tjwatson commented on code in PR #486:
URL: https://github.com/apache/felix-dev/pull/486#discussion_r2981401649
##########
scr/src/main/java/org/apache/felix/scr/impl/ComponentRegistry.java:
##########
@@ -718,42 +716,44 @@ public Dictionary<String, Object>
getServiceRegistrationProperties()
public void setRegistration(final
ServiceRegistration<ServiceComponentRuntime> reg)
{
- this.registration = reg;
+ long delay = m_configuration.serviceChangecountTimeout();
+ m_componentActor.scheduleWithFixedDelay(new
UpdateChangeCountProperty(reg), delay, delay, TimeUnit.MILLISECONDS);
Review Comment:
I was out all last week, just catching up ...
In full disclosure we only saw the spike in CPU usage with the Semeru/OpenJ9
JVM. The Temerin JVM did not show similar results. In more details, the issue
is in the ScheduledExecutor implementation vs the Timer implementation for
dispatching tasks.
Regardless of this detail, I still believe the current implementation is not
the best way. This is no fault of #419 since it did the same thing as the
Timer approach did. I see no reason to keep submitting a task over and over
for this though. I'll take a look at your #487 PR to see if it is the same as
I am thinking. But from what you describe it still sounds like it is creating
large numbers of tasks to submit to the scheduled executor, which I want to
avoid.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
