMetadataBean enrichment_module =
new EnrichmentControlMetadataBean(
"test_name",
Collections.emptyList(),
true,
null,
Arrays.asList("test_tech_id_stream", "test_module_id"),
null,
new LinkedHashMap<>());
final DataBucketBean bucket2 =
BeanTemplateUtils.clone(bucket)
.with(DataBucketBean::streaming_enrichment_topology, enrichment_module)
.done();
// 1) Normal operation
CompletableFuture<
Validation<BasicMessageBean, Map<String, Tuple2<SharedLibraryBean, String>>>>
reply_structure =
DataBucketChangeActor.cacheJars(
bucket2,
_service_context.getCoreManagementDbService(),
_service_context.getGlobalProperties(),
_service_context.getStorageService(),
_service_context,
"test1_source",
"test1_command");
if (reply_structure.get().isFail()) {
fail("About to crash with: " + reply_structure.get().fail().message());
}
assertTrue("cacheJars should return valid reply", reply_structure.get().isSuccess());
final Map<String, Tuple2<SharedLibraryBean, String>> reply_map =
reply_structure.get().success();
assertEquals(
"Should have 4 beans: " + reply_map.toString(),
4L,
reply_map.size()); // (both modules, 1x for _id and 1x for name)
// 3) Couple of error cases:
final EnrichmentControlMetadataBean enrichment_module2 =
new EnrichmentControlMetadataBean(
"test_name",
Collections.emptyList(),
true,
null,
Arrays.asList("test_tech_id_stream", "test_module_id", "failtest"),
null,
new LinkedHashMap<>());
final DataBucketBean bucket3 =
BeanTemplateUtils.clone(bucket)
.with(DataBucketBean::streaming_enrichment_topology, enrichment_module2)
.done();
CompletableFuture<
Validation<BasicMessageBean, Map<String, Tuple2<SharedLibraryBean, String>>>>
reply_structure3 =
DataBucketChangeActor.cacheJars(
bucket3,
_service_context.getCoreManagementDbService(),
_service_context.getGlobalProperties(),
_service_context.getStorageService(),
_service_context,
"test2_source",
"test2_command");
assertTrue("cacheJars should return error", reply_structure3.get().isFail());
} catch (Exception e) {
System.out.println(ErrorUtils.getLongForm("guice? {0}", e));
throw e;
}
}
public void start() {
final String hostname = _local_actor_context.getInformationService().getHostname();
final int MAX_ZK_ATTEMPTS = 6;
if (!_core_distributed_services.waitForAkkaJoin(
Optional.of(Duration.create(60L, TimeUnit.SECONDS)))) {
_core_distributed_services.getAkkaSystem().terminate(); // (last ditch attempt to recover)
throw new RuntimeException("Problem with CDS/Akka, try to terminate");
}
////////////////////////////////////////////////////////////////
// HARVEST
if (_service_config.harvest_enabled()) {
// Create a bucket change actor and register it vs the local message bus
final ActorRef handler =
_local_actor_context
.getActorSystem()
.actorOf(
Props.create(
com.ikanow.aleph2.data_import_manager.harvest.actors
.DataBucketHarvestChangeActor.class),
hostname + ActorNameUtils.HARVEST_BUCKET_CHANGE_SUFFIX);
_logger.info(
ErrorUtils.get(
"Attaching harvest DataBucketHarvestChangeActor {0} to bus {1}",
handler, ActorUtils.BUCKET_ACTION_EVENT_BUS));
_db_actor_context
.getBucketActionMessageBus()
.subscribe(handler, ActorUtils.BUCKET_ACTION_EVENT_BUS);
_logger.info(
ErrorUtils.get("Registering {1} with {0}", ActorUtils.BUCKET_ACTION_ZOOKEEPER, hostname));
for (int i = 0; i <= MAX_ZK_ATTEMPTS; ++i) {
try {
_core_distributed_services
.getCuratorFramework()
.create()
.creatingParentsIfNeeded()
.withMode(CreateMode.EPHEMERAL)
.forPath(ActorUtils.BUCKET_ACTION_ZOOKEEPER + "/" + hostname);
break;
} catch (Exception e) {
_logger.warn(
ErrorUtils.getLongForm(
"Failed to register with Zookeeper: {0}, retrying={1}", e, i < MAX_ZK_ATTEMPTS));
try {
Thread.sleep(10000L);
} catch (Exception __) {
}
}
}
Runtime.getRuntime()
.addShutdownHook(
new Thread(
Lambdas.wrap_runnable_u(
() -> {
_logger.info(
"Shutting down IkanowV1SynchronizationModule subservice=v1_sync_service");
_core_distributed_services
.getCuratorFramework()
.delete()
.deletingChildrenIfNeeded()
.forPath(ActorUtils.BUCKET_ACTION_ZOOKEEPER + "/" + hostname);
})));
_logger.info("Starting IkanowV1SynchronizationModule subservice=v1_sync_service");
}
////////////////////////////////////////////////////////////////
// ANALYTICS
if (_service_config.analytics_enabled()) {
// Create a analytics bucket change actor and register it vs the local message bus
final ActorRef analytics_handler =
_local_actor_context
.getActorSystem()
.actorOf(
Props.create(
com.ikanow.aleph2.data_import_manager.analytics.actors
.DataBucketAnalyticsChangeActor.class),
hostname + ActorNameUtils.ANALYTICS_BUCKET_CHANGE_SUFFIX);
_logger.info(
ErrorUtils.get(
"Attaching analytics DataBucketAnalyticsChangeActor {0} to bus {1}",
analytics_handler, ActorUtils.BUCKET_ANALYTICS_EVENT_BUS));
_db_actor_context
.getAnalyticsMessageBus()
.subscribe(analytics_handler, ActorUtils.BUCKET_ANALYTICS_EVENT_BUS);
// Create trigger supervisor and worker
final Optional<ActorRef> trigger_supervisor =
_core_distributed_services.createSingletonActor(
hostname + ActorNameUtils.ANALYTICS_TRIGGER_SUPERVISOR_SUFFIX,
ImmutableSet.<String>builder()
.add(
DistributedServicesPropertyBean.ApplicationNames.DataImportManager.toString())
.build(),
Props.create(
com.ikanow.aleph2.data_import_manager.analytics.actors
.AnalyticsTriggerSupervisorActor.class));
if (!trigger_supervisor.isPresent()) {
_logger.error(
"Analytic trigger supervisor didn't start, unknown reason (wrong CDS application_name?)");
}
final ActorRef trigger_worker =
_local_actor_context
.getActorSystem()
.actorOf(
Props.create(
com.ikanow.aleph2.data_import_manager.analytics.actors
.AnalyticsTriggerWorkerActor.class),
hostname + ActorNameUtils.ANALYTICS_TRIGGER_WORKER_SUFFIX);
_logger.info(
ErrorUtils.get(
"Attaching analytics AnalyticsTriggerWorkerActor {0} to bus {1}",
trigger_worker, ActorUtils.ANALYTICS_TRIGGER_BUS));
_db_actor_context
.getAnalyticsTriggerBus()
.subscribe(trigger_worker, ActorUtils.ANALYTICS_TRIGGER_BUS);
_logger.info(
ErrorUtils.get(
"Registering {1} with {0}", ActorUtils.BUCKET_ANALYTICS_ZOOKEEPER, hostname));
for (int i = 0; i <= MAX_ZK_ATTEMPTS; ++i) {
try {
_core_distributed_services
.getCuratorFramework()
.create()
.creatingParentsIfNeeded()
.withMode(CreateMode.EPHEMERAL)
.forPath(ActorUtils.BUCKET_ANALYTICS_ZOOKEEPER + "/" + hostname);
break;
} catch (Exception e) {
_logger.warn(
ErrorUtils.getLongForm(
"Failed to register with Zookeeper: {0}, retrying={1}", e, i < MAX_ZK_ATTEMPTS));
try {
Thread.sleep(10000L);
} catch (Exception __) {
}
}
}
Runtime.getRuntime()
.addShutdownHook(
new Thread(
Lambdas.wrap_runnable_u(
() -> {
_logger.info(
"Shutting down IkanowV1SynchronizationModule subservice=analytics");
_core_distributed_services
.getCuratorFramework()
.delete()
.deletingChildrenIfNeeded()
.forPath(ActorUtils.BUCKET_ANALYTICS_ZOOKEEPER + "/" + hostname);
})));
_logger.info("Starting IkanowV1SynchronizationModule subservice=analytics");
}
////////////////////////////////////////////////////////////////
// GOVERNANCE
if (_service_config.governance_enabled()) {
_core_distributed_services.createSingletonActor(
hostname + ".governance.actors.DataAgeOutSupervisor",
ImmutableSet.<String>builder()
.add(DistributedServicesPropertyBean.ApplicationNames.DataImportManager.toString())
.build(),
Props.create(DataAgeOutSupervisor.class));
_logger.info("Starting IkanowV1SynchronizationModule subservice=governance");
}
for (; ; ) {
try {
Thread.sleep(10000);
} catch (Exception e) {
}
}
}
/**
* Given a bucket ...returns either - a future containing the first error encountered, _or_ a map
* (both name and id as keys) of path names (and guarantee that the file has been cached when the
* future completes)
*
* @param bucket
* @param management_db
* @param globals
* @param fs
* @param handler_for_errors
* @param msg_for_errors
* @return a future containing the first error encountered, _or_ a map (both name and id as keys)
* of path names
*/
@SuppressWarnings("unchecked")
protected static <M>
CompletableFuture<
Validation<BasicMessageBean, Map<String, Tuple2<SharedLibraryBean, String>>>>
cacheJars(
final DataBucketBean bucket,
final IManagementDbService management_db,
final GlobalPropertiesBean globals,
final IStorageService fs,
final IServiceContext context,
final String handler_for_errors,
final M msg_for_errors) {
try {
MethodNamingHelper<SharedLibraryBean> helper =
BeanTemplateUtils.from(SharedLibraryBean.class);
final Optional<QueryComponent<SharedLibraryBean>> spec = getQuery(bucket);
if (!spec.isPresent()) {
return CompletableFuture.completedFuture(
Validation.<BasicMessageBean, Map<String, Tuple2<SharedLibraryBean, String>>>success(
Collections.emptyMap()));
}
return management_db
.getSharedLibraryStore()
.secured(context, new AuthorizationBean(bucket.owner_id()))
.getObjectsBySpec(spec.get())
.thenComposeAsync(
cursor -> {
// This is a map of futures from the cache call - either an error or the path name
// note we use a tuple of (id, name) as the key and then flatten out later
final Map<
Tuple2<String, String>,
Tuple2<
SharedLibraryBean,
CompletableFuture<Validation<BasicMessageBean, String>>>>
map_of_futures =
StreamSupport.stream(cursor.spliterator(), true)
.filter(
lib -> {
return true;
})
.collect(
Collectors
.<SharedLibraryBean, Tuple2<String, String>,
Tuple2<
SharedLibraryBean,
CompletableFuture<
Validation<BasicMessageBean, String>>>>
toMap(
// want to keep both the name and id versions - will
// flatten out below
lib -> Tuples._2T(lib.path_name(), lib._id()), // (key)
// spin off a future in which the file is being copied -
// save the shared library bean also
lib ->
Tuples._2T(
lib, // (value)
JarCacheUtils.getCachedJar(
globals.local_cached_jar_dir(),
lib,
fs,
handler_for_errors,
msg_for_errors))));
// denest from map of futures to future of maps, also handle any errors here:
// (some sort of "lift" function would be useful here - this are a somewhat
// inelegant few steps)
final CompletableFuture<Validation<BasicMessageBean, String>>[] futures =
(CompletableFuture<Validation<BasicMessageBean, String>>[])
map_of_futures
.values()
.stream()
.map(t2 -> t2._2())
.collect(Collectors.toList())
.toArray(new CompletableFuture[0]);
// (have to embed this thenApply instead of bringing it outside as part of the
// toCompose chain, because otherwise we'd lose map_of_futures scope)
return CompletableFuture.allOf(futures)
.<Validation<BasicMessageBean, Map<String, Tuple2<SharedLibraryBean, String>>>>
thenApply(
f -> {
try {
final Map<String, Tuple2<SharedLibraryBean, String>> almost_there =
map_of_futures
.entrySet()
.stream()
.flatMap(
kv -> {
final Validation<BasicMessageBean, String> ret =
kv.getValue()
._2()
.join(); // (must have already returned if
// here
return ret
.<Stream<
Tuple2<
String,
Tuple2<SharedLibraryBean, String>>>>
validation(
// Error:
err -> {
throw new RuntimeException(
err.message());
} // (not ideal, but will do)
,
// Normal:
s -> {
return Arrays.asList(
Tuples._2T(
kv.getKey()._1(),
Tuples._2T(
kv.getValue()._1(),
s)), // result object
// with path_name
Tuples._2T(
kv.getKey()._2(),
Tuples._2T(
kv.getValue()._1(),
s))) // result object
// with id
.stream();
});
})
.collect(
Collectors
.<Tuple2<String, Tuple2<SharedLibraryBean, String>>,
String, Tuple2<SharedLibraryBean, String>>
toMap(
idname_path -> idname_path._1(), // (key)
idname_path -> idname_path._2() // (value)
));
return Validation
.<BasicMessageBean,
Map<String, Tuple2<SharedLibraryBean, String>>>
success(almost_there);
} catch (
Exception e) { // handle the exception thrown above containing the
// message bean from whatever the original error was!
return Validation
.<BasicMessageBean,
Map<String, Tuple2<SharedLibraryBean, String>>>
fail(
SharedErrorUtils.buildErrorMessage(
handler_for_errors.toString(),
msg_for_errors,
e.getMessage()));
}
});
});
} catch (Throwable e) { // (can only occur if the DB call errors)
return CompletableFuture.completedFuture(
Validation.fail(
SharedErrorUtils.buildErrorMessage(
handler_for_errors.toString(),
msg_for_errors,
ErrorUtils.getLongForm(
SharedErrorUtils.ERROR_CACHING_SHARED_LIBS, e, bucket.full_name()))));
}
}
/**
* Talks to the topology module - this top level function just sets the classloader up and creates
* the module, then calls talkToStream to do the talking
*
* @param bucket
* @param libs
* @param harvest_tech_only
* @param m
* @param source
* @return
*/
protected static Validation<BasicMessageBean, IEnrichmentStreamingTopology> getStreamingTopology(
final DataBucketBean bucket,
final BucketActionMessage m,
final String source,
final Validation<BasicMessageBean, Map<String, Tuple2<SharedLibraryBean, String>>>
err_or_libs // "pipeline element"
) {
try {
return err_or_libs.<Validation<BasicMessageBean, IEnrichmentStreamingTopology>>validation(
// Error:
error -> Validation.fail(error),
// Normal
libs -> {
// Easy case, if streaming is turned off, just pass data through this layer
if (!Optional.ofNullable(bucket.streaming_enrichment_topology().enabled()).orElse(true))
return Validation.success(new PassthroughTopology());
// Easy case, if libs is empty then use the default streaming topology
if (libs.isEmpty()) {
return Validation.success(new PassthroughTopology());
}
final Tuple2<SharedLibraryBean, String> libbean_path =
libs.values()
.stream()
.filter(
t2 ->
(null != t2._1())
&& (null
!= Optional.ofNullable(
t2._1().streaming_enrichment_entry_point())
.orElse(t2._1().misc_entry_point())))
.findFirst()
.orElse(null);
if ((null == libbean_path)
|| (null
== libbean_path._2())) { // Nice easy error case, probably can't ever happen
return Validation.fail(
SharedErrorUtils.buildErrorMessage(
source,
m,
SharedErrorUtils.SHARED_LIBRARY_NAME_NOT_FOUND,
bucket.full_name(),
"(unknown)"));
}
final Validation<BasicMessageBean, IEnrichmentStreamingTopology> ret_val =
ClassloaderUtils.getFromCustomClasspath(
IEnrichmentStreamingTopology.class,
Optional.ofNullable(libbean_path._1().streaming_enrichment_entry_point())
.orElse(libbean_path._1().misc_entry_point()),
Optional.of(libbean_path._2()),
libs.values().stream().map(lp -> lp._2()).collect(Collectors.toList()),
source,
m);
return ret_val;
});
} catch (Throwable t) {
return Validation.fail(
SharedErrorUtils.buildErrorMessage(
source,
m,
ErrorUtils.getLongForm(
SharedErrorUtils.ERROR_LOADING_CLASS,
t,
bucket.harvest_technology_name_or_id())));
}
}
protected static CompletableFuture<BucketActionReplyMessage> talkToStream(
final IStormController storm_controller,
final DataBucketBean bucket,
final BucketActionMessage m,
final Validation<BasicMessageBean, IEnrichmentStreamingTopology> err_or_user_topology,
final Validation<BasicMessageBean, Map<String, Tuple2<SharedLibraryBean, String>>> err_or_map,
final String source,
final StreamingEnrichmentContext context,
final String yarn_config_dir,
final String cached_jars_dir) {
try {
// handle getting the user libs
final List<String> user_lib_paths =
err_or_map.<List<String>>validation(
fail -> Collections.emptyList() // (going to die soon anyway)
,
success ->
success
.values()
.stream()
.map(tuple -> tuple._2.replaceFirst("file:", ""))
.collect(Collectors.toList()));
return err_or_user_topology.<CompletableFuture<BucketActionReplyMessage>>validation(
// ERROR getting enrichment topology
error -> {
return CompletableFuture.completedFuture(new BucketActionHandlerMessage(source, error));
},
// NORMAL grab enrichment topology
enrichment_topology -> {
final String entry_point = enrichment_topology.getClass().getName();
context.setBucket(bucket);
context.setUserTopologyEntryPoint(entry_point);
// also set the library bean - note if here then must have been set, else
// IHarvestTechnologyModule wouldn't exist
err_or_map.forEach(
map -> {
context.setLibraryConfig(
map.values()
.stream()
.map(t2 -> t2._1())
.filter(
lib ->
entry_point.equals(lib.misc_entry_point())
|| entry_point.equals(lib.streaming_enrichment_entry_point()))
.findFirst()
.orElse(BeanTemplateUtils.build(SharedLibraryBean.class).done().get()));
// (else this is a passthrough topology, so just use a dummy library bean)
});
_logger.info(
"Set active class="
+ enrichment_topology.getClass()
+ " message="
+ m.getClass().getSimpleName()
+ " bucket="
+ bucket.full_name());
return Patterns.match(m)
.<CompletableFuture<BucketActionReplyMessage>>andReturn()
.when(
BucketActionMessage.DeleteBucketActionMessage.class,
msg -> {
return StormControllerUtil.stopJob(storm_controller, bucket);
})
.when(
BucketActionMessage.NewBucketActionMessage.class,
msg -> {
if (!msg.is_suspended())
return StormControllerUtil.startJob(
storm_controller,
bucket,
context,
user_lib_paths,
enrichment_topology,
cached_jars_dir);
else
return StormControllerUtil.stopJob(
storm_controller,
bucket); // (nothing to do but just do this to return something
// sensible)
})
.when(
BucketActionMessage.UpdateBucketActionMessage.class,
msg -> {
if (msg.is_enabled())
return StormControllerUtil.restartJob(
storm_controller,
bucket,
context,
user_lib_paths,
enrichment_topology,
cached_jars_dir);
else return StormControllerUtil.stopJob(storm_controller, bucket);
})
.when(
BucketActionMessage.TestBucketActionMessage.class,
msg -> {
// TODO (ALEPH-25): in the future run this test with local storm rather than
// remote storm_controller
return StormControllerUtil.restartJob(
storm_controller,
bucket,
context,
user_lib_paths,
enrichment_topology,
cached_jars_dir);
})
.otherwise(
msg -> {
return CompletableFuture.completedFuture(
new BucketActionHandlerMessage(
source,
new BasicMessageBean(
new Date(),
false,
null,
"Unknown message",
0,
"Unknown message",
null)));
});
});
} catch (Throwable e) { // (trying to use Validation to avoid this, but just in case...)
return CompletableFuture.completedFuture(
new BucketActionHandlerMessage(
source,
new BasicMessageBean(
new Date(),
false,
null,
ErrorUtils.getLongForm("Error loading streaming class: {0}", e),
0,
ErrorUtils.getLongForm("Error loading streaming class: {0}", e),
null)));
}
}
/* (non-Javadoc)
* @see akka.actor.AbstractActor#receive()
*/
@Override
public PartialFunction<Object, BoxedUnit> receive() {
return ReceiveBuilder.match(
BucketActionMessage.class,
m ->
!m.handling_clients().isEmpty()
&& !m.handling_clients()
.contains(_context.getInformationService().getHostname()),
__ -> {}) // (do nothing if it's not for me)
.match(
BucketActionOfferMessage.class,
m -> {
_logger.info(
ErrorUtils.get(
"Actor {0} received message {1} from {2} bucket {3}",
this.self(),
m.getClass().getSimpleName(),
this.sender(),
m.bucket().full_name()));
final ActorRef closing_sender = this.sender();
final ActorRef closing_self = this.self();
final String hostname = _context.getInformationService().getHostname();
// (this isn't async so doesn't require any futures)
final boolean accept_or_ignore =
new File(_globals.local_yarn_config_dir() + File.separator + "storm.yaml")
.exists();
final BucketActionReplyMessage reply =
accept_or_ignore
? new BucketActionReplyMessage.BucketActionWillAcceptMessage(hostname)
: new BucketActionReplyMessage.BucketActionIgnoredMessage(hostname);
closing_sender.tell(reply, closing_self);
})
.match(
BucketActionMessage.class,
m -> {
_logger.info(
ErrorUtils.get(
"Actor {0} received message {1} from {2} bucket={3}",
this.self(),
m.getClass().getSimpleName(),
this.sender(),
m.bucket().full_name()));
final ActorRef closing_sender = this.sender();
final ActorRef closing_self = this.self();
final String hostname = _context.getInformationService().getHostname();
// (cacheJars can't throw checked or unchecked in this thread, only from within
// exceptions)
cacheJars(
m.bucket(),
_management_db,
_globals,
_fs,
_context.getServiceContext(),
hostname,
m)
.thenComposeAsync(
err_or_map -> {
final StreamingEnrichmentContext e_context =
_context.getNewStreamingEnrichmentContext();
final Validation<BasicMessageBean, IEnrichmentStreamingTopology>
err_or_tech_module =
getStreamingTopology(m.bucket(), m, hostname, err_or_map);
final CompletableFuture<BucketActionReplyMessage> ret =
talkToStream(
_storm_controller,
m.bucket(),
m,
err_or_tech_module,
err_or_map,
hostname,
e_context,
_globals.local_yarn_config_dir(),
_globals.local_cached_jar_dir());
return ret;
})
.thenAccept(
reply -> { // (reply can contain an error or successful reply, they're the
// same bean type)
// Some information logging:
Patterns.match(reply)
.andAct()
.when(
BucketActionHandlerMessage.class,
msg ->
_logger.info(
ErrorUtils.get(
"Standard reply to message={0}, bucket={1}, success={2}",
m.getClass().getSimpleName(),
m.bucket().full_name(),
msg.reply().success())))
.when(
BucketActionReplyMessage.BucketActionWillAcceptMessage.class,
msg ->
_logger.info(
ErrorUtils.get(
"Standard reply to message={0}, bucket={1}",
m.getClass().getSimpleName(), m.bucket().full_name())))
.otherwise(
msg ->
_logger.info(
ErrorUtils.get(
"Unusual reply to message={0}, type={2}, bucket={1}",
m.getClass().getSimpleName(),
m.bucket().full_name(),
msg.getClass().getSimpleName())));
closing_sender.tell(reply, closing_self);
})
.exceptionally(
e -> { // another bit of error handling that shouldn't ever be called but is a
// useful backstop
// Some information logging:
_logger.warn(
"Unexpected error replying to '{0}': error = {1}, bucket={2}",
BeanTemplateUtils.toJson(m).toString(),
ErrorUtils.getLongForm("{0}", e),
m.bucket().full_name());
final BasicMessageBean error_bean =
SharedErrorUtils.buildErrorMessage(
hostname,
m,
ErrorUtils.getLongForm(
StreamErrorUtils.STREAM_UNKNOWN_ERROR,
e,
m.bucket().full_name()));
closing_sender.tell(
new BucketActionHandlerMessage(hostname, error_bean), closing_self);
return null;
});
})
.build();
}