Inline ReplicationPhase#<init>

Relates #16725
This commit is contained in:
Jason Tedor 2016-02-14 09:07:02 -05:00
parent 7a7f6055dc
commit 7101ecea94
1 changed files with 49 additions and 46 deletions

View File

@ -83,6 +83,7 @@ import java.util.Map;
import java.util.concurrent.ConcurrentMap;
import java.util.concurrent.atomic.AtomicBoolean;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.function.Consumer;
import java.util.function.Supplier;
/**
@ -874,23 +875,48 @@ public abstract class TransportReplicationAction<Request extends ReplicationRequ
// to the recovery target. If we use an old cluster state, we may miss a relocation that has started since then.
// If the index gets deleted after primary operation, we skip replication
final ClusterState state = clusterService.state();
final IndexRoutingTable index = state.getRoutingTable().index(shardId.getIndex());
final IndexShardRoutingTable shardRoutingTable = (index != null) ? index.shard(shardId.id()) : null;
final IndexShardRoutingTable shardRoutingTable = state.getRoutingTable().shardRoutingTableOrNull(shardId);
final IndexMetaData indexMetaData = state.getMetaData().index(shardId.getIndex());
this.shards = (shardRoutingTable != null) ? shardRoutingTable.shards() : Collections.emptyList();
this.executeOnReplica = (indexMetaData == null) || shouldExecuteReplication(indexMetaData.getSettings());
this.nodes = state.getNodes();
List<ShardRouting> shards = shards(shardRoutingTable);
boolean executeOnReplica = (indexMetaData == null) || shouldExecuteReplication(indexMetaData.getSettings());
DiscoveryNodes nodes = state.getNodes();
if (shards.isEmpty()) {
logger.debug("replication phase for request [{}] on [{}] is skipped due to index deletion after primary operation", replicaRequest, shardId);
}
// we calculate number of target nodes to send replication operations, including nodes with relocating shards
AtomicInteger numberOfPendingShardInstances = new AtomicInteger();
this.totalShards = countTotalAndPending(shards, executeOnReplica, nodes, numberOfPendingShardInstances);
this.pending = numberOfPendingShardInstances;
this.shards = shards;
this.executeOnReplica = executeOnReplica;
this.nodes = nodes;
if (logger.isTraceEnabled()) {
logger.trace("replication phase started. pending [{}], action [{}], request [{}], cluster state version used [{}]", pending.get(),
transportReplicaAction, replicaRequest, state.version());
}
}
private int countTotalAndPending(List<ShardRouting> shards, boolean executeOnReplica, DiscoveryNodes nodes, AtomicInteger pending) {
assert pending.get() == 0;
int numberOfIgnoredShardInstances = performOnShards(shards, executeOnReplica, nodes, shard -> pending.incrementAndGet(), shard -> pending.incrementAndGet());
// one for the local primary copy
return 1 + numberOfIgnoredShardInstances + pending.get();
}
private int performOnShards(List<ShardRouting> shards, boolean executeOnReplica, DiscoveryNodes nodes, Consumer<ShardRouting> onLocalShard, Consumer<ShardRouting> onRelocatingShard) {
int numberOfIgnoredShardInstances = 0;
int numberOfPendingShardInstances = 0;
for (ShardRouting shard : shards) {
// the following logic to select the shards to replicate to is mirrored and explained in the doRun method below
if (shard.primary() == false && executeOnReplica == false) {
// If the replicas use shadow replicas, there is no reason to
// perform the action on the replica, so skip it and
// immediately return
// this delays mapping updates on replicas because they have
// to wait until they get the new mapping through the cluster
// state, which is why we recommend pre-defined mappings for
// indices using shadow replicas
numberOfIgnoredShardInstances++;
continue;
}
@ -898,20 +924,26 @@ public abstract class TransportReplicationAction<Request extends ReplicationRequ
numberOfIgnoredShardInstances++;
continue;
}
// we index on a replica that is initializing as well since we might not have got the event
// yet that it was started. We will get an exception IllegalShardState exception if its not started
// and that's fine, we will ignore it
// we never execute replication operation locally as primary operation has already completed locally
// hence, we ignore any local shard for replication
if (nodes.localNodeId().equals(shard.currentNodeId()) == false) {
numberOfPendingShardInstances++;
onLocalShard.accept(shard);
}
// send operation to relocating shard
// local shard can be a relocation target of a primary that is in relocated state
if (shard.relocating() && nodes.localNodeId().equals(shard.relocatingNodeId()) == false) {
numberOfPendingShardInstances++;
onRelocatingShard.accept(shard);
}
}
// one for the local primary copy
this.totalShards = 1 + numberOfPendingShardInstances + numberOfIgnoredShardInstances;
this.pending = new AtomicInteger(numberOfPendingShardInstances);
if (logger.isTraceEnabled()) {
logger.trace("replication phase started. pending [{}], action [{}], request [{}], cluster state version used [{}]", pending.get(),
transportReplicaAction, replicaRequest, state.version());
}
return numberOfIgnoredShardInstances;
}
private List<ShardRouting> shards(IndexShardRoutingTable shardRoutingTable) {
return (shardRoutingTable != null) ? shardRoutingTable.shards() : Collections.emptyList();
}
/**
@ -951,36 +983,7 @@ public abstract class TransportReplicationAction<Request extends ReplicationRequ
doFinish();
return;
}
for (ShardRouting shard : shards) {
if (shard.primary() == false && executeOnReplica == false) {
// If the replicas use shadow replicas, there is no reason to
// perform the action on the replica, so skip it and
// immediately return
// this delays mapping updates on replicas because they have
// to wait until they get the new mapping through the cluster
// state, which is why we recommend pre-defined mappings for
// indices using shadow replicas
continue;
}
if (shard.unassigned()) {
continue;
}
// we index on a replica that is initializing as well since we might not have got the event
// yet that it was started. We will get an exception IllegalShardState exception if its not started
// and that's fine, we will ignore it
// we never execute replication operation locally as primary operation has already completed locally
// hence, we ignore any local shard for replication
if (nodes.localNodeId().equals(shard.currentNodeId()) == false) {
performOnReplica(shard);
}
// send operation to relocating shard
// local shard can be a relocation target of a primary that is in relocated state
if (shard.relocating() && nodes.localNodeId().equals(shard.relocatingNodeId()) == false) {
performOnReplica(shard.buildTargetRelocatingShard());
}
}
performOnShards(shards, executeOnReplica, nodes, shard -> performOnReplica(shard), shard -> performOnReplica(shard.buildTargetRelocatingShard()));
}
/**