Claus Ibsen created CAMEL-26555:
-----------------------------------
Summary: camel-core - DefaultShutdownStrategy: a shutdown task
cancelled on timeout keeps stopping the deferred consumers, on an interrupted
thread, at the same time as the forced shutdown
Key: CAMEL-26555
URL: https://issues.apache.org/jira/browse/CAMEL-26555
Project: Camel
Issue Type: Bug
Components: camel-core
Reporter: Claus Ibsen
Assignee: Claus Ibsen
Fix For: 4.24.0
h3. Problem
When a graceful shutdown times out, {{DefaultShutdownStrategy.doShutdown}}
cancels its {{ShutdownTask}} with {{future.cancel(true)}} and then, on the
calling thread, forces the routes to shutdown ({{shutdownRoutesNow}}), or
aborts, or ignores the remainder routes.
The cancelled task does not stop: its wait loop ends (on the timeout flag, or
on the interrupt with "Interrupted while waiting during graceful shutdown, will
force shutdown now."), and it goes on to prepare and stop the deferred
consumers. So the same consumers are stopped by two threads at the same time,
and the task does it on a thread whose interrupt flag it has just restored. A
consumer whose stop waits on something interruptible then fails to stop, for
example Jetty's {{ManagedSelector.doStop}}:
{noformat}
WARN DefaultShutdownStrategy - Error occurred while shutting down route: ...
This exception will be ignored.
java.lang.RuntimeException: java.lang.InterruptedException
at
org.eclipse.jetty.util.component.ContainerLifeCycle.remove(ContainerLifeCycle.java:730)
...
at
org.apache.camel.impl.engine.DefaultShutdownStrategy.shutdownNow(DefaultShutdownStrategy.java:434)
at
org.apache.camel.impl.engine.DefaultShutdownStrategy$ShutdownTask.run(DefaultShutdownStrategy.java:743)
{noformat}
Which thread reaches a consumer first is a race. In CAMEL-26554 this left a
camel-jetty port that no longer accepted connections.
h3. Fix
Once the shutdown timed out (or the task was interrupted), the task stops: the
calling thread owns what happens next (force, abort or ignore), and the
consumers of a stopped route are stopped by the route stop afterwards anyway.
Found while looking at CAMEL-26554.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
