private void innerStart(final DiscoveryNode masterNode) {
this.masterNode = masterNode;
this.retryCount = 0;
this.notifiedMasterFailure.set(false);
// try and connect to make sure we are connected
try {
transportService.connectToNode(masterNode);
} catch (final Exception e) {
// notify master failure (which stops also) and bail..
notifyMasterFailure(masterNode, "failed to perform initial connect [" + e.getMessage() + "]");
return;
}
if (masterPinger != null) {
masterPinger.stop();
}
this.masterPinger = new MasterPinger();
// start the ping process
threadPool.schedule(pingInterval, ThreadPool.Names.SAME, masterPinger);
}
private void innerJoinCluster() {
boolean retry = true;
while (retry) {
if (lifecycle.stoppedOrClosed()) {
return;
}
retry = false;
DiscoveryNode masterNode = findMaster();
if (masterNode == null) {
logger.trace("no masterNode returned");
retry = true;
continue;
}
if (localNode.equals(masterNode)) {
this.master = true;
nodesFD.start(); // start the nodes FD
clusterService.submitStateUpdateTask(
"zen-disco-join (elected_as_master)",
Priority.URGENT,
new ProcessedClusterStateUpdateTask() {
@Override
public ClusterState execute(ClusterState currentState) {
DiscoveryNodes.Builder builder =
new DiscoveryNodes.Builder()
.localNodeId(localNode.id())
.masterNodeId(localNode.id())
// put our local node
.put(localNode);
// update the fact that we are the master...
latestDiscoNodes = builder.build();
ClusterBlocks clusterBlocks =
ClusterBlocks.builder()
.blocks(currentState.blocks())
.removeGlobalBlock(NO_MASTER_BLOCK)
.build();
return ClusterState.builder(currentState)
.nodes(latestDiscoNodes)
.blocks(clusterBlocks)
.build();
}
@Override
public void onFailure(String source, Throwable t) {
logger.error("unexpected failure during [{}]", t, source);
}
@Override
public void clusterStateProcessed(
String source, ClusterState oldState, ClusterState newState) {
sendInitialStateEventIfNeeded();
}
});
} else {
this.master = false;
try {
// first, make sure we can connect to the master
transportService.connectToNode(masterNode);
} catch (Exception e) {
logger.warn("failed to connect to master [{}], retrying...", e, masterNode);
retry = true;
continue;
}
// send join request
try {
membership.sendJoinRequestBlocking(masterNode, localNode, pingTimeout);
} catch (Exception e) {
if (e instanceof ElasticsearchException) {
logger.info(
"failed to send join request to master [{}], reason [{}]",
masterNode,
((ElasticsearchException) e).getDetailedMessage());
} else {
logger.info(
"failed to send join request to master [{}], reason [{}]",
masterNode,
e.getMessage());
}
if (logger.isTraceEnabled()) {
logger.trace("detailed failed reason", e);
}
// failed to send the join request, retry
retry = true;
continue;
}
masterFD.start(masterNode, "initial_join");
// no need to submit the received cluster state, we will get it from the master when it
// publishes
// the fact that we joined
}
}
}