Examples with ClusterInfo org.opensearch.cluster.ClusterInfo used on opensource projects

Example 6 with ClusterInfo

use of org.opensearch.cluster.ClusterInfo in project OpenSearch by opensearch-project.

the class ClusterAllocationExplainIT method testUnassignedReplicaDelayedAllocation.

public void testUnassignedReplicaDelayedAllocation() throws Exception {
    logger.info("--> starting 3 nodes");
    internalCluster().startNodes(3);
    prepareIndex(1, 1);
    logger.info("--> stopping the node with the replica");
    internalCluster().stopRandomNode(InternalTestCluster.nameFilter(replicaNode().getName()));
    ensureStableCluster(2);
    assertBusy(() -> assertEquals(AllocationDecision.ALLOCATION_DELAYED, client().admin().cluster().prepareAllocationExplain().setIndex("idx").setShard(0).setPrimary(false).get().getExplanation().getShardAllocationDecision().getAllocateDecision().getAllocationDecision()));
    logger.info("--> observing delayed allocation...");
    boolean includeYesDecisions = randomBoolean();
    boolean includeDiskInfo = randomBoolean();
    ClusterAllocationExplanation explanation = runExplain(false, includeYesDecisions, includeDiskInfo);
    ShardId shardId = explanation.getShard();
    boolean isPrimary = explanation.isPrimary();
    ShardRoutingState shardRoutingState = explanation.getShardState();
    DiscoveryNode currentNode = explanation.getCurrentNode();
    UnassignedInfo unassignedInfo = explanation.getUnassignedInfo();
    ClusterInfo clusterInfo = explanation.getClusterInfo();
    AllocateUnassignedDecision allocateDecision = explanation.getShardAllocationDecision().getAllocateDecision();
    MoveDecision moveDecision = explanation.getShardAllocationDecision().getMoveDecision();
    // verify shard info
    assertEquals("idx", shardId.getIndexName());
    assertEquals(0, shardId.getId());
    assertFalse(isPrimary);
    // verify current node info
    assertNotEquals(ShardRoutingState.STARTED, shardRoutingState);
    assertNull(currentNode);
    // verify unassigned info
    assertNotNull(unassignedInfo);
    assertEquals(Reason.NODE_LEFT, unassignedInfo.getReason());
    assertEquals(AllocationStatus.NO_ATTEMPT, unassignedInfo.getLastAllocationStatus());
    // verify cluster info
    verifyClusterInfo(clusterInfo, includeDiskInfo, 2);
    // verify decision objects
    assertTrue(allocateDecision.isDecisionTaken());
    assertFalse(moveDecision.isDecisionTaken());
    assertEquals(AllocationDecision.ALLOCATION_DELAYED, allocateDecision.getAllocationDecision());
    assertThat(allocateDecision.getExplanation(), startsWith("cannot allocate because the cluster is still waiting"));
    assertThat(allocateDecision.getExplanation(), containsString("despite being allowed to allocate the shard to at least one other node"));
    assertNull(allocateDecision.getAllocationId());
    assertNull(allocateDecision.getTargetNode());
    assertEquals(60000L, allocateDecision.getConfiguredDelayInMillis());
    assertThat(allocateDecision.getRemainingDelayInMillis(), greaterThan(0L));
    assertEquals(2, allocateDecision.getNodeDecisions().size());
    String primaryNodeName = primaryNodeName();
    for (NodeAllocationResult result : allocateDecision.getNodeDecisions()) {
        assertNotNull(result.getNode());
        boolean nodeHoldingPrimary = result.getNode().getName().equals(primaryNodeName);
        if (nodeHoldingPrimary) {
            // shouldn't be able to allocate to the same node as the primary, the same shard decider should say no
            assertEquals(AllocationDecision.NO, result.getNodeDecision());
            assertThat(result.getShardStoreInfo().getMatchingBytes(), greaterThan(0L));
        } else {
            assertEquals(AllocationDecision.YES, result.getNodeDecision());
            assertNull(result.getShardStoreInfo());
        }
        if (includeYesDecisions) {
            assertThat(result.getCanAllocateDecision().getDecisions().size(), greaterThan(1));
        } else {
            // if we are not including YES decisions, then the node holding the primary should have 1 NO decision,
            // the other node should have zero NO decisions
            assertEquals(nodeHoldingPrimary ? 1 : 0, result.getCanAllocateDecision().getDecisions().size());
        }
        for (Decision d : result.getCanAllocateDecision().getDecisions()) {
            if (d.label().equals("same_shard") && nodeHoldingPrimary) {
                assertEquals(Decision.Type.NO, d.type());
                assertThat(d.getExplanation(), startsWith("a copy of this shard is already allocated to this node ["));
            } else {
                assertEquals(Decision.Type.YES, d.type());
                assertNotNull(d.getExplanation());
            }
        }
    }
    // verify JSON output
    try (XContentParser parser = getParser(explanation)) {
        verifyShardInfo(parser, false, includeDiskInfo, ShardRoutingState.UNASSIGNED);
        parser.nextToken();
        assertEquals("can_allocate", parser.currentName());
        parser.nextToken();
        assertEquals(AllocationDecision.ALLOCATION_DELAYED.toString(), parser.text());
        parser.nextToken();
        assertEquals("allocate_explanation", parser.currentName());
        parser.nextToken();
        assertThat(parser.text(), startsWith("cannot allocate because the cluster is still waiting"));
        parser.nextToken();
        assertEquals("configured_delay_in_millis", parser.currentName());
        parser.nextToken();
        assertEquals(60000L, parser.longValue());
        parser.nextToken();
        assertEquals("remaining_delay_in_millis", parser.currentName());
        parser.nextToken();
        assertThat(parser.longValue(), greaterThan(0L));
        Map<String, AllocationDecision> nodes = new HashMap<>();
        nodes.put(primaryNodeName, AllocationDecision.NO);
        String[] currentNodes = internalCluster().getNodeNames();
        nodes.put(currentNodes[0].equals(primaryNodeName) ? currentNodes[1] : currentNodes[0], AllocationDecision.YES);
        verifyNodeDecisions(parser, nodes, includeYesDecisions, true);
        assertEquals(Token.END_OBJECT, parser.nextToken());
    }
}

Example 7 with ClusterInfo

use of org.opensearch.cluster.ClusterInfo in project OpenSearch by opensearch-project.

the class ClusterAllocationExplainIT method testAllocationFilteringOnIndexCreation.

public void testAllocationFilteringOnIndexCreation() throws Exception {
    logger.info("--> starting 2 nodes");
    internalCluster().startNodes(2);
    logger.info("--> creating an index with 1 primary, 0 replicas, with allocation filtering so the primary can't be assigned");
    prepareIndex(IndexMetadata.State.OPEN, 1, 0, Settings.builder().put("index.routing.allocation.include._name", "non_existent_node").build(), ActiveShardCount.NONE);
    boolean includeYesDecisions = randomBoolean();
    boolean includeDiskInfo = randomBoolean();
    ClusterAllocationExplanation explanation = runExplain(true, includeYesDecisions, includeDiskInfo);
    ShardId shardId = explanation.getShard();
    boolean isPrimary = explanation.isPrimary();
    ShardRoutingState shardRoutingState = explanation.getShardState();
    DiscoveryNode currentNode = explanation.getCurrentNode();
    UnassignedInfo unassignedInfo = explanation.getUnassignedInfo();
    ClusterInfo clusterInfo = explanation.getClusterInfo();
    AllocateUnassignedDecision allocateDecision = explanation.getShardAllocationDecision().getAllocateDecision();
    MoveDecision moveDecision = explanation.getShardAllocationDecision().getMoveDecision();
    // verify shard info
    assertEquals("idx", shardId.getIndexName());
    assertEquals(0, shardId.getId());
    assertTrue(isPrimary);
    // verify current node info
    assertNotEquals(ShardRoutingState.STARTED, shardRoutingState);
    assertNull(currentNode);
    // verify unassigned info
    assertNotNull(unassignedInfo);
    assertEquals(Reason.INDEX_CREATED, unassignedInfo.getReason());
    assertEquals(AllocationStatus.DECIDERS_NO, unassignedInfo.getLastAllocationStatus());
    // verify cluster info
    verifyClusterInfo(clusterInfo, includeDiskInfo, 2);
    // verify decision objects
    assertTrue(allocateDecision.isDecisionTaken());
    assertFalse(moveDecision.isDecisionTaken());
    assertEquals(AllocationDecision.NO, allocateDecision.getAllocationDecision());
    assertEquals("cannot allocate because allocation is not permitted to any of the nodes", allocateDecision.getExplanation());
    assertNull(allocateDecision.getAllocationId());
    assertNull(allocateDecision.getTargetNode());
    assertEquals(0L, allocateDecision.getConfiguredDelayInMillis());
    assertEquals(0L, allocateDecision.getRemainingDelayInMillis());
    assertEquals(2, allocateDecision.getNodeDecisions().size());
    for (NodeAllocationResult result : allocateDecision.getNodeDecisions()) {
        assertNotNull(result.getNode());
        assertEquals(AllocationDecision.NO, result.getNodeDecision());
        if (includeYesDecisions) {
            assertThat(result.getCanAllocateDecision().getDecisions().size(), greaterThan(1));
        } else {
            assertEquals(1, result.getCanAllocateDecision().getDecisions().size());
        }
        for (Decision d : result.getCanAllocateDecision().getDecisions()) {
            if (d.label().equals("filter")) {
                assertEquals(Decision.Type.NO, d.type());
                assertEquals("node does not match index setting [index.routing.allocation.include] filters " + "[_name:\"non_existent_node\"]", d.getExplanation());
            }
        }
    }
    // verify JSON output
    try (XContentParser parser = getParser(explanation)) {
        verifyShardInfo(parser, true, includeDiskInfo, ShardRoutingState.UNASSIGNED);
        parser.nextToken();
        assertEquals("can_allocate", parser.currentName());
        parser.nextToken();
        String allocationDecision = parser.text();
        assertTrue(allocationDecision.equals(AllocationDecision.NO.toString()) || allocationDecision.equals(AllocationDecision.AWAITING_INFO.toString()));
        parser.nextToken();
        assertEquals("allocate_explanation", parser.currentName());
        parser.nextToken();
        if (allocationDecision.equals("awaiting_info")) {
            assertEquals("cannot allocate because information about existing shard data is still being retrieved " + "from some of the nodes", parser.text());
        } else {
            assertEquals("cannot allocate because allocation is not permitted to any of the nodes", parser.text());
        }
        Map<String, AllocationDecision> nodeDecisions = new HashMap<>();
        for (String nodeName : internalCluster().getNodeNames()) {
            nodeDecisions.put(nodeName, AllocationDecision.NO);
        }
        verifyNodeDecisions(parser, nodeDecisions, includeYesDecisions, false);
        assertEquals(Token.END_OBJECT, parser.nextToken());
    }
}

Example 8 with ClusterInfo

use of org.opensearch.cluster.ClusterInfo in project OpenSearch by opensearch-project.

the class ClusterAllocationExplainIT method testUnassignedPrimaryWithExistingIndex.

public void testUnassignedPrimaryWithExistingIndex() throws Exception {
    logger.info("--> starting 2 nodes");
    internalCluster().startNodes(2);
    prepareIndex(1, 0);
    logger.info("--> stopping the node with the primary");
    internalCluster().stopRandomNode(InternalTestCluster.nameFilter(primaryNodeName()));
    ensureStableCluster(1);
    boolean includeYesDecisions = randomBoolean();
    boolean includeDiskInfo = randomBoolean();
    ClusterAllocationExplanation explanation = runExplain(true, includeYesDecisions, includeDiskInfo);
    ShardId shardId = explanation.getShard();
    boolean isPrimary = explanation.isPrimary();
    ShardRoutingState shardState = explanation.getShardState();
    DiscoveryNode currentNode = explanation.getCurrentNode();
    UnassignedInfo unassignedInfo = explanation.getUnassignedInfo();
    ClusterInfo clusterInfo = explanation.getClusterInfo();
    AllocateUnassignedDecision allocateDecision = explanation.getShardAllocationDecision().getAllocateDecision();
    MoveDecision moveDecision = explanation.getShardAllocationDecision().getMoveDecision();
    // verify shard info
    assertEquals("idx", shardId.getIndexName());
    assertEquals(0, shardId.getId());
    assertTrue(isPrimary);
    // verify current node info
    assertNotEquals(ShardRoutingState.STARTED, shardState);
    assertNull(currentNode);
    // verify unassigned info
    assertNotNull(unassignedInfo);
    assertEquals(Reason.NODE_LEFT, unassignedInfo.getReason());
    assertTrue(unassignedInfo.getLastAllocationStatus() == AllocationStatus.FETCHING_SHARD_DATA || unassignedInfo.getLastAllocationStatus() == AllocationStatus.NO_VALID_SHARD_COPY);
    // verify cluster info
    verifyClusterInfo(clusterInfo, includeDiskInfo, 1);
    // verify decision objects
    assertTrue(allocateDecision.isDecisionTaken());
    assertFalse(moveDecision.isDecisionTaken());
    assertTrue(allocateDecision.getAllocationDecision() == AllocationDecision.NO_VALID_SHARD_COPY || allocateDecision.getAllocationDecision() == AllocationDecision.AWAITING_INFO);
    if (allocateDecision.getAllocationDecision() == AllocationDecision.NO_VALID_SHARD_COPY) {
        assertEquals("cannot allocate because a previous copy of the primary shard existed but can no longer be " + "found on the nodes in the cluster", allocateDecision.getExplanation());
    } else {
        assertEquals("cannot allocate because information about existing shard data is still being retrieved from some of the nodes", allocateDecision.getExplanation());
    }
    assertNull(allocateDecision.getAllocationId());
    assertNull(allocateDecision.getTargetNode());
    assertEquals(0L, allocateDecision.getConfiguredDelayInMillis());
    assertEquals(0L, allocateDecision.getRemainingDelayInMillis());
    if (allocateDecision.getAllocationDecision() == AllocationDecision.NO_VALID_SHARD_COPY) {
        assertEquals(1, allocateDecision.getNodeDecisions().size());
        // verify JSON output
        try (XContentParser parser = getParser(explanation)) {
            verifyShardInfo(parser, true, includeDiskInfo, ShardRoutingState.UNASSIGNED);
            parser.nextToken();
            assertEquals("can_allocate", parser.currentName());
            parser.nextToken();
            assertEquals(AllocationDecision.NO_VALID_SHARD_COPY.toString(), parser.text());
            parser.nextToken();
            assertEquals("allocate_explanation", parser.currentName());
            parser.nextToken();
            assertEquals("cannot allocate because a previous copy of the primary shard existed but can no longer be found " + "on the nodes in the cluster", parser.text());
            verifyStaleShardCopyNodeDecisions(parser, 1, Collections.emptySet());
        }
    }
}

Example 9 with ClusterInfo

use of org.opensearch.cluster.ClusterInfo in project OpenSearch by opensearch-project.

the class ClusterAllocationExplainIT method testUnassignedReplicaWithPriorCopy.

public void testUnassignedReplicaWithPriorCopy() throws Exception {
    logger.info("--> starting 3 nodes");
    List<String> nodes = internalCluster().startNodes(3);
    prepareIndex(1, 1);
    String primaryNodeName = primaryNodeName();
    nodes.remove(primaryNodeName);
    logger.info("--> shutting down all nodes except the one that holds the primary");
    Settings node0DataPathSettings = internalCluster().dataPathSettings(nodes.get(0));
    Settings node1DataPathSettings = internalCluster().dataPathSettings(nodes.get(1));
    internalCluster().stopRandomNode(InternalTestCluster.nameFilter(nodes.get(0)));
    internalCluster().stopRandomNode(InternalTestCluster.nameFilter(nodes.get(1)));
    ensureStableCluster(1);
    logger.info("--> setting allocation filtering to only allow allocation on the currently running node");
    client().admin().indices().prepareUpdateSettings("idx").setSettings(Settings.builder().put("index.routing.allocation.include._name", primaryNodeName)).get();
    logger.info("--> restarting the stopped nodes");
    internalCluster().startNode(Settings.builder().put("node.name", nodes.get(0)).put(node0DataPathSettings).build());
    internalCluster().startNode(Settings.builder().put("node.name", nodes.get(1)).put(node1DataPathSettings).build());
    ensureStableCluster(3);
    boolean includeYesDecisions = randomBoolean();
    boolean includeDiskInfo = randomBoolean();
    ClusterAllocationExplanation explanation = runExplain(false, includeYesDecisions, includeDiskInfo);
    ShardId shardId = explanation.getShard();
    boolean isPrimary = explanation.isPrimary();
    ShardRoutingState shardRoutingState = explanation.getShardState();
    DiscoveryNode currentNode = explanation.getCurrentNode();
    UnassignedInfo unassignedInfo = explanation.getUnassignedInfo();
    ClusterInfo clusterInfo = explanation.getClusterInfo();
    AllocateUnassignedDecision allocateDecision = explanation.getShardAllocationDecision().getAllocateDecision();
    MoveDecision moveDecision = explanation.getShardAllocationDecision().getMoveDecision();
    // verify shard info
    assertEquals("idx", shardId.getIndexName());
    assertEquals(0, shardId.getId());
    assertFalse(isPrimary);
    // verify current node info
    assertNotEquals(ShardRoutingState.STARTED, shardRoutingState);
    assertNull(currentNode);
    // verify unassigned info
    assertNotNull(unassignedInfo);
    assertEquals(Reason.NODE_LEFT, unassignedInfo.getReason());
    assertEquals(AllocationStatus.NO_ATTEMPT, unassignedInfo.getLastAllocationStatus());
    // verify cluster info
    verifyClusterInfo(clusterInfo, includeDiskInfo, 3);
    // verify decision objects
    assertTrue(allocateDecision.isDecisionTaken());
    assertFalse(moveDecision.isDecisionTaken());
    AllocationDecision decisionToAllocate = allocateDecision.getAllocationDecision();
    assertTrue(decisionToAllocate == AllocationDecision.AWAITING_INFO || decisionToAllocate == AllocationDecision.NO);
    if (decisionToAllocate == AllocationDecision.AWAITING_INFO) {
        assertEquals("cannot allocate because information about existing shard data is still being retrieved from some of the nodes", allocateDecision.getExplanation());
    } else {
        assertEquals("cannot allocate because allocation is not permitted to any of the nodes", allocateDecision.getExplanation());
    }
    assertNull(allocateDecision.getAllocationId());
    assertNull(allocateDecision.getTargetNode());
    assertEquals(0L, allocateDecision.getConfiguredDelayInMillis());
    assertEquals(0L, allocateDecision.getRemainingDelayInMillis());
    assertEquals(3, allocateDecision.getNodeDecisions().size());
    for (NodeAllocationResult result : allocateDecision.getNodeDecisions()) {
        assertNotNull(result.getNode());
        boolean nodeHoldingPrimary = result.getNode().getName().equals(primaryNodeName);
        assertEquals(AllocationDecision.NO, result.getNodeDecision());
        if (includeYesDecisions) {
            assertThat(result.getCanAllocateDecision().getDecisions().size(), greaterThan(1));
        } else {
            assertEquals(1, result.getCanAllocateDecision().getDecisions().size());
        }
        for (Decision d : result.getCanAllocateDecision().getDecisions()) {
            if (d.label().equals("same_shard") && nodeHoldingPrimary) {
                assertEquals(Decision.Type.NO, d.type());
                assertThat(d.getExplanation(), startsWith("a copy of this shard is already allocated to this node ["));
            } else if (d.label().equals("filter") && nodeHoldingPrimary == false) {
                assertEquals(Decision.Type.NO, d.type());
                assertEquals("node does not match index setting [index.routing.allocation.include] " + "filters [_name:\"" + primaryNodeName + "\"]", d.getExplanation());
            } else {
                assertEquals(Decision.Type.YES, d.type());
                assertNotNull(d.getExplanation());
            }
        }
    }
    // verify JSON output
    try (XContentParser parser = getParser(explanation)) {
        verifyShardInfo(parser, false, includeDiskInfo, ShardRoutingState.UNASSIGNED);
        parser.nextToken();
        assertEquals("can_allocate", parser.currentName());
        parser.nextToken();
        String allocationDecision = parser.text();
        assertTrue(allocationDecision.equals(AllocationDecision.NO.toString()) || allocationDecision.equals(AllocationDecision.AWAITING_INFO.toString()));
        parser.nextToken();
        assertEquals("allocate_explanation", parser.currentName());
        parser.nextToken();
        if (allocationDecision.equals("awaiting_info")) {
            assertEquals("cannot allocate because information about existing shard data is still being retrieved " + "from some of the nodes", parser.text());
        } else {
            assertEquals("cannot allocate because allocation is not permitted to any of the nodes", parser.text());
        }
        Map<String, AllocationDecision> nodeDecisions = new HashMap<>();
        for (String nodeName : internalCluster().getNodeNames()) {
            nodeDecisions.put(nodeName, AllocationDecision.NO);
        }
        verifyNodeDecisions(parser, nodeDecisions, includeYesDecisions, true);
        assertEquals(Token.END_OBJECT, parser.nextToken());
    }
}

Example 10 with ClusterInfo

use of org.opensearch.cluster.ClusterInfo in project OpenSearch by opensearch-project.

the class TransportClusterAllocationExplainAction method masterOperation.

@Override
protected void masterOperation(final ClusterAllocationExplainRequest request, final ClusterState state, final ActionListener<ClusterAllocationExplainResponse> listener) {
    final RoutingNodes routingNodes = state.getRoutingNodes();
    final ClusterInfo clusterInfo = clusterInfoService.getClusterInfo();
    final RoutingAllocation allocation = new RoutingAllocation(allocationDeciders, routingNodes, state, clusterInfo, snapshotsInfoService.snapshotShardSizes(), System.nanoTime());
    ShardRouting shardRouting = findShardToExplain(request, allocation);
    logger.debug("explaining the allocation for [{}], found shard [{}]", request, shardRouting);
    ClusterAllocationExplanation cae = explainShard(shardRouting, allocation, request.includeDiskInfo() ? clusterInfo : null, request.includeYesDecisions(), allocationService);
    listener.onResponse(new ClusterAllocationExplainResponse(cae));
}