Apache Camel: Unable to get the Exception Body - apache-camel

Whenever there is normal flow in my Camel Routes I am able to get the body in the next component. But whenever there is an exception(Http 401 or 500) I am unable to get the exception body. I just get a java exception in my server logs.
I have also tried onException().. Using that the flow goes into it on error, but still I do not get the error response body that was sent by the web service(which I get when using POSTMAN directly), I only get the request in the body that I had sent to the web service.
Also adding the route:
from("direct:contractUpdateAds")
.to("log:inside_direct:contractUpdateAds_route_CompleteLog?level=INFO&showAll=true&multiline=true")
.streamCaching()
.setHeader(Exchange.HTTP_METHOD, constant("POST"))
.setHeader(Exchange.CONTENT_TYPE, constant("application/json"))
.log("before calling ADS for ContractUpdate:\nBody:${body}")
.to("{{AdsContractUpdateEndpoint}}")
.log("after calling ADS for ContractUpdate:\nBody:${body}")
.convertBodyTo(String.class)
.end();

Option 1: handle failure status codes yourself
The throwExceptionOnFailure=false endpoint option (available at least for camel-http and camel-http4 endpoints) is probably what you want. With this option, camel-http will no longer consider an HTTP Status >= 300 as an error, and will let you decide what to do - including processing the response body however you see fit.
Something along those lines should work :
from("...")
.to("http://{{hostName}}?throwExceptionOnFailure=false")
.choice()
.when(header(Exchange.HTTP_RESPONSE_CODE).isLessThan(300))
// HTTP status < 300
.to("...")
.otherwise()
// HTTP status >= 300 : would throw an exception if we had "throwExceptionOnFailure=true"
.log("Error response: ${body}")
.to("...");
This is an interesting approach if you want to have special handling for certains status codes for example. Note that the logic can be reused in several routes by using direct endpoints, just like any other piece of Camel route logic.
Option 2 : Access the HttpOperationFailedException in the onException
If you want to keep the default error handling, but you want to access the response body in the exception handling code for some reason, you just need to access the responseBody property on the HttpOperationFailedException.
Here's an example:
onException(HttpOperationFailedException.class)
.process(new Processor() {
#Override
public void process(Exchange exchange) throws Exception {
// e won't be null because we only catch HttpOperationFailedException;
// otherwise, we'd need to check for null.
final HttpOperationFailedException e =
exchange.getProperty(Exchange.EXCEPTION_CAUGHT, HttpOperationFailedException.class);
// Do something with the responseBody
final String responseBody = e.getResponseBody();
}
});

Related

Camel - onException & redelivery

I am trying to implement an error handler & re-oauth mechanism for a Route which performs a GraphQL API call. However, I seem to be misunderstanding the usage of the onException handler and can't seem to get it to work.
Here's the scenario I am trying to cover:
Execute a Route which performs a GraphQL API call (including the setting of variables etc.). The API call itself will always respond with a HTTP 200.
Once the response is received, check the response payload for any error records
If errors exist with a code 401, throw an AuthorizationException
The onException handler catches the exception and is supposed to:
trigger the oauth route to refresh the token (updating the exchangeProperty)
re-run the Route (re-executing the GraphQL call)
After reaching the configured maximumRedeliveries in the onException handler, fail the execution.
Here's the code I have so far:
onException(AuthorizationException.class)
.handled(true)
.redeliveryDelay(1000)
.maximumRedeliveries(3)
.useOriginalMessage()
.to(ROUTE_OAUTH_URI) // updates the token (?)
.to(ROUTE_QUERY_URI) // re-executes the route (?)
;
from(ROUTE_QUERY_URI)
.process(exchange -> {
...set variables
})
.toD("graphql://...")
.setProperty("graphqlResponseErrors").jsonpath("$.errors.length()", true)
.choice()
.when(exchangeProperty("graphqlResponseErrors").isNotEqualTo(0))
.setProperty("graphqlResponseCode").jsonpath("$.errors[0].code", true)
.choice()
.when(exchangeProperty("graphqlResponseCode").isEqualTo(401))
.throwException(new AuthorizationException(
"Authorization exception"))
.end()
.end();
I have tried a few different combinations, including onRedeliver instead of the oauth route, remove the useOriginalMessage and re-trigger the route again but I couldn't get it to work.
Any help or hints would be much appreciated. Thanks!
Update:
Adding the below code to the onException, does retrieve and set a new token, however now the route seems go into an infinite loop:
.onRedelivery(exchange -> {
ProducerTemplate producerTemplate = exchange.getContext().createProducerTemplate();
Exchange oauthExchange = producerTemplate.send(ROUTE_OAUTH_URI, exchange);
exchange.setProperty("accessToken", oauthExchange.getProperty("accessToken"));
});

Keep part of URI encoded in camel route

I am new to camel, so this may be a simple problem to solve.
I have a spring-boot application with camel components which interacts with GitLab API.
My problem is that I need to keep the endpoint URIs in camel routes encoded, for example:
from("direct:start")
.setHeader("PRIVATE-TOKEN",constant("myToken"))
.to("https://gitlab.com/api/v4/projects/12345/repository/files/folder%2Ffile%2Eextension/raw?ref=master")
When the route starts, the message is sent to
"https://gitlab.com/api/v4/projects/12345/repository/files/folder/file.extension/raw?ref=master"
which returns 404, because the parameter file_path has to be encoded, as said in the GitLab doc (I've cheked with a GET from curl: with the first URI a json is returned, with the second 404).
I tried to pass the last part of the URI as HTTP_QUERY, but in this case there is the "?" between it and the URI and I get 404 again:
https://gitlab.com/api/v4/projects/12345/repository/files/?folder%2Ffile%2Eextension/raw?ref=master
I tried adding the URI with the headerHTTP_URI: this time the URI is reached correctly, but I get null body instead of the json answer.
Any idea to solve this issue?
I see that you already tried using HTTP_URI header. How did you set it? Try this:
from("direct:start")
.setHeader("PRIVATE-TOKEN", constant("myToken"))
.setHeader(Exchange.HTTP_URI, simple("https://gitlab.com/api/v4/projects/12345/repository/files/folder%2Ffile%2Eextension/raw?ref=master"))
.to("http:dummy");
This way you set the URI during the route execution, not in endpoint definition. According to docs:
Exchange.HTTP_URI: URI to call. Will override existing URI set directly on the endpoint. This URI is the URI of the HTTP server to call. Its not the same as the Camel endpoint URI, where you can configure endpoint options such as security etc. This header does not support that, its only the URI of the HTTP server.
Don't forget the dependency:
<dependency>
<groupId>org.apache.camel</groupId>
<artifactId>camel-http</artifactId>
</dependency>
The test:
#Override
protected RoutesBuilder createRouteBuilder() throws Exception {
return new RouteBuilder() {
#Override
public void configure() throws Exception {
from("direct:start")
.setHeader("PRIVATE-TOKEN", constant("myToken"))
.setHeader(Exchange.HTTP_URI, simple("http://0.0.0.0:8080?param=folder%2Ffile%2Eextension/raw&ref=master"))
.to("http:dummy");
from("jetty:http://0.0.0.0:8080?matchOnUriPrefix=true")
.setBody(constant("{ key: value }"))
.setHeader(Exchange.CONTENT_TYPE, constant(MediaType.APPLICATION_JSON_VALUE))
.to("mock:result");
}
};
}
#Test
public void test() throws InterruptedException {
getMockEndpoint("mock:result").expectedHeaderReceived(Exchange.HTTP_QUERY, "param=folder%2Ffile%2Eextension/raw&ref=master");
final Exchange response = template.send("direct:start", new Processor() {
public void process(Exchange exchange) throws Exception {
// nothing
}
});
assertThat(response, notNullValue());
assertThat(response.getIn().getHeader(Exchange.HTTP_URI).toString(), containsString("folder%2Ffile%2"));
assertThat(response.getOut().getBody(String.class), containsString("{ key: value }"));
assertMockEndpointsSatisfied();
}
I tried adding the URI with the headerHTTP_URI: this time the URI is reached correctly, but I get null body instead of the json answer.
Keep in mind that the response should be stored at the OUT body:
Camel will store the HTTP response from the external server on the OUT body. All headers from the IN message will be copied to the OUT message, so headers are preserved during routing. Additionally Camel will add the HTTP response headers as well to the OUT message headers.

Camel Reslet Component with async processing

I have a requirement which is as follows:
Accept HTTP POST requests containing XML to a certain URL.
Perform pre-requisite actions such as saving the request XML to a file.
Validate the incoming XML matches the corresponding schema.
If the schema validation fails, synchronously respond with a HTTP 400 response code.
If the schema validation passes, synchronously respond with a HTTP 200 response code.
Pass the XML message on for further processing.
When this further processing completes, asynchronously respond to the caller with a HTTP 200 response code.
This is currently how I have the route configured:
onException(IOException.class)
.log(LoggingLevel.INFO, "Schema validation error on incoming message: ${id}")
.handled(true)
.maximumRedeliveries(0)
.process(schemaValidationErrorProcessor);
from("restlet:http://localhost:" + portNum + "/api/XX/XXX?restletMethod=POST")
.log(LoggingLevel.INFO, "Received message")
.convertBodyTo(String.class)
.multicast()
.parallelProcessing()
.to(SAVE_REQUEST_TO_FILE_QUEUE, PROCESS_PROVISIONING_REQUEST_QUEUE);
from(SAVE_REQUEST_TO_FILE_QUEUE)
.log(LoggingLevel.INFO, "Storing message: ${id}")
.to("file://" + requestLogFolder);
from(PROCESS_PROVISIONING_REQUEST_QUEUE)
.log(LoggingLevel.INFO, "Processing provisioning request: ${id}")
.process(requestGate)
.choice()
.when(header(SYSTEM_STATUS_HEADER).isEqualTo(true))
.unmarshal(xmlParser)
.inOnly("bean:requestHandler?method=handle")
.when(header(SYSTEM_STATUS_HEADER).isEqualTo(false))
.log(LoggingLevel.INFO, "Intentially dropping message")
.endChoice();
The schema validation part is achieved via the .unmarshal(xmlParser) line (I have a JaxbDataFormat object configured elsewhere with the schema set in that). When schema validation fails, an IOException is thrown and this is handled by my schemaValidationErrorProcessor which adds the HTTP 400 to the response.
That is all working fine.
The problem I am having is passing the XML message on for further processing. Basically, I need this to be done asynchronously because when the schema validation passes I need to synchronously respond with a 200 response. The processing that I need to do is in the .inOnly("bean:requestHandler?method=handle") line.
I naively thought that setting the routing to my bean to inOnly would set this to be asynchronous and that main route would not wait for a response. However, this is not the case as when the requestHandler.handle method throws an exception, this is thrown back to the caller of the REST endpoint. I don't want this to happen as I want all of this processing to be done in 'the background' as the consumer will have already received a 200 response.
So, my question is, how would I go about achieving such behaviour? I have thought about using queues etc but ideally would like to avoid such components if possible.
Use Camel Websocket component for asynchronously respond to the caller.
From the Camel documentation:
from("activemq:topic:newsTopic")
.routeId("fromJMStoWebSocket")
.to("websocket://localhost:8443/newsTopic?sendToAll=true&staticResources=classpath:webapp");

camel with netty4-http - EncoderException

I'm using camel 2.14.0 with netty4-http
and I get the following exception.
the scenario is this:
I have a route that sends a request, waits for the response (inOut) and then sends another request.
the first request works, and then the second one fails.
also, if I do it quickly enough after the failure - the first request will also fail.
while debugging a bit (HttpObjectEncoder) - I saw that in the working flow the state of the request is: state = ST_INIT (0)
and in the request that failed it is: ST_CONTENT_NON_CHUNK (1)
which causes the illegal state when the type of message is HttpMessage
is this a bug or is there anything I can configure to fix it?
Caused by: io.netty.handler.codec.EncoderException: java.lang.IllegalStateException: unexpected message type: DefaultFullHttpRequest
at io.netty.handler.codec.MessageToMessageEncoder.write(MessageToMessageEncoder.java:107)
at io.netty.channel.CombinedChannelDuplexHandler.write(CombinedChannelDuplexHandler.java:192)
at io.netty.channel.AbstractChannelHandlerContext.invokeWrite(AbstractChannelHandlerContext.java:658)
at io.netty.channel.AbstractChannelHandlerContext.access$2000(AbstractChannelHandlerContext.java:32)
at io.netty.channel.AbstractChannelHandlerContext$AbstractWriteTask.write(AbstractChannelHandlerContext.java:939)
at io.netty.channel.AbstractChannelHandlerContext$WriteAndFlushTask.write(AbstractChannelHandlerContext.java:991)
at io.netty.channel.AbstractChannelHandlerContext$AbstractWriteTask.run(AbstractChannelHandlerContext.java:924)
at io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:380)
at io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:357)
at io.netty.util.concurrent.SingleThreadEventExecutor$2.run(SingleThreadEventExecutor.java:116)
at java.lang.Thread.run(Thread.java:745)
Caused by: java.lang.IllegalStateException: unexpected message type: DefaultFullHttpRequest
at io.netty.handler.codec.http.HttpObjectEncoder.encode(HttpObjectEncoder.java:63)
at io.netty.handler.codec.http.HttpClientCodec$Encoder.encode(HttpClientCodec.java:106)
at io.netty.handler.codec.MessageToMessageEncoder.write(MessageToMessageEncoder.java:89)
... 10 more
I managed to identify the problem:
the first request I sent was a GET request with null body.
in the class org.apache.camel.component.netty4.http.NettyHttpProducer -
the method getRequestBody(Exchange exchange) is creating the actual request object from the exchange.
in it - the method "toNettyRequest" in class org.apache.camel.component.netty4.http.DefaultNettyHttpBinding
checks if the body is null, and if so - it is creating a DefaultHttpRequest, and not DefaultHttpFullRequest
when the request reaches the encoder as a result of a writeAndFlush call - the encoder does not clean its state because of this part of the code:
if (msg instanceof LastHttpContent) {
state = ST_INIT;
}
the DefaultHttpRequest is not instanceof LastHttpContent, so the state remains ST_CONTENT_NON_CHUNK and the next request will get an IllegalStateException because the state is not ST_INIT
this bug did not exist in netty-http, it only happened when I moved to use netty4-http
the workaround is simple - use an empty String ("") as body

Camel message redelivery not behaving as expected

I have a route in Camel that I want to retry when an exception occurs, but I want to set a property so that the route can do something slightly differently the second time to try to stop the error happening again on the retry. Here's a route that illustrates the idea I'm trying at the moment.
from("direct:onExceptionTest")
.onException(Exception.class)
.maximumRedeliveries(1)
.log("Retrying")
.setProperty("retrying", constant(true))
.end()
.log("Start")
.choice()
.when(property("retrying").isNull())
.log("Throwing")
.throwException(new Exception("Hello world"))
.end()
.end()
.log("Done")
Obviously this isn't the real route; the whole choice body just simulates my component erroring in certain cases. I'm expecting to see the following messages logged:
Start
Throwing
Retrying
Start
Done
But what I actually see is:
Start
Throwing
Retrying
Failed delivery for (MessageId: ... on ExchangeId: ...). Exhausted after delivery attempt: 2 caught: java.lang.Exception: Hello world. Processed by failure processor: FatalFallbackErrorHandler[Pipeline[[Channel[Log(onExceptionTest)[Retrying]], Channel[setProperty(retrying, true)]]]]
I've tried adding handled(true) to the exception handler, but all this does is suppress the error message. I don't see the second Start or Done log message.
Why doesn't my route behave as I expect, and what do I need to do to get it to behave the way I want?
Update
#ProgrammerDan points out that the problem is that redelivery isn't intended for what I'm trying to achieve, which would explain why my route doesn't work! So I need to do the work in my handler, but my route calls a web service and has a few other steps and I don't want to duplicate all this in the handler. I've come up with this, which works as expected but it involves the route calling itself again from the start. Is this a bad idea? Will I get myself into knots with this approach?
from("direct:onExceptionTest")
.onException(Exception.class)
.onWhen(property("retrying").isNull()) // don't retry forever
.log("Retrying")
.setProperty("retrying", constant(true))
.handled(true)
.to("direct:onExceptionTest") // is recursion bad?
.end()
.log("Start")
.choice()
.when(property("retrying").isNull())
.log("Throwing")
.throwException(new Exception("Hello world"))
.end()
.end()
.log("Done")
Use onRedelivery with a Processor to set the property:
String KEY = "retrying";
from("direct:onExceptionTest")
.onException(RuntimeException.class)
.onRedelivery(new Processor() { // Sets a processor that should be processed before a redelivery attempt.
#Override
public void process(final Exchange exchange) throws Exception {
LOG.info("Retrying");
exchange.setProperty(KEY, true);
}
})
.maximumRedeliveries(1)
.handled(true)
.end()
.log("Start")
.process(new Processor() {
#Override
public void process(final Exchange exchange) throws Exception {
LOG.info("No problem");
}
})
.process(new Processor() {
#Override
public void process(final Exchange exchange) throws Exception {
if (exchange.getProperty(KEY) == null) {
LOG.info("Throwing");
throw new RuntimeException("Hello World");
}
else {
LOG.info("No throwing");
}
}
})
.log("Done");
This prints
[ main] route1 INFO Start
[ main] OnExceptionHandler INFO No problem
[ main] OnExceptionHandler INFO Throwing
[ main] OnExceptionHandler INFO Retrying
[ main] OnExceptionHandler INFO No throwing
[ main] route1 INFO Done
As #ProgrammerDan noted, only the processor that failed is re-executed but not the first processor that passed without any problems.
Edit:
If all the processing has to be re-done then you may use a sub-route with doTry and doCatch as follows:
from("direct:onExceptionTest")
.doTry()
.to("direct:subroute")
.doCatch(RuntimeException.class)
.setProperty(KEY, constant(true))
.to("direct:subroute")
.end()
.log("Done");
from("direct:subroute")
.log("Start")
.process(new Processor() {
#Override
public void process(final Exchange exchange) throws Exception {
LOG.info("No problem");
}
})
.process(new Processor() {
#Override
public void process(final Exchange exchange) throws Exception {
if (exchange.getProperty(KEY) == null) {
LOG.info("Throwing");
throw new RuntimeException("Hello World");
}
else {
LOG.info("No throwing");
}
}
});
From the Camel Docs:
When using doTry .. doCatch .. doFinally then the regular Camel Error Handler does not apply. That means any onException or the likes does not trigger. The reason is that doTry .. doCatch .. doFinally is in fact its own error handler and that it aims to mimic and work like how try/catch/finally works in Java.
Couple of points to consider about Camel's redelivery mechanism. First, check out the docs on the topic which might challenge your assumptions about how Camel handles redelivery. The point I've linked to is that Camel attempts redelivery at point of failure, it does not start over from the beginning of the route (as you appear to assume). If I'm understanding the docs correctly (I haven't tried this pattern in a while) you are basically telling it to retry throwing an exception several times, which I doubt is what you want to test.
Second, I'd recommend just doing the alternate handling directly in the onException() processor chain, as demonstrated a little further down in the same docs. Basically, you could specify how you want the message handled via a custom processor, and use both handled(true) and stop() to indicate that no further processing is necessary.
To sum it up, redelivery is generally meant to handle typical endpoint delivery failures, like intermittent connectivity drops, receiving server momentary unavailability, etc. where it makes the most sense to just "try again" and have a reasonable expectation of success. If you need more complex logic to handle retries, use a custom processor or series of processors within your onException() processor chain.

Resources