Sunil created CAMEL-12718:
-----------------------------
Summary: Camel retry optimization on large volume
Key: CAMEL-12718
URL: https://issues.apache.org/jira/browse/CAMEL-12718
Project: Camel
Issue Type: Improvement
Components: camel-core
Reporter: Sunil
We are using camel retry in our project. It works fine when number of
transactions/period is low.
Our transaction are a ftp process - occassionally we encounter connections
issues(servers are down etc) - so we have configured for a specific number of
retry/delay - after which the flows goes to the error processor.
When there is a large number of transactions (say > 25) that fails and comes to
retry loop all the same time - then the retry period of each transactions goes
up...it takes more time for each of transactions to reach to the error
processor.
We have configured a redeliveryprocessor to gather time statistics - as of now
what we understand is that -
a) a transaction goes further and further back in the queue for retry each time
other failed transactions are added to the queue.
b) the trigger for retry happens every second . We have added a
scheduledexecutor service (10 threads to the deadletterChannel builder) -
however the number of threads does not see increase the frequency of retry -
since there are lots of transactions waiting for retry - expectation is there
should be a retry happening as soon as transactions are available for retry and
not necessarily every 1 sec.
Below is the blueprint snippet for this
<bean class="org.apache.camel.builder.DeadLetterChannelBuilder"
id="myDeadLetterErrorHandler">
<property name="deadLetterUri" value="direct:deadletter"/>
<property name="redeliveryPolicy" ref="myRedeliveryPolicyConfig"/>
<property name="executorServiceRef" value="redeliveryThreadPool"/>
<property name="onRedelivery" ref="redelProcessor"/> <!-- we have used
this for getting statistics of the retry --->
</bean>
<bean class="org.apache.camel.processor.RedeliveryPolicy"
id="myRedeliveryPolicyConfig">
<property name="maximumRedeliveries" value="${maximum.Redeliveries}"/>
<property name="redeliveryDelay" value="${redelivery.Delay}"/>
<property name="asyncDelayedRedelivery" value="true"/>
<property name="maximumRedeliveryDelay"
value="${maximum.RedeliveryDelay}"/>
<property name="backOffMultiplier" value="0"/>
</bean>
<bean class="java.util.concurrent.ScheduledThreadPoolExecutor"
id="redeliveryThreadPool">
<argument type="int" value="${int}"/>
<property name="maximumPoolSize" value="${maximum.PoolSize}"/>
</bean>
Need inputs/advise on how to limit retry delay when large volume of failed
transactions within a short period
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
