[jira] [Created] (SLING-9030) SimpleDistributionAgentQueueProcessor does not distinguish between recoverable and non-recoverable exceptions

["Mohit Arora (Jira)" <ji...@apache.org>]

Mohit Arora created SLING-9030:
----------------------------------

             Summary: SimpleDistributionAgentQueueProcessor does not distinguish between recoverable and non-recoverable exceptions
                 Key: SLING-9030
                 URL: https://issues.apache.org/jira/browse/SLING-9030
             Project: Sling
          Issue Type: Bug
          Components: Content Distribution
            Reporter: Mohit Arora
             Fix For: Content Distribution Core 0.4.2


[SimpleDistributionAgentQueueProcessor.java|https://github.com/apache/sling-org-apache-sling-distribution-core/blob/master/src/main/java/org/apache/sling/distribution/agent/impl/SimpleDistributionAgentQueueProcessor.java#L83] is responsible for processing a queueItem which is then passed on to [RemoteDistributionPackageImporter#importPackage()|https://github.com/apache/sling-org-apache-sling-distribution-core/blob/master/src/main/java/org/apache/sling/distribution/packaging/impl/importer/RemoteDistributionPackageImporter.java#L59] which in turn selects a valid transporter and send the POST request through [SimpleHttpDistributionTransport#deliverPackage|https://github.com/apache/sling-org-apache-sling-distribution-core/blob/master/src/main/java/org/apache/sling/distribution/transport/impl/SimpleHttpDistributionTransport.java#L108]. There can be 2 type of exceptions thrown by this deliverPackage() function. One is a [RecoverableDistributionException|https://github.com/apache/sling-org-apache-sling-distribution-core/blob/master/src/main/java/org/apache/sling/distribution/common/RecoverableDistributionException.java] which is a type of DistributionException and another is DistributionException. As the name suggests, a RecoverableDistributionException is where the transport is tried again. But it seems there is currently no cap on the number of retries.

For example, if the endpoint is not accessible at the moment, the error logs of the caller application will be flooded with constant retries of the DistributionPackages in queue, until the endpoint comes up and the distribution is successful. The reason being, [the verbose logging done here|https://github.com/apache/sling-org-apache-sling-distribution-core/blob/master/src/main/java/org/apache/sling/distribution/agent/impl/SimpleDistributionAgentQueueProcessor.java#L100] does not distinguish between a RecoverableDistributionException and a normal DistributionException. This would lead to sharp increase in disk size of the caller application. Perhaps the logging can be less verbose and can be logged at *WARN* level for RecoverableDistributionException.

cc - [~ashishc], [~marett]



--
This message was sent by Atlassian Jira
(v8.3.4#803005)