/**
* Finds the routing source node for peer recovery, return null if its not found. Note, this
* method expects the shard routing to *require* peer recovery, use {@link
* #isPeerRecovery(org.elasticsearch.cluster.routing.ShardRouting)} to check if its needed or not.
*/
private DiscoveryNode findSourceNodeForPeerRecovery(
RoutingTable routingTable, DiscoveryNodes nodes, ShardRouting shardRouting) {
DiscoveryNode sourceNode = null;
if (!shardRouting.primary()) {
IndexShardRoutingTable shardRoutingTable =
routingTable.index(shardRouting.index()).shard(shardRouting.id());
for (ShardRouting entry : shardRoutingTable) {
if (entry.primary() && entry.active()) {
// only recover from started primary, if we can't find one, we will do it next round
sourceNode = nodes.get(entry.currentNodeId());
if (sourceNode == null) {
logger.trace(
"can't find replica source node because primary shard {} is assigned to an unknown node.",
entry);
return null;
}
break;
}
}
if (sourceNode == null) {
logger.trace(
"can't find replica source node for {} because a primary shard can not be found.",
shardRouting.shardId());
}
} else if (shardRouting.relocatingNodeId() != null) {
sourceNode = nodes.get(shardRouting.relocatingNodeId());
if (sourceNode == null) {
logger.trace(
"can't find relocation source node for shard {} because it is assigned to an unknown node [{}].",
shardRouting.shardId(),
shardRouting.relocatingNodeId());
}
} else {
throw new IllegalStateException(
"trying to find source node for peer recovery when routing state means no peer recovery: "
+ shardRouting);
}
return sourceNode;
}
private void cleanFailedShards(final ClusterChangedEvent event) {
RoutingTable routingTable = event.state().routingTable();
RoutingNodes.RoutingNodeIterator routingNode =
event.state().readOnlyRoutingNodes().routingNodeIter(event.state().nodes().localNodeId());
if (routingNode == null) {
failedShards.clear();
return;
}
DiscoveryNodes nodes = event.state().nodes();
long now = System.currentTimeMillis();
String localNodeId = nodes.localNodeId();
Iterator<Map.Entry<ShardId, FailedShard>> iterator = failedShards.entrySet().iterator();
shards:
while (iterator.hasNext()) {
Map.Entry<ShardId, FailedShard> entry = iterator.next();
FailedShard failedShard = entry.getValue();
IndexRoutingTable indexRoutingTable = routingTable.index(entry.getKey().getIndex());
if (indexRoutingTable != null) {
IndexShardRoutingTable shardRoutingTable = indexRoutingTable.shard(entry.getKey().id());
if (shardRoutingTable != null) {
for (ShardRouting shardRouting : shardRoutingTable.assignedShards()) {
if (localNodeId.equals(shardRouting.currentNodeId())) {
// we have a timeout here just to make sure we don't have dangled failed shards for
// some reason
// its just another safely layer
if (shardRouting.version() == failedShard.version
&& ((now - failedShard.timestamp) < TimeValue.timeValueMinutes(60).millis())) {
// It's the same failed shard - keep it if it hasn't timed out
continue shards;
} else {
// Different version or expired, remove it
break;
}
}
}
}
}
iterator.remove();
}
}
