use of org.elasticsearch.cluster.DiskUsage in project elasticsearch by elastic.
the class FsProbe method stats.
public FsInfo stats(FsInfo previous, @Nullable ClusterInfo clusterInfo) throws IOException {
if (!nodeEnv.hasNodeFile()) {
return new FsInfo(System.currentTimeMillis(), null, new FsInfo.Path[0]);
}
NodePath[] dataLocations = nodeEnv.nodePaths();
FsInfo.Path[] paths = new FsInfo.Path[dataLocations.length];
for (int i = 0; i < dataLocations.length; i++) {
paths[i] = getFSInfo(dataLocations[i]);
}
FsInfo.IoStats ioStats = null;
if (Constants.LINUX) {
Set<Tuple<Integer, Integer>> devicesNumbers = new HashSet<>();
for (int i = 0; i < dataLocations.length; i++) {
if (dataLocations[i].majorDeviceNumber != -1 && dataLocations[i].minorDeviceNumber != -1) {
devicesNumbers.add(Tuple.tuple(dataLocations[i].majorDeviceNumber, dataLocations[i].minorDeviceNumber));
}
}
ioStats = ioStats(devicesNumbers, previous);
}
DiskUsage leastDiskEstimate = null;
DiskUsage mostDiskEstimate = null;
if (clusterInfo != null) {
leastDiskEstimate = clusterInfo.getNodeLeastAvailableDiskUsages().get(nodeEnv.nodeId());
mostDiskEstimate = clusterInfo.getNodeMostAvailableDiskUsages().get(nodeEnv.nodeId());
}
return new FsInfo(System.currentTimeMillis(), ioStats, paths, leastDiskEstimate, mostDiskEstimate);
}
use of org.elasticsearch.cluster.DiskUsage in project elasticsearch by elastic.
the class DiskThresholdDecider method canRemain.
@Override
public Decision canRemain(ShardRouting shardRouting, RoutingNode node, RoutingAllocation allocation) {
if (shardRouting.currentNodeId().equals(node.nodeId()) == false) {
throw new IllegalArgumentException("Shard [" + shardRouting + "] is not allocated on node: [" + node.nodeId() + "]");
}
final ClusterInfo clusterInfo = allocation.clusterInfo();
final ImmutableOpenMap<String, DiskUsage> usages = clusterInfo.getNodeLeastAvailableDiskUsages();
final Decision decision = earlyTerminate(allocation, usages);
if (decision != null) {
return decision;
}
// subtractLeavingShards is passed as true here, since this is only for shards remaining, we will *eventually* have enough disk
// since shards are moving away. No new shards will be incoming since in canAllocate we pass false for this check.
final DiskUsage usage = getDiskUsage(node, allocation, usages, true);
final String dataPath = clusterInfo.getDataPath(shardRouting);
// If this node is already above the high threshold, the shard cannot remain (get it off!)
final double freeDiskPercentage = usage.getFreeDiskAsPercentage();
final long freeBytes = usage.getFreeBytes();
if (logger.isTraceEnabled()) {
logger.trace("node [{}] has {}% free disk ({} bytes)", node.nodeId(), freeDiskPercentage, freeBytes);
}
if (dataPath == null || usage.getPath().equals(dataPath) == false) {
return allocation.decision(Decision.YES, NAME, "this shard is not allocated on the most utilized disk and can remain");
}
if (freeBytes < diskThresholdSettings.getFreeBytesThresholdHigh().getBytes()) {
if (logger.isDebugEnabled()) {
logger.debug("less than the required {} free bytes threshold ({} bytes free) on node {}, shard cannot remain", diskThresholdSettings.getFreeBytesThresholdHigh(), freeBytes, node.nodeId());
}
return allocation.decision(Decision.NO, NAME, "the shard cannot remain on this node because it is above the high watermark cluster setting [%s=%s] " + "and there is less than the required [%s] free space on node, actual free: [%s]", CLUSTER_ROUTING_ALLOCATION_HIGH_DISK_WATERMARK_SETTING.getKey(), diskThresholdSettings.getHighWatermarkRaw(), diskThresholdSettings.getFreeBytesThresholdHigh(), new ByteSizeValue(freeBytes));
}
if (freeDiskPercentage < diskThresholdSettings.getFreeDiskThresholdHigh()) {
if (logger.isDebugEnabled()) {
logger.debug("less than the required {}% free disk threshold ({}% free) on node {}, shard cannot remain", diskThresholdSettings.getFreeDiskThresholdHigh(), freeDiskPercentage, node.nodeId());
}
return allocation.decision(Decision.NO, NAME, "the shard cannot remain on this node because it is above the high watermark cluster setting [%s=%s] " + "and there is less than the required [%s%%] free disk on node, actual free: [%s%%]", CLUSTER_ROUTING_ALLOCATION_HIGH_DISK_WATERMARK_SETTING.getKey(), diskThresholdSettings.getHighWatermarkRaw(), diskThresholdSettings.getFreeDiskThresholdHigh(), freeDiskPercentage);
}
return allocation.decision(Decision.YES, NAME, "there is enough disk on this node for the shard to remain, free: [%s]", new ByteSizeValue(freeBytes));
}
use of org.elasticsearch.cluster.DiskUsage in project elasticsearch by elastic.
the class DiskThresholdDecider method freeDiskPercentageAfterShardAssigned.
/**
* Given the DiskUsage for a node and the size of the shard, return the
* percentage of free disk if the shard were to be allocated to the node.
* @param usage A DiskUsage for the node to have space computed for
* @param shardSize Size in bytes of the shard
* @return Percentage of free space after the shard is assigned to the node
*/
double freeDiskPercentageAfterShardAssigned(DiskUsage usage, Long shardSize) {
shardSize = (shardSize == null) ? 0 : shardSize;
DiskUsage newUsage = new DiskUsage(usage.getNodeId(), usage.getNodeName(), usage.getPath(), usage.getTotalBytes(), usage.getFreeBytes() - shardSize);
return newUsage.getFreeDiskAsPercentage();
}
use of org.elasticsearch.cluster.DiskUsage in project elasticsearch by elastic.
the class DiskThresholdDeciderTests method testDiskThresholdWithAbsoluteSizes.
public void testDiskThresholdWithAbsoluteSizes() {
Settings diskSettings = Settings.builder().put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_DISK_THRESHOLD_ENABLED_SETTING.getKey(), true).put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_LOW_DISK_WATERMARK_SETTING.getKey(), "30b").put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_HIGH_DISK_WATERMARK_SETTING.getKey(), "9b").build();
ImmutableOpenMap.Builder<String, DiskUsage> usagesBuilder = ImmutableOpenMap.builder();
// 90% used
usagesBuilder.put("node1", new DiskUsage("node1", "n1", "/dev/null", 100, 10));
// 90% used
usagesBuilder.put("node2", new DiskUsage("node2", "n2", "/dev/null", 100, 10));
// 40% used
usagesBuilder.put("node3", new DiskUsage("node3", "n3", "/dev/null", 100, 60));
// 20% used
usagesBuilder.put("node4", new DiskUsage("node4", "n4", "/dev/null", 100, 80));
// 15% used
usagesBuilder.put("node5", new DiskUsage("node5", "n5", "/dev/null", 100, 85));
ImmutableOpenMap<String, DiskUsage> usages = usagesBuilder.build();
ImmutableOpenMap.Builder<String, Long> shardSizesBuilder = ImmutableOpenMap.builder();
// 10 bytes
shardSizesBuilder.put("[test][0][p]", 10L);
shardSizesBuilder.put("[test][0][r]", 10L);
ImmutableOpenMap<String, Long> shardSizes = shardSizesBuilder.build();
final ClusterInfo clusterInfo = new DevNullClusterInfo(usages, usages, shardSizes);
ClusterSettings clusterSettings = new ClusterSettings(Settings.EMPTY, ClusterSettings.BUILT_IN_CLUSTER_SETTINGS);
AllocationDeciders deciders = new AllocationDeciders(Settings.EMPTY, new HashSet<>(Arrays.asList(new SameShardAllocationDecider(Settings.EMPTY, clusterSettings), makeDecider(diskSettings))));
ClusterInfoService cis = new ClusterInfoService() {
@Override
public ClusterInfo getClusterInfo() {
logger.info("--> calling fake getClusterInfo");
return clusterInfo;
}
@Override
public void addListener(Listener listener) {
// noop
}
};
AllocationService strategy = new AllocationService(Settings.builder().put("cluster.routing.allocation.node_concurrent_recoveries", 10).put(ClusterRebalanceAllocationDecider.CLUSTER_ROUTING_ALLOCATION_ALLOW_REBALANCE_SETTING.getKey(), "always").put("cluster.routing.allocation.cluster_concurrent_rebalance", -1).build(), deciders, new TestGatewayAllocator(), new BalancedShardsAllocator(Settings.EMPTY), cis);
MetaData metaData = MetaData.builder().put(IndexMetaData.builder("test").settings(settings(Version.CURRENT)).numberOfShards(1).numberOfReplicas(2)).build();
RoutingTable initialRoutingTable = RoutingTable.builder().addAsNew(metaData.index("test")).build();
ClusterState clusterState = ClusterState.builder(ClusterName.CLUSTER_NAME_SETTING.getDefault(Settings.EMPTY)).metaData(metaData).routingTable(initialRoutingTable).build();
logger.info("--> adding node1 and node2 node");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder().add(newNode("node1")).add(newNode("node2"))).build();
clusterState = strategy.reroute(clusterState, "reroute");
logShardStates(clusterState);
// Primary should initialize, even though both nodes are over the limit initialize
assertThat(clusterState.getRoutingNodes().shardsWithState(INITIALIZING).size(), equalTo(1));
String nodeWithPrimary, nodeWithoutPrimary;
if (clusterState.getRoutingNodes().node("node1").size() == 1) {
nodeWithPrimary = "node1";
nodeWithoutPrimary = "node2";
} else {
nodeWithPrimary = "node2";
nodeWithoutPrimary = "node1";
}
logger.info("--> nodeWithPrimary: {}", nodeWithPrimary);
logger.info("--> nodeWithoutPrimary: {}", nodeWithoutPrimary);
// Make node without the primary now habitable to replicas
usagesBuilder = ImmutableOpenMap.builder(usages);
// 65% used
usagesBuilder.put(nodeWithoutPrimary, new DiskUsage(nodeWithoutPrimary, "", "/dev/null", 100, 35));
usages = usagesBuilder.build();
final ClusterInfo clusterInfo2 = new DevNullClusterInfo(usages, usages, shardSizes);
cis = new ClusterInfoService() {
@Override
public ClusterInfo getClusterInfo() {
logger.info("--> calling fake getClusterInfo");
return clusterInfo2;
}
@Override
public void addListener(Listener listener) {
// noop
}
};
strategy = new AllocationService(Settings.builder().put("cluster.routing.allocation.node_concurrent_recoveries", 10).put(ClusterRebalanceAllocationDecider.CLUSTER_ROUTING_ALLOCATION_ALLOW_REBALANCE_SETTING.getKey(), "always").put("cluster.routing.allocation.cluster_concurrent_rebalance", -1).build(), deciders, new TestGatewayAllocator(), new BalancedShardsAllocator(Settings.EMPTY), cis);
clusterState = strategy.reroute(clusterState, "reroute");
logShardStates(clusterState);
// Now the replica should be able to initialize
assertThat(clusterState.getRoutingNodes().shardsWithState(INITIALIZING).size(), equalTo(2));
logger.info("--> start the shards (primaries)");
clusterState = strategy.applyStartedShards(clusterState, clusterState.getRoutingNodes().shardsWithState(INITIALIZING));
logShardStates(clusterState);
// Assert that we're able to start the primary and replica, since they were both initializing
assertThat(clusterState.getRoutingNodes().shardsWithState(ShardRoutingState.STARTED).size(), equalTo(2));
// Assert that node1 got a single shard (the primary), even though its disk usage is too high
assertThat(clusterState.getRoutingNodes().node("node1").size(), equalTo(1));
// Assert that node2 got a single shard (a replica)
assertThat(clusterState.getRoutingNodes().node("node2").size(), equalTo(1));
// Assert that one replica is still unassigned
//assertThat(clusterState.routingNodes().shardsWithState(ShardRoutingState.UNASSIGNED).size(), equalTo(1));
logger.info("--> adding node3");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder(clusterState.nodes()).add(newNode("node3"))).build();
clusterState = strategy.reroute(clusterState, "reroute");
logShardStates(clusterState);
// Assert that the replica is initialized now that node3 is available with enough space
assertThat(clusterState.getRoutingNodes().shardsWithState(ShardRoutingState.STARTED).size(), equalTo(2));
assertThat(clusterState.getRoutingNodes().shardsWithState(ShardRoutingState.INITIALIZING).size(), equalTo(1));
logger.info("--> start the shards (replicas)");
clusterState = strategy.applyStartedShards(clusterState, clusterState.getRoutingNodes().shardsWithState(INITIALIZING));
logShardStates(clusterState);
// Assert that all replicas could be started
assertThat(clusterState.getRoutingNodes().shardsWithState(ShardRoutingState.STARTED).size(), equalTo(3));
assertThat(clusterState.getRoutingNodes().node("node1").size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().node("node2").size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().node("node3").size(), equalTo(1));
logger.info("--> changing decider settings");
// Set the low threshold to 60 instead of 70
// Set the high threshold to 70 instead of 80
// node2 now should not have new shards allocated to it, but shards can remain
diskSettings = Settings.builder().put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_DISK_THRESHOLD_ENABLED_SETTING.getKey(), true).put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_LOW_DISK_WATERMARK_SETTING.getKey(), "40b").put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_HIGH_DISK_WATERMARK_SETTING.getKey(), "30b").build();
deciders = new AllocationDeciders(Settings.EMPTY, new HashSet<>(Arrays.asList(new SameShardAllocationDecider(Settings.EMPTY, clusterSettings), makeDecider(diskSettings))));
strategy = new AllocationService(Settings.builder().put("cluster.routing.allocation.node_concurrent_recoveries", 10).put(ClusterRebalanceAllocationDecider.CLUSTER_ROUTING_ALLOCATION_ALLOW_REBALANCE_SETTING.getKey(), "always").put("cluster.routing.allocation.cluster_concurrent_rebalance", -1).build(), deciders, new TestGatewayAllocator(), new BalancedShardsAllocator(Settings.EMPTY), cis);
clusterState = strategy.reroute(clusterState, "reroute");
logShardStates(clusterState);
// Shards remain started
assertThat(clusterState.getRoutingNodes().shardsWithState(STARTED).size(), equalTo(3));
assertThat(clusterState.getRoutingNodes().node("node1").size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().node("node2").size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().node("node3").size(), equalTo(1));
logger.info("--> changing settings again");
// Set the low threshold to 50 instead of 60
// Set the high threshold to 60 instead of 70
// node2 now should not have new shards allocated to it, and shards cannot remain
diskSettings = Settings.builder().put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_DISK_THRESHOLD_ENABLED_SETTING.getKey(), true).put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_LOW_DISK_WATERMARK_SETTING.getKey(), "50b").put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_HIGH_DISK_WATERMARK_SETTING.getKey(), "40b").build();
deciders = new AllocationDeciders(Settings.EMPTY, new HashSet<>(Arrays.asList(new SameShardAllocationDecider(Settings.EMPTY, clusterSettings), makeDecider(diskSettings))));
strategy = new AllocationService(Settings.builder().put("cluster.routing.allocation.node_concurrent_recoveries", 10).put(ClusterRebalanceAllocationDecider.CLUSTER_ROUTING_ALLOCATION_ALLOW_REBALANCE_SETTING.getKey(), "always").put("cluster.routing.allocation.cluster_concurrent_rebalance", -1).build(), deciders, new TestGatewayAllocator(), new BalancedShardsAllocator(Settings.EMPTY), cis);
clusterState = strategy.reroute(clusterState, "reroute");
logShardStates(clusterState);
// Shards remain started
assertThat(clusterState.getRoutingNodes().shardsWithState(STARTED).size(), equalTo(3));
assertThat(clusterState.getRoutingNodes().node("node1").size(), equalTo(1));
// Shard hasn't been moved off of node2 yet because there's nowhere for it to go
assertThat(clusterState.getRoutingNodes().node("node2").size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().node("node3").size(), equalTo(1));
logger.info("--> adding node4");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder(clusterState.nodes()).add(newNode("node4"))).build();
clusterState = strategy.reroute(clusterState, "reroute");
logShardStates(clusterState);
// Shards remain started
assertThat(clusterState.getRoutingNodes().shardsWithState(STARTED).size(), equalTo(2));
// One shard is relocating off of node1
assertThat(clusterState.getRoutingNodes().shardsWithState(RELOCATING).size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().shardsWithState(INITIALIZING).size(), equalTo(1));
logger.info("--> apply INITIALIZING shards");
clusterState = strategy.applyStartedShards(clusterState, clusterState.getRoutingNodes().shardsWithState(INITIALIZING));
logShardStates(clusterState);
// primary shard already has been relocated away
assertThat(clusterState.getRoutingNodes().node(nodeWithPrimary).size(), equalTo(0));
// node with increased space still has its shard
assertThat(clusterState.getRoutingNodes().node(nodeWithoutPrimary).size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().node("node3").size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().node("node4").size(), equalTo(1));
logger.info("--> adding node5");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder(clusterState.nodes()).add(newNode("node5"))).build();
clusterState = strategy.reroute(clusterState, "reroute");
logShardStates(clusterState);
// Shards remain started on node3 and node4
assertThat(clusterState.getRoutingNodes().shardsWithState(STARTED).size(), equalTo(2));
// One shard is relocating off of node2 now
assertThat(clusterState.getRoutingNodes().shardsWithState(RELOCATING).size(), equalTo(1));
// Initializing on node5
assertThat(clusterState.getRoutingNodes().shardsWithState(INITIALIZING).size(), equalTo(1));
logger.info("--> apply INITIALIZING shards");
clusterState = strategy.applyStartedShards(clusterState, clusterState.getRoutingNodes().shardsWithState(INITIALIZING));
logger.info("--> final cluster state:");
logShardStates(clusterState);
// Node1 still has no shards because it has no space for them
assertThat(clusterState.getRoutingNodes().node("node1").size(), equalTo(0));
// Node5 is available now, so the shard is moved off of node2
assertThat(clusterState.getRoutingNodes().node("node2").size(), equalTo(0));
assertThat(clusterState.getRoutingNodes().node("node3").size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().node("node4").size(), equalTo(1));
assertThat(clusterState.getRoutingNodes().node("node5").size(), equalTo(1));
}
use of org.elasticsearch.cluster.DiskUsage in project elasticsearch by elastic.
the class DiskThresholdDeciderTests method testDiskThresholdWithShardSizes.
public void testDiskThresholdWithShardSizes() {
Settings diskSettings = Settings.builder().put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_DISK_THRESHOLD_ENABLED_SETTING.getKey(), true).put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_LOW_DISK_WATERMARK_SETTING.getKey(), 0.7).put(DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_HIGH_DISK_WATERMARK_SETTING.getKey(), "71%").build();
ImmutableOpenMap.Builder<String, DiskUsage> usagesBuilder = ImmutableOpenMap.builder();
// 69% used
usagesBuilder.put("node1", new DiskUsage("node1", "n1", "/dev/null", 100, 31));
// 99% used
usagesBuilder.put("node2", new DiskUsage("node2", "n2", "/dev/null", 100, 1));
ImmutableOpenMap<String, DiskUsage> usages = usagesBuilder.build();
ImmutableOpenMap.Builder<String, Long> shardSizesBuilder = ImmutableOpenMap.builder();
// 10 bytes
shardSizesBuilder.put("[test][0][p]", 10L);
ImmutableOpenMap<String, Long> shardSizes = shardSizesBuilder.build();
final ClusterInfo clusterInfo = new DevNullClusterInfo(usages, usages, shardSizes);
AllocationDeciders deciders = new AllocationDeciders(Settings.EMPTY, new HashSet<>(Arrays.asList(new SameShardAllocationDecider(Settings.EMPTY, new ClusterSettings(Settings.EMPTY, ClusterSettings.BUILT_IN_CLUSTER_SETTINGS)), makeDecider(diskSettings))));
ClusterInfoService cis = new ClusterInfoService() {
@Override
public ClusterInfo getClusterInfo() {
logger.info("--> calling fake getClusterInfo");
return clusterInfo;
}
@Override
public void addListener(Listener listener) {
// noop
}
};
AllocationService strategy = new AllocationService(Settings.builder().put("cluster.routing.allocation.node_concurrent_recoveries", 10).put(ClusterRebalanceAllocationDecider.CLUSTER_ROUTING_ALLOCATION_ALLOW_REBALANCE_SETTING.getKey(), "always").put("cluster.routing.allocation.cluster_concurrent_rebalance", -1).build(), deciders, new TestGatewayAllocator(), new BalancedShardsAllocator(Settings.EMPTY), cis);
MetaData metaData = MetaData.builder().put(IndexMetaData.builder("test").settings(settings(Version.CURRENT)).numberOfShards(1).numberOfReplicas(0)).build();
RoutingTable routingTable = RoutingTable.builder().addAsNew(metaData.index("test")).build();
ClusterState clusterState = ClusterState.builder(ClusterName.CLUSTER_NAME_SETTING.getDefault(Settings.EMPTY)).metaData(metaData).routingTable(routingTable).build();
logger.info("--> adding node1");
clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder().add(newNode("node1")).add(// node2 is added because DiskThresholdDecider automatically ignore single-node clusters
newNode("node2"))).build();
routingTable = strategy.reroute(clusterState, "reroute").routingTable();
clusterState = ClusterState.builder(clusterState).routingTable(routingTable).build();
logger.info("--> start the shards (primaries)");
routingTable = strategy.applyStartedShards(clusterState, clusterState.getRoutingNodes().shardsWithState(INITIALIZING)).routingTable();
clusterState = ClusterState.builder(clusterState).routingTable(routingTable).build();
logShardStates(clusterState);
// Shard can't be allocated to node1 (or node2) because it would cause too much usage
assertThat(clusterState.getRoutingNodes().shardsWithState(INITIALIZING).size(), equalTo(0));
// No shards are started, no nodes have enough disk for allocation
assertThat(clusterState.getRoutingNodes().shardsWithState(STARTED).size(), equalTo(0));
}
Aggregations