Vladislav Pyatkov created IGNITE-26396:
------------------------------------------
Summary: A node restart might not send a message to the initiator.
Key: IGNITE-26396
URL: https://issues.apache.org/jira/browse/IGNITE-26396
Project: Ignite
Issue Type: Improvement
Reporter: Vladislav Pyatkov
h3. Motivation
After a ticket about acknowledgment batching was closed, we do not guarantee a
message will be delivered when the service future is completed. So the
restarting process can start, but the other side does not receive a successful
response yet.
{code}
private void handleResetClusterMessage(ResetClusterMessage message,
ClusterNode sender, long correlationId) {
restartExecutor.execute(() -> {
storage.saveResetClusterMessage(message);
messagingService.respond(sender, successResponseMessage(),
correlationId)
.thenRunAsync(() -> {
if (!thisNodeName.equals(sender.name())) {
restarter.initiateRestart();
}
}, restartExecutor)
.whenComplete((res, ex) -> {
if (ex != null) {
LOG.error("Error when handling a ResetClusterMessage",
ex);
}
});
});
}
{code}
h3. Definition of done.
The initiator side has to answer that the successful message is received
explicitly.
And the node should wait for the message before the restart process starts.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
