I have observed that even if i stop a node or a participant by pressing ctr + C, i.e. terminate it...it still shows in the Admin console or 2 minutes or so. Why isnt it immediately derigestered. Is there a way to do so ?
The Global Leader in DDS. Your Systems working as one.
Hey,
Since you are not gracefully terminating the application, its participant(s) have no way of alerting other participants that they're "dead".
How ever, RTI is able to identify that a participant died after a period of time that depends on a few things:
1. How often are participant discovery messages sent (default is 30 seconds)
2. How long are remote participants considered "alive" after their most recent participant discovery message was received (I don't remember the default)
3. How often the thread responsible for cleaning the internal data structures RTI maintains on each node runs
So it may take about 2 minutes for RTI to realize a participant has died.
BUT THAT'S TOO LONG, you may think.
This is why RTI enables, on a per topic basis, using a liveliness mechanism which will alert readers when a writer is presumed not "alive" using harsher standards (basically, liveliness messages are sent at rate X and the other side expects to get them at a rate that is greater than Y).
I'm uncertain if enabling this would affect admin console as you wish it to but this is how you could detect application faults more quickly (another, more applicative approach, would be to periodically send applicative status messages and then, if required, maintain a "liveliness" map which verifies remote applications are alive).
I hope this helps,
Roy.
Hi,
Just for completeness, you could apply specific QoS profiles to Admin Console. So if QoS changes are needed as Roy mentioned, you could apply them in the tool too.
All the best,
Sara