These are chat archives for akkadotnet/akka.net

26th
Mar 2015
Stefan Sedich
@stefansedich
Mar 26 2015 02:01
Say I am listening to a terminated child to be able to fail a parent process, how does one get information on why that child died? should a child be sending a message to the parent upon termination? can I hook into that by any chance?
Stefan Sedich
@stefansedich
Mar 26 2015 03:26
I want to know why my child died to be able to update a record that my parent keeps track of, say it is an AWS exception I want to update with "file not found" for example.
Roger Johansson
@rogeralsing
Mar 26 2015 09:35
@stefansedich protected override void PreRestart(Exception reason, object message) should work?
or you mean before termination?
David Smith
@smalldave
Mar 26 2015 15:54
@Aaronontheweb looking at restartnodespec. it downs one system and then brings up another but on the same port. this isn't working. the new system seems to work fine but the cluster leader can't send InitJoinAck to the new system so it never joins. I've tracked it down to the endpointwriter on the cluster leader for the new systems (and presumably the old systems) endpoint. it never changes message handler from Initialized and just keeps buffering the message. Any guess what is happening here? Is the endpoint writer likely to be the endpoint writer for the first system that has been shutdown. Should this matter?
David Smith
@smalldave
Mar 26 2015 18:01
@Aaronontheweb ignore all that
Is there an equivalent of the JVM system.registerOnTermination? Rather nasty bug. When the actor system shuts down it doesn't leave the cluster!
Stefan Sedich
@stefansedich
Mar 26 2015 22:12
yeah @rogeralsing before termination.
argh what the hell have they done to the new mongodb c# driver, put some load on my api and CPU get pinned, all from the linq parsing.