We Keep Coding

Webflux, reactor connection reset by peer unhandled

So basically I am simulating a 'Connection reset by peer' locally using Toxiproxy or Wiremock (same behaviour for both of them) and I get the exception below.
The tricky part is that it will bypass any defined exception handlers and there's no way to revert the flow if there's a failure.
I tried configuring the WebClient as mentioned here: https://github.com/reactor/reactor-netty/issues/388 by either removing the connection pool or simply setting the HttpClient keepAlive property to false.
I defined a global exception handling mechanism by extending ErrorWebExceptionHandler but it doesn't go through.
Do you have any suggestions on how to manage this one? Or a proper configuration which I could try for either TcpClient / HttpClient or WebClient? What am I missing?
There must be a way to handle this properly..
[26/04/22 20:02:28] lvl=ERROR [ioEventLoopGroup-4-3] r.n.r.PooledConnectionProvider - [id: 0xaf559a88, L:0.0.0.0/0.0.0.0:56755 ! R:localhost/127.0.0.1:8989] Pooled connection observed an error
java.io.IOException: Connection reset by peer
at java.base/sun.nio.ch.FileDispatcherImpl.read0(Native Method)
at java.base/sun.nio.ch.SocketDispatcher.read(SocketDispatcher.java:39)
at java.base/sun.nio.ch.IOUtil.readIntoNativeBuffer(IOUtil.java:276)
at java.base/sun.nio.ch.IOUtil.read(IOUtil.java:233)
at java.base/sun.nio.ch.IOUtil.read(IOUtil.java:223)
at java.base/sun.nio.ch.SocketChannelImpl.read(SocketChannelImpl.java:358)
at io.netty.buffer.PooledByteBuf.setBytes(PooledByteBuf.java:247)
at io.netty.buffer.AbstractByteBuf.writeBytes(AbstractByteBuf.java:1140)
at io.netty.channel.socket.nio.NioSocketChannel.doReadBytes(NioSocketChannel.java:347)
at io.netty.channel.nio.AbstractNioByteChannel$NioByteUnsafe.read(AbstractNioByteChannel.java:148)
at io.netty.channel.nio.NioEventLoop.processSelectedKey(NioEventLoop.java:697)
at io.netty.channel.nio.NioEventLoop.processSelectedKeysOptimized(NioEventLoop.java:632)
at io.netty.channel.nio.NioEventLoop.processSelectedKeys(NioEventLoop.java:549)
at io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:511)
at io.netty.util.concurrent.SingleThreadEventExecutor$5.run(SingleThreadEventExecutor.java:918)
at io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)
at io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)

java.net.BindException: Address already in use: connect (I did read the others before posting)

I know there are a ton of these posts, but this is a little different. We are using vended code for part of our data processing system, and part of the system sends emails to clients if certain events take place on data insertion or deletion. Recently we have started getting address already in use exceptions. We checked the repository history, and nothing has changed in our code in the last 6 months for this system. We have already tried the typical solutions for this issue including increasing the number of connections allowed to the port with little success. We had a meeting with the vendor, and I asked if anything had changed in their code, and if they would assure that all connections in their code are explicitly closed. They indicated that they are explicitly closing all sockets. However they didn't show us the code so there is no way for us to know if this is true other than taking their word for it. So, the only thing I can think of to do is continue to increase the number of connections to the port until we stop getting bind exceptions. So, what is the industry standard for max number of connections to port 25; is there one? Also if anyone has any other suggestions I would greatly appreciate it? Thanks so much in advance, Robert
20210505112127.716 ERROR m.fiserv.ppx.business.notification.EmailNotifier : MessagingException from notify
javax.mail.MessagingException: Could not connect to SMTP host: SERVER.URL.COM, port: 25;
nested exception is:
java.net.BindException: Address already in use: connect
at com.sun.mail.smtp.SMTPTransport.openServer(SMTPTransport.java:1545)
at com.sun.mail.smtp.SMTPTransport.protocolConnect(SMTPTransport.java:453)
Caused by:
java.net.BindException: Address already in use: connect
at java.net.DualStackPlainSocketImpl.connect0(Native Method)
at java.net.DualStackPlainSocketImpl.socketConnect(DualStackPlainSocketImpl.java:90)
at java.net.AbstractPlainSocketImpl.doConnect(AbstractPlainSocketImpl.java:380)
20210505131529.950 ERROR erv.ppx.web.controller.AuditReportViewController : Error while generating HTML
net.sf.jasperreports.engine.JRException: Error writing to OutputStream writer : CorpAdminAuditReport
at net.sf.jasperreports.engine.export.JRHtmlExporter.exportReport(JRHtmlExporter.java:496)
at com.fiserv.ppx.web.controller.AuditReportViewController.generateReport(AuditReportViewController.java:184)
Caused by:
com.ibm.wsspi.webcontainer.ClosedConnectionException: OutputStream encountered error during write
at com.ibm.ws.webcontainer.channel.WCCByteBufferOutputStream.write(WCCByteBufferOutputStream.java:188)
at com.ibm.ws.webcontainer.srt.SRTOutputStream.write(SRTOutputStream.java:97)
20210505140706.240 ERROR com.fiserv.ppx.business.db.DBConnectionUtil : Exception in getting for AppServer connection from DataSource.
com.ibm.websphere.ce.cm.ConnectionWaitTimeoutException: J2CA1010E: Connection not available; timed out waiting for 180,005 seconds.
at com.ibm.ws.rsadapter.AdapterUtil.toSQLException(AdapterUtil.java:1680)
at com.ibm.ws.rsadapter.jdbc.WSJdbcDataSource.getConnection(WSJdbcDataSource.java:661)
at com.ibm.ws.rsadapter.jdbc.WSJdbcDataSource.getConnection(WSJdbcDataSource.java:611)
Caused by:
com.ibm.websphere.ce.j2c.ConnectionWaitTimeoutException: J2CA1010E: Connection not available; timed out waiting for 180,005 seconds.
at com.ibm.ejs.j2c.FreePool.createOrWaitForConnection(FreePool.java:1781)
at com.ibm.ejs.j2c.PoolManager.reserve(PoolManager.java:3834)
at com.ibm.ejs.j2c.PoolManager.reserve(PoolManager.java:3082)
20210505140731.341 ERROR com.fiserv.ppx.business.db.DBConnectionUtil : Exception in getting for AppServer connection from DataSource.
com.ibm.websphere.ce.cm.ConnectionWaitTimeoutException: J2CA1010E: Connection not available; timed out waiting for 180,010 seconds.
at com.ibm.ws.rsadapter.AdapterUtil.toSQLException(AdapterUtil.java:1680)
at com.ibm.ws.rsadapter.jdbc.WSJdbcDataSource.getConnection(WSJdbcDataSource.java:661)
at com.ibm.ws.rsadapter.jdbc.WSJdbcDataSource.getConnection(WSJdbcDataSource.java:611)
Caused by:
com.ibm.websphere.ce.j2c.ConnectionWaitTimeoutException: J2CA1010E: Connection not available; timed out waiting for 180,010 seconds.
at com.ibm.ejs.j2c.FreePool.createOrWaitForConnection(FreePool.java:1781)
at com.ibm.ejs.j2c.PoolManager.reserve(PoolManager.java:3904)
at com.ibm.ejs.j2c.PoolManager.reserve(PoolManager.java:3082)
20210505140731.341 ERROR com.fiserv.ppx.sso.controller.SSOController : SSO Configuration error
java.lang.NullPointerException
at com.fiserv.ppx.business.db.PPXDbTransactionManager.<init>(PPXDbTransactionManager.java:60)
at com.fiserv.ppx.sso.impl.SSOLoginAuthenticator.authenticateSSOUser(SSOLoginAuthenticator.java:157)

How to handle in Reactor Netty an io.netty.channel.ConnectTimeoutException

I'm trying to use Reactor Netty TcpClient in reactive way to interact with hosts, that may be unreachable. Here is an example of a channel initialization logic:
ConnectionProvider connectionProvider = ConnectionProvider.fixed("fixed", 50);
TcpClient.create(connectionProvider)
.host(host).port(port)
.wiretap(true)
.option(ChannelOption.CONNECT_TIMEOUT_MILLIS, 50)
.doOnConnect(x -> log.trace("Connect to {}:{}", host, port))
.doOnConnected(conn -> log.trace("Connected {}", conn.channel()))
.connect()
.subscribe(this::utilizeConnection);
the output, that i receiving :
2019-09-04 08:23:13.612 TRACE 71988 --- [ioEventLoop-4-3] c.c.pcb.poc.network.tcp.NettyTcpSender : Connect to 192.168.88.210:2000
2019-09-04 08:23:13.684 WARN 71988 --- [actor-tcp-nio-4] io.netty.util.concurrent.DefaultPromise : An exception was thrown by reactor.netty.resources.PooledConnectionProvider$DisposableAcquire.operationComplete()
reactor.core.Exceptions$ErrorCallbackNotImplemented: io.netty.channel.ConnectTimeoutException: connection timed out: /192.168.88.210:2000
Caused by: io.netty.channel.ConnectTimeoutException: connection timed out: /192.168.88.210:2000
at io.netty.channel.nio.AbstractNioChannel$AbstractNioUnsafe$1.run(AbstractNioChannel.java:267) ~[netty-transport-4.1.36.Final.jar:4.1.36.Final]
at io.netty.util.concurrent.PromiseTask$RunnableAdapter.call(PromiseTask.java:38) ~[netty-common-4.1.36.Final.jar:4.1.36.Final]
at io.netty.util.concurrent.ScheduledFutureTask.run(ScheduledFutureTask.java:127) ~[netty-common-4.1.36.Final.jar:4.1.36.Final]
at io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:163) [netty-common-4.1.36.Final.jar:4.1.36.Final]
at io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:405) [netty-common-4.1.36.Final.jar:4.1.36.Final]
at io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500) [netty-transport-4.1.36.Final.jar:4.1.36.Final]
at io.netty.util.concurrent.SingleThreadEventExecutor$5.run(SingleThreadEventExecutor.java:906) [netty-common-4.1.36.Final.jar:4.1.36.Final]
at io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74) [netty-common-4.1.36.Final.jar:4.1.36.Final]
at java.lang.Thread.run(Thread.java:748) [na:1.8.0_181]
Suppressed: reactor.core.publisher.FluxOnAssembly$OnAssemblyException:
Assembly trace from producer [reactor.core.publisher.MonoCreate] :
reactor.core.publisher.Mono.create(Mono.java:183)
reactor.netty.resources.PooledConnectionProvider.acquire(PooledConnectionProvider.java:130)
Error has been observed by the following operator(s):
|_ Mono.create ⇢ reactor.netty.resources.PooledConnectionProvider.acquire(PooledConnectionProvider.java:130)
|_ Mono.doOnSubscribe ⇢ reactor.netty.tcp.TcpClientDoOn.connect(TcpClientDoOn.java:58)
The 'inbound' and 'outbound' are having a dedicated method to handle their errors, but they works on top of a Connection instance that won't be created if you got the 'connection timeout'.
I tried:
The exception, that i receiving is wrapped in 'ErrorCallbackNotImplemented'. But I wasn't able to find any way to implement any 'ErrorCallback'
The log contains a warning message from 'io.netty.util.concurrent.DefaultPromise' . but I wasn't able to find a way how to make own Promise to handle it in a right way.
No any configurations i've found that may somehow intercept connection timeouts.
workaround. The blocked approach to create a connection ( .block() instead of .subscribe()) will allow me to catch any Connection creating exceptions within plain try-catch block, but i'll lose the benefits of reactive approach with such workaround.
Do somebody may suggest me at least something to help me to find a right way to handle a 'io.netty.channel.ConnectTimeoutException'?

Do not forget to implement your error callback
Usually reactor.core.Exceptions$ErrorCallbackNotImplemented happens when there is subscription over labmda based .subscribe method (same for Mono and Flux).
If you are going to look at the sources here and here, you will find the place where reactor.core.Exceptions$ErrorCallbackNotImplemented is thrown!
Action Points
In order to handle the original io.netty.channel.ConnectTimeoutException I would recommend looking at Handling Errors section of the original Project Reactor documentation

How to make the distinction between different exceptions in Netty's onChannelFailure callback

I am writing an HTTP client with Netty 4.1.12.Final and I have unit tests simulating the crash of the HTTP server in order to be able to handle it.
I noticed that, when it happens, the exceptionCaught callback method of my inbound handler is called with:
java.io.IOException: Une connexion existante a dû être fermée par l’hôte distant
at sun.nio.ch.SocketDispatcher.read0(Native Method)
at sun.nio.ch.SocketDispatcher.read(Unknown Source)
at sun.nio.ch.IOUtil.readIntoNativeBuffer(Unknown Source)
at sun.nio.ch.IOUtil.read(Unknown Source)
at sun.nio.ch.SocketChannelImpl.read(Unknown Source)
at io.netty.buffer.PooledUnsafeDirectByteBuf.setBytes(PooledUnsafeDirectByteBuf.java:288)
at io.netty.buffer.AbstractByteBuf.writeBytes(AbstractByteBuf.java:1100)
at io.netty.channel.socket.nio.NioSocketChannel.doReadBytes(NioSocketChannel.java:372)
at io.netty.channel.nio.AbstractNioByteChannel$NioByteUnsafe.read(AbstractNioByteChannel.java:123)
at io.netty.channel.nio.NioEventLoop.processSelectedKey(NioEventLoop.java:644)
at io.netty.channel.nio.NioEventLoop.processSelectedKeysOptimized(NioEventLoop.java:579)
at io.netty.channel.nio.NioEventLoop.processSelectedKeys(NioEventLoop.java:496)
at io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:458)
at io.netty.util.concurrent.SingleThreadEventExecutor$5.run(SingleThreadEventExecutor.java:858)
at io.netty.util.concurrent.DefaultThreadFactory$DefaultRunnableDecorator.run(DefaultThreadFactory.java:138)
Where the english equivalent of the exception message is quite probably:
java.io.IOException: An existing connection was forcibly closed by the remote host
Since this callback method is also called when an exception is thrown from my channelRead0 method of my inbound handler, I am asking a few questions:
Should I always consider an IOException "received" in the exceptionCaught callback as an indication that there is no point in continuing using the channel?
Since channelRead0 is declared to throw Exception, should I catch all IOException inside it in order to be sure that, when "receiving" an IOException in the exceptionCaught callback, it is related to the Channel?
Is there a way to know if an exception "received" in the exceptionCaught callback is related to I/O operations or to handlers operations?
Thank you for any hint!

1) If we are talking about a TCP connection then yes every IOException will result in having the connection closed automatically by netty as there is no way to recover.
2) I think I don't understand the question completely as each exception passed through the exceptionCaught(...) method is related the the channel which can be obtained by ctx.channel()
3) No there is no way in general. That said if its a TCP connection and its an IOException and its triggered by the actual transport we will close the connection.

spring boot redis operation throw broken pipe error

We use redis in spring boot project. After running a period of time, the redis operation MAY throw a broken pipe error, but sometimes it will succeed. Restarting the service will resolve this problem, but it's not a good idea.
I can't tell the reason why it happens. It seems that some redis connections in the pool are unusable, but aren't closed and evicted from the pool.
my questions are:
possible reason causing the broken pipe error?
if there isn't redis operation for a long period, will the idle connection in the pool become unusable?
will the connection be closed and evicted from pool when broken pipe error happen?
pom.xml:
<dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-data-redis</artifactId>
</dependency>
application.yml:
spring:
redis:
database: 0
host: ${REDIS_HOST:127.0.0.1}
password: ${REDIS_PASSWORD:password}
port: ${REDIS_PORT:6379}
timeout: ${REDIS_TIMEOUT:1000}
pool:
max-active: ${REDIS_MAX_ACTIVE:100}
max-wait: ${REDIS_MAX_WAIT:500}
max-idle: ${REDIS_MAX_IDLE:20}
min-idle: ${REDIS_MIN_IDLE:5}
error message:
org.springframework.data.redis.RedisConnectionFailureException: java.net.SocketException: Broken pipe (Write failed); nested exception is redis.clients.jedis.exceptions.JedisConnectionException: java.net.SocketException: Broken pipe (Write failed)
at org.springframework.data.redis.connection.jedis.JedisExceptionConverter.convert(JedisExceptionConverter.java:67) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.connection.jedis.JedisExceptionConverter.convert(JedisExceptionConverter.java:41) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.PassThroughExceptionTranslationStrategy.translate(PassThroughExceptionTranslationStrategy.java:37) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.FallbackExceptionTranslationStrategy.translate(FallbackExceptionTranslationStrategy.java:37) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.connection.jedis.JedisConnection.convertJedisAccessException(JedisConnection.java:212) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.connection.jedis.JedisConnection.hSet(JedisConnection.java:2810) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.core.DefaultHashOperations$9.doInRedis(DefaultHashOperations.java:173) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.core.RedisTemplate.execute(RedisTemplate.java:204) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.core.RedisTemplate.execute(RedisTemplate.java:166) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.core.AbstractOperations.execute(AbstractOperations.java:88) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
at org.springframework.data.redis.core.DefaultHashOperations.put(DefaultHashOperations.java:170) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
Caused by: redis.clients.jedis.exceptions.JedisConnectionException: java.net.SocketException: Broken pipe (Write failed)
at redis.clients.jedis.Connection.flush(Connection.java:291) ~[jedis-2.8.2.jar!/:na]
at redis.clients.jedis.Connection.getIntegerReply(Connection.java:220) ~[jedis-2.8.2.jar!/:na]
at redis.clients.jedis.BinaryJedis.hset(BinaryJedis.java:749) ~[jedis-2.8.2.jar!/:na]
at org.springframework.data.redis.connection.jedis.JedisConnection.hSet(JedisConnection.java:2808) ~[spring-data-redis-1.7.6.RELEASE.jar!/:na]
... 115 common frames omitted
Caused by: java.net.SocketException: Broken pipe (Write failed)
at java.net.SocketOutputStream.socketWrite0(Native Method) ~[na:1.8.0_111]
at java.net.SocketOutputStream.socketWrite(SocketOutputStream.java:109) ~[na:1.8.0_111]
at java.net.SocketOutputStream.write(SocketOutputStream.java:153) ~[na:1.8.0_111]
at redis.clients.util.RedisOutputStream.flushBuffer(RedisOutputStream.java:52) ~[jedis-2.8.2.jar!/:na]
at redis.clients.util.RedisOutputStream.flush(RedisOutputStream.java:216) ~[jedis-2.8.2.jar!/:na]
at redis.clients.jedis.Connection.flush(Connection.java:288) ~[jedis-2.8.2.jar!/:na]
... 118 common frames omitted

Answer my question:
why the broken pipe error happen?
TransactionSynchronizationManager will hold the RedisConnection in thread, and wont close it or return it to pool, see RedisTemplate.java and RedisConnectionUtils.java. After restarting the redis server, operation on the holded RedisConnection in thread will throw broken pipe error.
how to resolve it?
Adding try/catch for all redis operation, if the error happens , unbinding it from thread, and could get a new connection from pool and execute redis operation again.
private static final ExceptionTranslationStrategy EXCEPTION_TRANSLATION =
new FallbackExceptionTranslationStrategy(JedisConverters.exceptionConverter());
public Object req(RedisRequest req) {
try {
return req.request();
} catch (Exception ex) {
if (ex instanceof NullPointerException) {
throw ex;
}
DataAccessException exception = EXCEPTION_TRANSLATION.translate(ex);
if (exception instanceof RedisConnectionFailureException) {
RedisConnectionUtils.unbindConnection(factory);
/** retry again */
return req.request();
} else {
throw ex;
}
}
}

This can happen for n number of reasons, one of them could be when you use a long-living connection (e. g. connect to Redis on application start and then use the connection over and over).
Some of the things to do are:
Reconnect if the connection is broken (needs some try/catch magic to prevent that errors are propagated to your application logic)
or a better was is to use
TestOnBorrow - Sends a PING request when you ask for the resource.
TestOnReturn - Sends a PING when you return a resource to the pool.
TestWhileIdle - Sends periodic PINGS from idle resources in the pool.
Connect at the moment you need the connection and disconnect afterwards
Regarding
if there isn't redis operation for a long period, will the idle connection in the pool become unusable?
maxidle means at any given time the system allows 'maxIdle' many number of connections to be idle the rest will be constantly checked, closed and returned to the pool.
I dont know a reason when an idle connection be unusable. Anyways this can be got rid of by using ways mentioned above.