/** The actor constructor - at some point all these things should be inserted by injection */
public DataBucketChangeActor() {
_context = DataImportActorContext.get();
_core_distributed_services = _context.getDistributedServices();
_actor_system = _core_distributed_services.getAkkaSystem();
_management_db = _context.getServiceContext().getCoreManagementDbService().readOnlyVersion();
_globals = _context.getGlobalProperties();
_fs = _context.getServiceContext().getStorageService();
_storm_controller = _context.getStormController();
}
@Test
public void test_actor()
throws IllegalArgumentException, InterruptedException, ExecutionException, TimeoutException,
IOException {
// Create a bucket
final DataBucketBean bucket = createBucket("test_actor");
// Create an actor:
final ActorRef handler =
_db_actor_context
.getActorSystem()
.actorOf(Props.create(DataBucketChangeActor.class), "test_host");
_db_actor_context
.getStreamingEnrichmentMessageBus()
.subscribe(handler, ActorUtils.STREAMING_ENRICHMENT_EVENT_BUS);
// create the inbox:
final Inbox inbox = Inbox.create(_actor_context.getActorSystem());
// 1) A message that it will ignore because it's not for this actor
{
final BucketActionMessage.DeleteBucketActionMessage delete =
new BucketActionMessage.DeleteBucketActionMessage(
bucket, new HashSet<String>(Arrays.asList("a", "b")));
inbox.send(handler, delete);
try {
inbox.receive(Duration.create(1L, TimeUnit.SECONDS));
fail("should have timed out");
} catch (Exception e) {
assertEquals(TimeoutException.class, e.getClass());
}
}
// 2a) Send an offer (ignored, no storm properties)
{
try {
new File(
_service_context.getGlobalProperties().local_yarn_config_dir()
+ File.separator
+ "storm.yaml")
.delete();
} catch (Exception e) {
// (don't care if fails, probably just first time through)
}
final BucketActionMessage.BucketActionOfferMessage broadcast =
new BucketActionMessage.BucketActionOfferMessage(bucket);
_db_actor_context
.getStreamingEnrichmentMessageBus()
.publish(new BucketActionEventBusWrapper(inbox.getRef(), broadcast));
final Object msg = inbox.receive(Duration.create(5L, TimeUnit.SECONDS));
assertEquals(BucketActionReplyMessage.BucketActionIgnoredMessage.class, msg.getClass());
}
// 2b) Send an offer (accepted, create file)
{
new File(
_service_context.getGlobalProperties().local_yarn_config_dir()
+ File.separator
+ "storm.yaml")
.createNewFile();
final BucketActionMessage.BucketActionOfferMessage broadcast =
new BucketActionMessage.BucketActionOfferMessage(bucket);
_db_actor_context
.getStreamingEnrichmentMessageBus()
.publish(new BucketActionEventBusWrapper(inbox.getRef(), broadcast));
final Object msg = inbox.receive(Duration.create(5L, TimeUnit.SECONDS));
assertEquals(BucketActionReplyMessage.BucketActionWillAcceptMessage.class, msg.getClass());
}
// 3) Send a message, currently just check it arrives back
{
final BucketActionMessage.UpdateBucketActionMessage update =
new BucketActionMessage.UpdateBucketActionMessage(
bucket,
false,
bucket,
new HashSet<String>(
Arrays.asList(_actor_context.getInformationService().getHostname())));
final CompletableFuture<BucketActionReplyMessage> reply4 =
AkkaFutureUtils.efficientWrap(
Patterns.ask(handler, update, 5000L),
_db_actor_context.getActorSystem().dispatcher());
@SuppressWarnings("unused")
final BucketActionReplyMessage msg4 = reply4.get();
}
}
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) {
}
}
}
/* (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();
}