Apache Ignite cluster node events not Triggered

classic Classic list List threaded Threaded
4 messages Options
BasavarajB BasavarajB
Reply | Threaded
Open this post in threaded view
|

Apache Ignite cluster node events not Triggered

CONTENTS DELETED
The author has deleted this message.
ilya.kasnacheev ilya.kasnacheev
Reply | Threaded
Open this post in threaded view
|

Re: Apache Ignite cluster node events not Triggered

Hello!

Maybe it's because retrying connections to server nodes takes prolonged period of time (due to network null routing) and node only really finishes trying to reconnect where it has reconnected?

Client nodes basically have soft reconnect and hard reconnect. Soft reconnect is when they just switch to another server node for discovery, no CLIENT_NODE_DISCONNECTED, no IgniteClientNodeDisconnectedException, no operations are interrupted.

Regards,
--
Ilya Kasnacheev


чт, 29 авг. 2019 г. в 11:21, BasavarajB <[hidden email]>:
Hi,



working with ignite with kubernetes ipfinder

org.apache.ignite.spi.discovery.tcp.ipfinder.kubernetes.TcpDiscoveryKubernetesIpFinder

in local machine i have started 2 local server instance with IP finder..

org.apache.ignite.spi.discovery.tcp.ipfinder.multicast.TcpDiscoveryMulticastIpFinder

and i have registered local listener for events

ignite.events().localListen(evt -> {},EventType.EVTS_DISCOVERY);

in local machine if local server stopped, ignite EVT_NODE_LEFT will trigger.
if all server node left EVT_CLIENT_NODE_DISCONNECTED event will trigger.

but in testing server with kubernetes IP finder if i stopped all pods in
which ignite is running there no ignite event triggering
CLIENT_NODE_DISCONNECTED.

but if restart ignite pods then the previous CLIENT_NODE_DISCONNECTED event
triggered after that CLIENT_NODE_RECONNECTING,NODE_JOIN event is triggered.

need some suggestion why event CLIENT_NODE_DISCONNECTED is not trigger once
pods stopped and why CLIENT_NODE_DISCONNECTED event is triggered after pods
are up,is there any queue if pods get down.




--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/
BasavarajB BasavarajB
Reply | Threaded
Open this post in threaded view
|

Re: Apache Ignite cluster node events not Triggered

This post was updated on .
CONTENTS DELETED
The author has deleted this message.
ilya.kasnacheev ilya.kasnacheev
Reply | Threaded
Open this post in threaded view
|

Re: Apache Ignite cluster node events not Triggered

Hello!

Client node will not disconnect immediately nor it will terminate. It will try to reconnect instead.

If you want client node to disconnect and fire CLIENT_NODE_DISCONNECTED immediately, you can try setting TcpDiscoverySpi.clientReconnectDisabled to true.

Regards,
--
Ilya Kasnacheev


вт, 3 сент. 2019 г. в 10:08, BasavarajB <[hidden email]>:
Hello,

Thank for reply.

in my case terminated(kiiled) all ignite server pods manually in that case
all Nodes will left the cluster and client node should terminate and it will
raise the event CLIENT_NODE_DISCONNECTED.

if any one server get disconnected then event NODE_LEFT event will raise and
client will connect to other server mode.

my concern is--

if all ignite server pods left(all server left from cluster) then client
should disconnect cluster and it should trigger the event
CLIENT_NODE_DISCONNECTED.

but there is no events triggering it is always trying to reconnect,but why
CLIENT_NODE_DISCONNECTED event triggering once all server pods are started.

In local machine this case is working with all events notification.



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/