I have a gRPC server written in C++ and a client written in Java.
Everything was working fine using a blocking stub. Then I decided that I want to change one of the calls to be asynchronous, so I created an additional stub in my client, this one is created with newStub(channel) as opposed to newBlockingStub(channel). I didn't make any changes on the server side. This is a simple unary RPC call.
So I changed
Empty response = blockingStub.callMethod(request);
to
asyncStub.callMethod(request, new StreamObserver<Empty>() {
#Override
public void onNext(Empty response) {
logInfo("asyncStub.callMethod.onNext");
}
#Override
public void onError(Throwable throwable) {
logError("asyncStub.callMethod.onError " + throwable.getMessage());
}
#Override
public void onCompleted() {
logInfo("asyncStub.callMethod.onCompleted");
}
});
Ever since then, onError is called when I use this RPC (Most of the time) and the error it gives is "CANCELLED: io.grpc.Context was cancelled without error". I read about forking Context objects when making an RPC call from within an RPC call, but that's not the case here. Also, the Context seems to be a server side object, I don't see how it relates to the client. Is this a server side error propagating back to the client? On the server side everything seems to complete successfully, so I'm at a loss as to why this is happening. Inserting a 1ms sleep after calling asyncStub.callMethod seems to make this issue go away, but defeats the purpose. Any and all help in understanding this would be greatly appreciated.
Some notes:
The processing time on the server side is around 1 microsecond
Until now, the round trip time for the blocking call was several hundred microseconds (This is the time I'm trying to cut down, as this is essentially a void function, so I don't need to wait for a response)
This method is called multiple times in a row, so before it used to wait until the previous one finished, now they just fire off one after the other.
Some snippets from the proto file:
service EventHandler {
rpc callMethod(Msg) returns (Empty) {}
}
message Msg {
uint64 fieldA = 1;
int32 fieldB = 2;
string fieldC = 3;
string fieldD = 4;
}
message Empty {
}
So it turns out that I was wrong. The context object is used by the client too.
The solution was to do the following:
Context newContext = Context.current().fork();
Context origContext = newContext.attach();
try {
// Call async RPC here
} finally {
newContext.detach(origContext);
}
Hopefully this can help someone else in the future.
Related
I am trying to solve an issue whereby I need to combine third party events with the eventBus send and reply approach that Vertx provides for Standard and Worker Verticle setups. I am not sure if what I have laid out below is necessarily the correct approach.
Problem Statement:
I want to have standard verticle that sends a message to a worker verticle.
The worker verticle does some preprocessing and then uses a client method provided by a third party state management lib to publish an even (in an async manner). The result of which is only whether or not the event was successfully received or not (but does not contain any further info around processing etc).
Further processing takes place when the third party state management lib receives the event(this all happens on a separate thread) and a success or failure can occur at which point another event will be published to the cluster management tools output channel.
From the output channel listener I then want to be able to use the event to somehow use the message.reply() on the worker verticle to send back a response to the standard verticle that made the original request, thereby closing the loop of the entire request lifecycle but also using the async approach that vertx is built to use.
Now I conceptually know how to do 90% of what is described here but the missing piece for me is how to coordinate the event on the output channel listener and connect this to the worker verticle so that I can trigger the message.reply.
I have looked at possibly using SharedData and Clustering that Vertx has but was wondering if there is possibly another approach.
I have put a possible example implementation but would really appreciate if anyone has any insights/thoughts into how this can be accomplished and if I am on the right track.
class Order(val id: String)
class OrderCommand(val order: Order) : Serializable {
companion object {
const val name = "CreateOrderCommand"
}
}
class SuccessOrderEvent(val id: String) : Serializable {
companion object {
const val name = "OrderSuccessfulEvent"
}
}
interface StateManagementLib {
fun <T> send(
value: T,
success: Handler<AsyncResult<Int>>,
failure: Handler<AsyncResult<Exception>>
) {
val output = publish(value)
if (output == 1) {
success.handle(Future.succeededFuture())
} else {
failure.handle(Future.failedFuture("Failed"))
}
}
// non-blocking
fun <T> publish(value: T): Int // returns success/failure only
}
class WorkVerticle constructor(private val lib: StateManagementLib) : AbstractVerticle() {
override fun start(startPromise: Promise<Void>) {
workerHandler()
startPromise.complete()
}
private fun workerHandler() {
val consumer = vertx.eventBus().consumer<OrderCommand>(OrderCommand.name)
consumer.handler { message: Message<OrderCommand> ->
try {
vertx.sharedData().getClusterWideMap<String, Message<OrderCommand>>("OrderRequest") { mapIt ->
if (mapIt.succeeded()) {
lib.send(message.body(), {
// The StateManagementLib successfully propagated the event so, we try and store in this map (id -> Message)
mapIt.result().put(message.body().order.id, message)
}, {
message.fail(400, it.result().message)
})
}
}
} catch (e: Exception) {
message.fail(
HttpResponseStatus.INTERNAL_SERVER_ERROR.code(), "Failed to encode data."
)
}
}
// A consumer that will pick up an event that is received from the clusterTool output channel
vertx.eventBus().addInboundInterceptor { context: DeliveryContext<SuccessOrderEvent> ->
// Retrieve cluster map to get the previously stored message and try to respond with Message.reply
// This should go back to the Standard Verticle that sent the message
vertx.sharedData().getClusterWideMap<String, Message<OrderCommand>>("OrderRequest") {
if (it.succeeded()) {
val id = context.message().body().id
val mapResult = it.result().get(id)
it.result().remove(id)
// Try and reply so the original eventloop thread can pickup and respond to calling client
mapResult.result().reply(id)
}
}
}
}
}
I'm quite new to the Microservice world and particularly vertX. I want my verticle to start anyway even there is no database connection available (e.g. database URL missing in configuration). I already managed to do this and my verticle is starting.
The issue now is that I want my verticle to notice when the database connection is available again and connect to it. How can I do this ?
I thought about creating another Verticle "DatabaseVerticle.java" which would send the current DB config on the event bus and my initial verticle would consume this message and check whether the config info is consistent (reply with success) or still missing some data (reply with fail and make the DatabaseVerticle check again).
This might work (and might not) but does not seem to be the optimal solution for me.
I'd be very glad if someone could suggest a better solution. Thank you !
For your use case, I'd recommend to use the vertx-config. In particular, have a look at the Listening to configuration changes section of the Vert.x Config documentation.
You could create a config retriever and set a handler for changes:
ConfigRetrieverOptions options = new ConfigRetrieverOptions()
.setScanPeriod(2000)
.addStore(myConfigStore);
ConfigRetriever retriever = ConfigRetriever.create(vertx, options);
retriever.getConfig(json -> {
// If DB config available, start the DB client
// Otherwise set a "dbStarted" variable to false
});
retriever.listen(change -> {
// If "dbStarted" is still set to false
// Check the config and start the DB client if possible
// Set "dbStarted" to true when done
});
The ideal way would be some other service telling your service about database connection. Either through event bus or HTTP, what you can do is when someone tries to access your database when connection is not made just try to make some DB call and handle the exception, return a boolean as false. Now when you get a message on event bus, consume it and save it in some config pojo. Now when someone tries to access your database, look for config and if available make a connection.
Your consumer:
public void start(){
EventBus eb = vertx.eventBus();
eb.consumer("database", message -> {
config.setConfig(message.body());
});
}
Your db client(Mongo for this eg):
public class MongoService{
private MongoClient client;
public boolean isAvailable = false;
MongoService(Vertx vertx){
if(config().getString("connection")){
client = MongoClient.createShared(vertx, config().getString("connection"));
isAvailable = true;
}
}
}
Not everything in Vertx should be solved by another verticle.
In this case, you can use .periodic()
http://vertx.io/docs/vertx-core/java/#_don_t_call_us_we_ll_call_you
I assume you have some function that checks the DB for the first time.
Let's call it checkDB()
class PeriodicVerticle extends AbstractVerticle {
private Long timerId;
#Override
public void start() {
System.out.println("Started");
// Should be called each time DB goes offline
final Long timerId = this.vertx.setPeriodic(1000, (l) -> {
final boolean result = checkDB();
// Set some variable telling verticle that DB is back online
if (result) {
cancelTimer();
}
});
setTimerId(timerId);
}
private void cancelTimer() {
System.out.println("Cancelling");
getVertx().cancelTimer(this.timerId);
}
private void setTimerId(final Long timerId) {
this.timerId = timerId;
}
}
Here I play a bit with timerId, since we cannot pass it to cancelTimer() right away. But otherwise, it's quite simple.
I m sending several retrofit calls via SyncAdapter onPerformSync and I m trying to regulate http calls by sending out via a try/catch sleep statement. However, this is blocking the UI and will be not responsive only after all calls are done.
What is a better way to regulate network calls (with a sleep timer) in background in onPerformSync without blocking UI?
#Override
public void onPerformSync(Account account, Bundle extras, String authority, ContentProviderClient provider, SyncResult syncResult) {
String baseUrl = BuildConfig.API_BASE_URL;
Retrofit retrofit = new Retrofit.Builder()
.baseUrl(baseUrl)
.addConverterFactory(GsonConverterFactory.create())
.build();
service = retrofit.create(HTTPService.class);
Call<RetroFitModel> RetroFitModelCall = service.getRetroFit(apiKey, sortOrder);
RetroFitModelCall.enqueue(new Callback<RetroFitModel>() {
#Override
public void onResponse(Response<RetroFitModel> response) {
if (!response.isSuccess()) {
} else {
List<RetroFitResult> retrofitResultList = response.body().getResults();
Utility.storeList(getContext(), retrofitResultList);
for (final RetroFitResult result : retrofitResultList) {
RetroFitReview(result.getId(), service);
try {
// Sleep for SLEEP_TIME before running RetroFitReports & RetroFitTime
Thread.sleep(SLEEP_TIME);
} catch (InterruptedException e) {
}
RetroFitReports(result.getId(), service);
RetroFitTime(result.getId(), service);
}
}
}
#Override
public void onFailure(Throwable t) {
Log.e(LOG_TAG, "Error: " + t.getMessage());
}
});
}
}
The "onPerformSync" code is executed within the "SyncAdapterThread" thread, not within the Main UI thread. However this could change when making asynchronous calls with callbacks (which is our case here).
Here you are using an asynchronous call of the Retrofit "call.enqueue" method, and this has an impact on thread execution. The question we need to ask at this point:
Where callback methods are going to be executed?
To get the answer to this question, we have to determine which Looper is going to be used by the Handler that will post callbacks.
In case we are playing with handlers ourselves, we can define the looper, the handler and how to process messages/runnables between handlers. But this time it is different because we are using a third party framework (Retrofit). So we have to know which looper used by Retrofit?
Please note that if Retrofit didn't already define his looper, you
could have caught an exception saying that you need a looper to
process callbacks. In other words, an asynchronous call needs to be in
a looper thread in order to post callbacks back to the thread from
where it was executed.
According to the code source of Retrofit (Platform.java):
static class Android extends Platform {
#Override CallAdapter.Factory defaultCallAdapterFactory(Executor callbackExecutor) {
if (callbackExecutor == null) {
callbackExecutor = new MainThreadExecutor();
}
return new ExecutorCallAdapterFactory(callbackExecutor);
}
static class MainThreadExecutor implements Executor {
private final Handler handler = new Handler(Looper.getMainLooper());
#Override public void execute(Runnable r) {
handler.post(r);
}
}
}
You can notice "Looper.getMainLooper()", which means that Retrofit will post messages/runnables into the main thread message queue (you can do research on this for further detailed explanation). Thus the posted message/runnable will be handled by the main thread.
So that being said, the onResponse/onFailure callbacks will be executed in the main thread. And it's going to block the UI, if you are doing too much work (Thread.sleep(SLEEP_TIME);). You can check it by yourself: just make a breakpoint in "onResponse" callback and check in which thread it is running.
So how to handle this situation? (the answer to your question about Retrofit use)
Since we are already in a background thread (SyncAdapterThread), so there is no need to make asynchronous calls in your case. Just make a Retrofit synchronous call and then process the result, or log a failure. This way, you will not block the UI.
My app receives an HTTP request, and in the middle of the pipeline, a call is made to another server for supporting information. The initial HTTP request can't continue through the pipeline until that response comes back. I can't use the awaitUninterruptability() from an I/O thread, so what's the best approach to make these calls so I don't block Netty's event loop, but put the client's pipeline on hold until my call out returns and I tell the pipeline to continue on?
Ryan this does not sound like a good idea..
I think you should better use something like that:
public class HttpHandler extends SimpleChannelUpstreamHandler{
#Override
public void messageReceived(final ChannelHandlerContext ctx, final MessageEvent e) throws Exception {
otherChannel.write(yourRequet).addListener(new ChannelFutureListener() {
public void operationComplete(ChannelFuture future) throws Exception {
// once the write is done we can continue in the pipeline
ctx.sendUpstream(e);
}
});
// the event stops here to get processed
}
}
If you need to wait for the response then you will need to handle it in another SimpleChannelUpstreamHandler. But I think you get the idea..
I guess you need an ExecutionHandler.
I am developing custom HTTP server with netty 3.3.1.
I need to implement something like this
HTTP Server receives request
HTTP Server parses it and invokes HTTP request as a client to other machine
HTTP Server waits for the response of request sent in (2)
HTTP Server sends response to request from (1) based on what had received in (3)
It means that client request (2) has to behave as synchronous.
What I wrote is based on HttpSnoopClient example but it does not work, because I receive
java.lang.IllegalStateException:
await*() in I/O thread causes a dead lock or sudden performance drop. Use addListener() instead or call await*() from a different thread.
I've refactored the code from the example mentioned above and now it looks more less like this (starting from line 7f of HttpSnoopClient):
ChannelFuture future = bootstrap.connect(new InetSocketAddress(host, port));
future.addListener(new ChannelFutureListener() {
public void operationComplete(ChannelFuture future) {
if (!future.isSuccess()) {
System.err.println("Cannot connect");
future.getCause().printStackTrace();
bootstrap.releaseExternalResources();
return;
}
System.err.println("Connected");
Channel channel = future.getChannel();
// Send the HTTP request.
channel.write(request);
channel.close();
// Wait for the server to close the connection.
channel.getCloseFuture().addListener(new ChannelFutureListener() {
public void operationComplete(ChannelFuture future) {
System.err.println("Disconnected");
bootstrap.releaseExternalResources(); // DOES NOT WORK?
}
});
}
});
}
}
The run() command from the above example is invoked in the messageReceived function of my herver handler.
So it became asynchronous and avoid await* functions. Request is invoked properly. But - for uknown reason for me - the line
bootstrap.releaseExternalResources(); // DOES NOT WORK?
does not work. It throws an exception saying that I cannot kill the thread I am currently using (which sounds reasonable, but still does not give me an answer how to do that in a different way).
I am also not sure is this a correct approach?
Maybe you can recommend a tutorial of such event programming techniques in netty? How to deal - in general - with a few asynchronous requests that has to be invoked in specified order and wait for each other?
Thank you,
If you really want to release the bootstrap on close you can do it like this:
channel.getCloseFuture().addListener(new ChannelFutureListener() {
public void operationComplete(ChannelFuture future) {
System.err.println("Disconnected");
new Thread(new Runnable() {
public void run() {
bootstrap.releaseExternalResources();
}
}).start();
}
});