You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@activemq.apache.org by rdvries <ro...@triodos.com.INVALID> on 2019/12/13 11:27:22 UTC

Getting "Succesfully connected" all the time when failover is setup

Hi all! 

We've recently been connecting to another party's broker, with a failover
setup. 
We soon discovered we're getting a lot of these:
11:51:36,115 INFO 
[org.apache.activemq.transport.failover.FailoverTransport] (ActiveMQ Task-1)
Successfully connected to tcp://<host>:61616
Sometimes even a couple per second. 

It seems to be related to the failover. 

I can reproduce this by just running a local broker with a failover URL like
so (no messaging involved)
<resource-adapter id="activemq-ra">
                    <archive>
                        activemq-rar.rar
                    </archive>
                    <config-property name="ServerUrl">
                       
failover:(tcp://localhost:61616,tcp://localhost:61617)?randomize=false
                    </config-property>
                    <connection-definitions>
                        <connection-definition
class-name="org.apache.activemq.ra.ActiveMQManagedConnectionFactory"
jndi-name="activemq/ConnectionFactory" pool-name="activemq">
                            <recovery no-recovery="true"/>
                        </connection-definition>
                    </connection-definitions>
                </resource-adapter>

Using version 5.15.9, with Jboss 7.2.

Locally, the message / (re?)connection seems to be triggered by the periodic
recovery period of the transaction manager. 
So setting the tx manager recovery like following, I can locally exactly see
the 30/10 pattern in the loggings:
-DRecoveryEnvironmentBean.periodicRecoveryPeriod=30
-DRecoveryEnvironmentBean.recoveryBackoffPeriod=10

We've been tinkering and staring at trace/debug logging for too long now...
time to ask help :)







--
Sent from: http://activemq.2283324.n4.nabble.com/ActiveMQ-User-f2341805.html

Re: Getting "Succesfully connected" all the time when failover is setup

Posted by Robbert de Vries <Ro...@triodos.com.INVALID>.
Doesn't anyone have any idea? 
Didn't I provide enough information? Is the problem too hard to help out with? 

It would really help to get some thoughts on disabling the XA recovery. Whether this is a problem, and whether we're doing it in the right way? 
See my last post.

On 16/12/2019, 16:00, "Robbert de Vries" <Ro...@triodos.com.INVALID> wrote:

    Also, I discovered this warning in the startup of JBoss. 
    
    15:47:36,686 WARN  [org.jboss.as.connector.deployers.RaXmlDeployer] (MSC service thread 1-7) IJ020016: Missing <recovery> element. XA recovery disabled for: java:/activemq/ConnectionFactory
    I even get that when I explicitly add <recovery no-recovery="true"/> to the connection-definition. 
    
    So I guess XA recovery is disabled anyway, for this connection factory...  
    
    Is it save to just disable periodic recovery in this setup? Like so 
    -DRecoveryEnvironmentBean.periodicRecoveryPeriod=0 
    -DRecoveryEnvironmentBean.recoveryBackoffPeriod=0
    
    We don't want tx management between datasource and jms messages anyway. Any millisecond in the app will count, and XA takes performance penalty.

Re: Getting "Succesfully connected" all the time when failover is setup

Posted by Robbert de Vries <Ro...@triodos.com.INVALID>.
Also, I discovered this warning in the startup of JBoss. 

15:47:36,686 WARN  [org.jboss.as.connector.deployers.RaXmlDeployer] (MSC service thread 1-7) IJ020016: Missing <recovery> element. XA recovery disabled for: java:/activemq/ConnectionFactory
I even get that when I explicitly add <recovery no-recovery="true"/> to the connection-definition. 

So I guess XA recovery is disabled anyway, for this connection factory...  

Is it save to just disable periodic recovery in this setup? Like so 
-DRecoveryEnvironmentBean.periodicRecoveryPeriod=0 
-DRecoveryEnvironmentBean.recoveryBackoffPeriod=0

We don't want tx management between datasource and jms messages anyway. Any millisecond in the app will count, and XA takes performance penalty.

Re: Getting "Succesfully connected" all the time when failover is setup

Posted by Robbert de Vries <Ro...@triodos.com.INVALID>.
Maybe some more information; 
Here's a recent local logging

14:37:17,865 DEBUG [org.apache.activemq.transport.failover.FailoverTransport] (Periodic Recovery) Stopped tcp://localhost:61616
14:37:17,866 DEBUG [org.apache.activemq.transport.failover.FailoverTransport] (Periodic Recovery) Reconnect was triggered but transport is not started yet. Wait for start to connect the transport.
14:37:17,866 DEBUG [org.apache.activemq.transport.failover.FailoverTransport] (Periodic Recovery) Started unconnected
14:37:17,866 DEBUG [org.apache.activemq.transport.failover.FailoverTransport] (Periodic Recovery) Waking up reconnect task
14:37:17,867 DEBUG [org.apache.activemq.transport.failover.FailoverTransport] (ActiveMQ Task-1) urlList connectionList:[tcp://localhost:61616], from: [tcp://localhost:61616]
14:37:17,867 DEBUG [org.apache.activemq.transport.failover.FailoverTransport] (ActiveMQ Task-1) Attempting 0th connect to: tcp://localhost:61616
14:37:17,867 DEBUG [org.apache.activemq.transport.failover.FailoverTransport] (ActiveMQ Task-1) Connection established
14:37:17,868 INFO  [org.apache.activemq.transport.failover.FailoverTransport] (ActiveMQ Task-1) Successfully connected to tcp://localhost:61616

On 13/12/2019, 18:13, "rdvries" <ro...@triodos.com.INVALID> wrote:

    Hi all! 
    
    We've recently been connecting to another party's broker, with a failover
    setup. 
    We soon discovered we're getting a lot of these:
    11:51:36,115 INFO 
    [org.apache.activemq.transport.failover.FailoverTransport] (ActiveMQ Task-1)
    Successfully connected to tcp://<host>:61616
    Sometimes even a couple per second. 
    
    It seems to be related to the failover. 
    
    I can reproduce this by just running a local broker with a failover URL like
    so (no messaging involved)
    <resource-adapter id="activemq-ra">
                        <archive>
                            activemq-rar.rar
                        </archive>
                        <config-property name="ServerUrl">
                           
    failover:(tcp://localhost:61616,tcp://localhost:61617)?randomize=false
                        </config-property>
                        <connection-definitions>
                            <connection-definition
    class-name="org.apache.activemq.ra.ActiveMQManagedConnectionFactory"
    jndi-name="activemq/ConnectionFactory" pool-name="activemq">
                                <recovery no-recovery="true"/>
                            </connection-definition>
                        </connection-definitions>
                    </resource-adapter>
    
    Using version 5.15.9, with Jboss 7.2.
    
    Locally, the message / (re?)connection seems to be triggered by the periodic
    recovery period of the transaction manager. 
    So setting the tx manager recovery like following, I can locally exactly see
    the 30/10 pattern in the loggings:
    -DRecoveryEnvironmentBean.periodicRecoveryPeriod=30
    -DRecoveryEnvironmentBean.recoveryBackoffPeriod=10
    
    We've been tinkering and staring at trace/debug logging for too long now...
    time to ask help :)
    
    
    
    
    
    
    
    --
    Sent from: http://activemq.2283324.n4.nabble.com/ActiveMQ-User-f2341805.html