[
https://issues.apache.org/jira/browse/UIMA-1292?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jerry Cwiklik closed UIMA-1292.
-------------------------------
Resolution: Fixed
The listener failed to recover from a lost broker because of the
maxInactivityDuration=0 parameter in the broker URL. This effectively turns off
ability to detect socket closure and causing a service to sit on a stale
connection. No messages flow into the service even if a new broker becomes
available. Modified the listener to accept broker URL as provided by a user.
Removed check for existence of maxInactivityDuration=0. If there are problems
with socket closing due to inactivity use keepAlive=true parameter in the
broker URL instead of maxInactivityDuration=0. Example:
tcp://host:61616?keepAlive=true. This turns on low level socket keep alive
protocol which sends small packets and awaits ACKs. The frequency of these
packets can be changed on linux by editing this file:
/proc/sys/net/ipv4/tcp_keepalive_intvl. The default frequency on linux is 75
seconds
> Uima AS Jms Listener Should Support Retry When Establishing Connection To a
> Broker
> ----------------------------------------------------------------------------------
>
> Key: UIMA-1292
> URL: https://issues.apache.org/jira/browse/UIMA-1292
> Project: UIMA
> Issue Type: Bug
> Components: Async Scaleout
> Reporter: Jerry Cwiklik
> Assignee: Jerry Cwiklik
>
> UIMA AS service should support retry when a listener on the service input
> queue fails due to a broker not being available. The listener should log a
> message stating the fact that the broker is not available and it should enter
> a loop attempting to reconnect. When a broker becomes available the listener
> should reconnect, log a message that the connection was made, and proceed to
> complete initialization.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
