[jira] [Assigned] (CAMEL-23291) EventHubs and Google PubSub consumers lack ShutdownAware support, causing potential message loss during graceful shutdown

["Claus Ibsen (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/CAMEL-23291?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Claus Ibsen reassigned CAMEL-23291:
-----------------------------------

    Assignee: Claus Ibsen

> EventHubs and Google PubSub consumers lack ShutdownAware support, causing potential message loss during graceful shutdown
> -------------------------------------------------------------------------------------------------------------------------
>
>                 Key: CAMEL-23291
>                 URL: https://issues.apache.org/jira/browse/CAMEL-23291
>             Project: Camel
>          Issue Type: Bug
>          Components: camel-azure-eventhubs, camel-google-pubsub
>    Affects Versions: 4.18.1
>            Reporter: Federico Mariani
>            Assignee: Claus Ibsen
>            Priority: Minor
>             Fix For: 4.18.3, 4.21.0
>
>
> _EventHubsConsumer_ and _GooglePubsubConsumer_ do not implement ShutdownAware. Both consumers close/stop their underlying client immediately in _doStop()_, without waiting for in-flight exchanges to complete. This is the same root cause that was fixed for ServiceBusConsumer in [CAMEL-23260|https://issues.apache.org/jira/browse/CAMEL-23260].
> When a SIGTERM is received while messages are being processed:
> - EventHubsConsumer.doStop() calls processorClient.stop() and sets it to null immediately
> - GooglePubsubConsumer.doStop() calls stopAsync() on all subscribers and shuts down the executor
> In both cases, in-flight exchanges may fail to ACK/NACK their messages because the underlying client connection is torn down before the exchange completion callbacks run. This can lead to message redelivery and duplicate processing.
> Both consumers should implement ShutdownAware following the pattern established in CAMEL-23260:
> - deferShutdown() should stop the message pump while keeping connections alive
> - getPendingExchangesSize() should track in-flight exchanges
> - Client resources should only be fully closed in doShutdown(), after all exchanges have drained



--
This message was sent by Atlassian Jira
(v8.20.10#820010)