use of org.elasticsearch.action.admin.cluster.snapshots.status.SnapshotsStatusResponse in project elasticsearch by elastic.
the class SnapshotBlocksIT method testSnapshotStatusWithBlocks.
public void testSnapshotStatusWithBlocks() {
// This test checks that the Snapshot Status operation is never blocked, even if the cluster is read only.
try {
setClusterReadOnly(true);
SnapshotsStatusResponse response = client().admin().cluster().prepareSnapshotStatus(REPOSITORY_NAME).setSnapshots(SNAPSHOT_NAME).execute().actionGet();
assertThat(response.getSnapshots(), hasSize(1));
assertThat(response.getSnapshots().get(0).getState().completed(), equalTo(true));
} finally {
setClusterReadOnly(false);
}
}
use of org.elasticsearch.action.admin.cluster.snapshots.status.SnapshotsStatusResponse in project elasticsearch by elastic.
the class SharedClusterSnapshotRestoreIT method testDataFileFailureDuringSnapshot.
public void testDataFileFailureDuringSnapshot() throws Exception {
Client client = client();
logger.info("--> creating repository");
assertAcked(client.admin().cluster().preparePutRepository("test-repo").setType("mock").setSettings(Settings.builder().put("location", randomRepoPath()).put("random", randomAsciiOfLength(10)).put("random_data_file_io_exception_rate", 0.3)));
createIndex("test-idx");
ensureGreen();
logger.info("--> indexing some data");
for (int i = 0; i < 100; i++) {
index("test-idx", "doc", Integer.toString(i), "foo", "bar" + i);
}
refresh();
assertThat(client.prepareSearch("test-idx").setSize(0).get().getHits().getTotalHits(), equalTo(100L));
logger.info("--> snapshot");
CreateSnapshotResponse createSnapshotResponse = client.admin().cluster().prepareCreateSnapshot("test-repo", "test-snap").setWaitForCompletion(true).setIndices("test-idx").get();
if (createSnapshotResponse.getSnapshotInfo().totalShards() == createSnapshotResponse.getSnapshotInfo().successfulShards()) {
logger.info("--> no failures");
// If we are here, that means we didn't have any failures, let's check it
assertThat(getFailureCount("test-repo"), equalTo(0L));
} else {
logger.info("--> some failures");
assertThat(getFailureCount("test-repo"), greaterThan(0L));
assertThat(createSnapshotResponse.getSnapshotInfo().shardFailures().size(), greaterThan(0));
for (SnapshotShardFailure shardFailure : createSnapshotResponse.getSnapshotInfo().shardFailures()) {
assertThat(shardFailure.nodeId(), notNullValue());
assertThat(shardFailure.index(), equalTo("test-idx"));
}
GetSnapshotsResponse getSnapshotsResponse = client.admin().cluster().prepareGetSnapshots("test-repo").addSnapshots("test-snap").get();
assertThat(getSnapshotsResponse.getSnapshots().size(), equalTo(1));
SnapshotInfo snapshotInfo = getSnapshotsResponse.getSnapshots().get(0);
assertThat(snapshotInfo.state(), equalTo(SnapshotState.PARTIAL));
assertThat(snapshotInfo.shardFailures().size(), greaterThan(0));
assertThat(snapshotInfo.totalShards(), greaterThan(snapshotInfo.successfulShards()));
// Verify that snapshot status also contains the same failures
SnapshotsStatusResponse snapshotsStatusResponse = client.admin().cluster().prepareSnapshotStatus("test-repo").addSnapshots("test-snap").get();
assertThat(snapshotsStatusResponse.getSnapshots().size(), equalTo(1));
SnapshotStatus snapshotStatus = snapshotsStatusResponse.getSnapshots().get(0);
assertThat(snapshotStatus.getIndices().size(), equalTo(1));
SnapshotIndexStatus indexStatus = snapshotStatus.getIndices().get("test-idx");
assertThat(indexStatus, notNullValue());
assertThat(indexStatus.getShardsStats().getFailedShards(), equalTo(snapshotInfo.failedShards()));
assertThat(indexStatus.getShardsStats().getDoneShards(), equalTo(snapshotInfo.successfulShards()));
assertThat(indexStatus.getShards().size(), equalTo(snapshotInfo.totalShards()));
int numberOfFailures = 0;
for (SnapshotIndexShardStatus shardStatus : indexStatus.getShards().values()) {
if (shardStatus.getStage() == SnapshotIndexShardStage.FAILURE) {
assertThat(shardStatus.getFailure(), notNullValue());
numberOfFailures++;
} else {
assertThat(shardStatus.getFailure(), nullValue());
}
}
assertThat(indexStatus.getShardsStats().getFailedShards(), equalTo(numberOfFailures));
}
}
use of org.elasticsearch.action.admin.cluster.snapshots.status.SnapshotsStatusResponse in project elasticsearch by elastic.
the class DedicatedClusterSnapshotRestoreIT method testRestoreIndexWithMissingShards.
public void testRestoreIndexWithMissingShards() throws Exception {
logger.info("--> start 2 nodes");
internalCluster().startNode();
internalCluster().startNode();
cluster().wipeIndices("_all");
logger.info("--> create an index that will have some unallocated shards");
assertAcked(prepareCreate("test-idx-some", 2, Settings.builder().put("number_of_shards", 6).put("number_of_replicas", 0)));
ensureGreen();
logger.info("--> indexing some data into test-idx-some");
for (int i = 0; i < 100; i++) {
index("test-idx-some", "doc", Integer.toString(i), "foo", "bar" + i);
}
refresh();
assertThat(client().prepareSearch("test-idx-some").setSize(0).get().getHits().getTotalHits(), equalTo(100L));
logger.info("--> shutdown one of the nodes");
internalCluster().stopRandomDataNode();
assertThat(client().admin().cluster().prepareHealth().setWaitForEvents(Priority.LANGUID).setTimeout("1m").setWaitForNodes("<2").execute().actionGet().isTimedOut(), equalTo(false));
logger.info("--> create an index that will have all allocated shards");
assertAcked(prepareCreate("test-idx-all", 1, Settings.builder().put("number_of_shards", 6).put("number_of_replicas", 0)));
ensureGreen("test-idx-all");
logger.info("--> create an index that will be closed");
assertAcked(prepareCreate("test-idx-closed", 1, Settings.builder().put("number_of_shards", 4).put("number_of_replicas", 0)));
ensureGreen("test-idx-closed");
logger.info("--> indexing some data into test-idx-all");
for (int i = 0; i < 100; i++) {
index("test-idx-all", "doc", Integer.toString(i), "foo", "bar" + i);
index("test-idx-closed", "doc", Integer.toString(i), "foo", "bar" + i);
}
// don't refresh test-idx-some it will take 30 sec until it times out...
refresh("test-idx-closed", "test-idx-all");
assertThat(client().prepareSearch("test-idx-all").setSize(0).get().getHits().getTotalHits(), equalTo(100L));
assertAcked(client().admin().indices().prepareClose("test-idx-closed"));
logger.info("--> create an index that will have no allocated shards");
assertAcked(prepareCreate("test-idx-none", 1, Settings.builder().put("number_of_shards", 6).put("index.routing.allocation.include.tag", "nowhere").put("number_of_replicas", 0)).setWaitForActiveShards(ActiveShardCount.NONE).get());
assertTrue(client().admin().indices().prepareExists("test-idx-none").get().isExists());
logger.info("--> creating repository");
PutRepositoryResponse putRepositoryResponse = client().admin().cluster().preparePutRepository("test-repo").setType("fs").setSettings(Settings.builder().put("location", randomRepoPath())).execute().actionGet();
assertThat(putRepositoryResponse.isAcknowledged(), equalTo(true));
logger.info("--> start snapshot with default settings and closed index - should be blocked");
assertBlocked(client().admin().cluster().prepareCreateSnapshot("test-repo", "test-snap-1").setIndices("test-idx-all", "test-idx-none", "test-idx-some", "test-idx-closed").setWaitForCompletion(true), MetaDataIndexStateService.INDEX_CLOSED_BLOCK);
logger.info("--> start snapshot with default settings without a closed index - should fail");
CreateSnapshotResponse createSnapshotResponse = client().admin().cluster().prepareCreateSnapshot("test-repo", "test-snap-1").setIndices("test-idx-all", "test-idx-none", "test-idx-some").setWaitForCompletion(true).execute().actionGet();
assertThat(createSnapshotResponse.getSnapshotInfo().state(), equalTo(SnapshotState.FAILED));
assertThat(createSnapshotResponse.getSnapshotInfo().reason(), containsString("Indices don't have primary shards"));
if (randomBoolean()) {
logger.info("checking snapshot completion using status");
client().admin().cluster().prepareCreateSnapshot("test-repo", "test-snap-2").setIndices("test-idx-all", "test-idx-none", "test-idx-some").setWaitForCompletion(false).setPartial(true).execute().actionGet();
assertBusy(new Runnable() {
@Override
public void run() {
SnapshotsStatusResponse snapshotsStatusResponse = client().admin().cluster().prepareSnapshotStatus("test-repo").setSnapshots("test-snap-2").get();
List<SnapshotStatus> snapshotStatuses = snapshotsStatusResponse.getSnapshots();
assertEquals(snapshotStatuses.size(), 1);
logger.trace("current snapshot status [{}]", snapshotStatuses.get(0));
assertTrue(snapshotStatuses.get(0).getState().completed());
}
}, 1, TimeUnit.MINUTES);
SnapshotsStatusResponse snapshotsStatusResponse = client().admin().cluster().prepareSnapshotStatus("test-repo").setSnapshots("test-snap-2").get();
List<SnapshotStatus> snapshotStatuses = snapshotsStatusResponse.getSnapshots();
assertThat(snapshotStatuses.size(), equalTo(1));
SnapshotStatus snapshotStatus = snapshotStatuses.get(0);
logger.info("State: [{}], Reason: [{}]", createSnapshotResponse.getSnapshotInfo().state(), createSnapshotResponse.getSnapshotInfo().reason());
assertThat(snapshotStatus.getShardsStats().getTotalShards(), equalTo(18));
assertThat(snapshotStatus.getShardsStats().getDoneShards(), lessThan(12));
assertThat(snapshotStatus.getShardsStats().getDoneShards(), greaterThan(6));
// There is slight delay between snapshot being marked as completed in the cluster state and on the file system
// After it was marked as completed in the cluster state - we need to check if it's completed on the file system as well
assertBusy(new Runnable() {
@Override
public void run() {
GetSnapshotsResponse response = client().admin().cluster().prepareGetSnapshots("test-repo").setSnapshots("test-snap-2").get();
assertThat(response.getSnapshots().size(), equalTo(1));
SnapshotInfo snapshotInfo = response.getSnapshots().get(0);
assertTrue(snapshotInfo.state().completed());
assertEquals(SnapshotState.PARTIAL, snapshotInfo.state());
}
}, 1, TimeUnit.MINUTES);
} else {
logger.info("checking snapshot completion using wait_for_completion flag");
createSnapshotResponse = client().admin().cluster().prepareCreateSnapshot("test-repo", "test-snap-2").setIndices("test-idx-all", "test-idx-none", "test-idx-some").setWaitForCompletion(true).setPartial(true).execute().actionGet();
logger.info("State: [{}], Reason: [{}]", createSnapshotResponse.getSnapshotInfo().state(), createSnapshotResponse.getSnapshotInfo().reason());
assertThat(createSnapshotResponse.getSnapshotInfo().totalShards(), equalTo(18));
assertThat(createSnapshotResponse.getSnapshotInfo().successfulShards(), lessThan(12));
assertThat(createSnapshotResponse.getSnapshotInfo().successfulShards(), greaterThan(6));
assertThat(client().admin().cluster().prepareGetSnapshots("test-repo").setSnapshots("test-snap-2").execute().actionGet().getSnapshots().get(0).state(), equalTo(SnapshotState.PARTIAL));
}
assertAcked(client().admin().indices().prepareClose("test-idx-some", "test-idx-all").execute().actionGet());
logger.info("--> restore incomplete snapshot - should fail");
assertThrows(client().admin().cluster().prepareRestoreSnapshot("test-repo", "test-snap-2").setRestoreGlobalState(false).setWaitForCompletion(true).execute(), SnapshotRestoreException.class);
logger.info("--> restore snapshot for the index that was snapshotted completely");
RestoreSnapshotResponse restoreSnapshotResponse = client().admin().cluster().prepareRestoreSnapshot("test-repo", "test-snap-2").setRestoreGlobalState(false).setIndices("test-idx-all").setWaitForCompletion(true).execute().actionGet();
assertThat(restoreSnapshotResponse.getRestoreInfo(), notNullValue());
assertThat(restoreSnapshotResponse.getRestoreInfo().totalShards(), equalTo(6));
assertThat(restoreSnapshotResponse.getRestoreInfo().successfulShards(), equalTo(6));
assertThat(restoreSnapshotResponse.getRestoreInfo().failedShards(), equalTo(0));
assertThat(client().prepareSearch("test-idx-all").setSize(0).get().getHits().getTotalHits(), equalTo(100L));
logger.info("--> restore snapshot for the partial index");
cluster().wipeIndices("test-idx-some");
restoreSnapshotResponse = client().admin().cluster().prepareRestoreSnapshot("test-repo", "test-snap-2").setRestoreGlobalState(false).setIndices("test-idx-some").setPartial(true).setWaitForCompletion(true).get();
assertThat(restoreSnapshotResponse.getRestoreInfo(), notNullValue());
assertThat(restoreSnapshotResponse.getRestoreInfo().totalShards(), equalTo(6));
assertThat(restoreSnapshotResponse.getRestoreInfo().successfulShards(), allOf(greaterThan(0), lessThan(6)));
assertThat(restoreSnapshotResponse.getRestoreInfo().failedShards(), greaterThan(0));
assertThat(client().prepareSearch("test-idx-some").setSize(0).get().getHits().getTotalHits(), allOf(greaterThan(0L), lessThan(100L)));
logger.info("--> restore snapshot for the index that didn't have any shards snapshotted successfully");
cluster().wipeIndices("test-idx-none");
restoreSnapshotResponse = client().admin().cluster().prepareRestoreSnapshot("test-repo", "test-snap-2").setRestoreGlobalState(false).setIndices("test-idx-none").setPartial(true).setWaitForCompletion(true).get();
assertThat(restoreSnapshotResponse.getRestoreInfo(), notNullValue());
assertThat(restoreSnapshotResponse.getRestoreInfo().totalShards(), equalTo(6));
assertThat(restoreSnapshotResponse.getRestoreInfo().successfulShards(), equalTo(0));
assertThat(restoreSnapshotResponse.getRestoreInfo().failedShards(), equalTo(6));
assertThat(client().prepareSearch("test-idx-some").setSize(0).get().getHits().getTotalHits(), allOf(greaterThan(0L), lessThan(100L)));
}
use of org.elasticsearch.action.admin.cluster.snapshots.status.SnapshotsStatusResponse in project elasticsearch by elastic.
the class SharedClusterSnapshotRestoreIT method testSnapshotStatus.
public void testSnapshotStatus() throws Exception {
Client client = client();
Path repositoryLocation = randomRepoPath();
logger.info("--> creating repository");
PutRepositoryResponse putRepositoryResponse = client.admin().cluster().preparePutRepository("test-repo").setType("mock").setSettings(Settings.builder().put("location", repositoryLocation).put("random", randomAsciiOfLength(10)).put("wait_after_unblock", 200)).get();
assertThat(putRepositoryResponse.isAcknowledged(), equalTo(true));
// Create index on 2 nodes and make sure each node has a primary by setting no replicas
assertAcked(prepareCreate("test-idx", 2, Settings.builder().put("number_of_replicas", 0)));
logger.info("--> indexing some data");
for (int i = 0; i < 100; i++) {
index("test-idx", "doc", Integer.toString(i), "foo", "bar" + i);
}
refresh();
assertThat(client.prepareSearch("test-idx").setSize(0).get().getHits().getTotalHits(), equalTo(100L));
// Pick one node and block it
String blockedNode = blockNodeWithIndex("test-repo", "test-idx");
logger.info("--> snapshot");
client.admin().cluster().prepareCreateSnapshot("test-repo", "test-snap").setWaitForCompletion(false).setIndices("test-idx").get();
logger.info("--> waiting for block to kick in");
waitForBlock(blockedNode, "test-repo", TimeValue.timeValueSeconds(60));
logger.info("--> execution was blocked on node [{}], checking snapshot status with specified repository and snapshot", blockedNode);
SnapshotsStatusResponse response = client.admin().cluster().prepareSnapshotStatus("test-repo").execute().actionGet();
assertThat(response.getSnapshots().size(), equalTo(1));
SnapshotStatus snapshotStatus = response.getSnapshots().get(0);
assertThat(snapshotStatus.getState(), equalTo(SnapshotsInProgress.State.STARTED));
// We blocked the node during data write operation, so at least one shard snapshot should be in STARTED stage
assertThat(snapshotStatus.getShardsStats().getStartedShards(), greaterThan(0));
for (SnapshotIndexShardStatus shardStatus : snapshotStatus.getIndices().get("test-idx")) {
if (shardStatus.getStage() == SnapshotIndexShardStage.STARTED) {
assertThat(shardStatus.getNodeId(), notNullValue());
}
}
logger.info("--> checking snapshot status for all currently running and snapshot with empty repository");
response = client.admin().cluster().prepareSnapshotStatus().execute().actionGet();
assertThat(response.getSnapshots().size(), equalTo(1));
snapshotStatus = response.getSnapshots().get(0);
assertThat(snapshotStatus.getState(), equalTo(SnapshotsInProgress.State.STARTED));
// We blocked the node during data write operation, so at least one shard snapshot should be in STARTED stage
assertThat(snapshotStatus.getShardsStats().getStartedShards(), greaterThan(0));
for (SnapshotIndexShardStatus shardStatus : snapshotStatus.getIndices().get("test-idx")) {
if (shardStatus.getStage() == SnapshotIndexShardStage.STARTED) {
assertThat(shardStatus.getNodeId(), notNullValue());
}
}
logger.info("--> checking that _current returns the currently running snapshot");
GetSnapshotsResponse getResponse = client.admin().cluster().prepareGetSnapshots("test-repo").setCurrentSnapshot().execute().actionGet();
assertThat(getResponse.getSnapshots().size(), equalTo(1));
SnapshotInfo snapshotInfo = getResponse.getSnapshots().get(0);
assertThat(snapshotInfo.state(), equalTo(SnapshotState.IN_PROGRESS));
logger.info("--> unblocking blocked node");
unblockNode("test-repo", blockedNode);
snapshotInfo = waitForCompletion("test-repo", "test-snap", TimeValue.timeValueSeconds(600));
logger.info("Number of failed shards [{}]", snapshotInfo.shardFailures().size());
logger.info("--> done");
logger.info("--> checking snapshot status again after snapshot is done");
response = client.admin().cluster().prepareSnapshotStatus("test-repo").addSnapshots("test-snap").execute().actionGet();
snapshotStatus = response.getSnapshots().get(0);
assertThat(snapshotStatus.getIndices().size(), equalTo(1));
SnapshotIndexStatus indexStatus = snapshotStatus.getIndices().get("test-idx");
assertThat(indexStatus, notNullValue());
assertThat(indexStatus.getShardsStats().getInitializingShards(), equalTo(0));
assertThat(indexStatus.getShardsStats().getFailedShards(), equalTo(snapshotInfo.failedShards()));
assertThat(indexStatus.getShardsStats().getDoneShards(), equalTo(snapshotInfo.successfulShards()));
assertThat(indexStatus.getShards().size(), equalTo(snapshotInfo.totalShards()));
logger.info("--> checking snapshot status after it is done with empty repository");
response = client.admin().cluster().prepareSnapshotStatus().execute().actionGet();
assertThat(response.getSnapshots().size(), equalTo(0));
logger.info("--> checking that _current no longer returns the snapshot");
assertThat(client.admin().cluster().prepareGetSnapshots("test-repo").addSnapshots("_current").execute().actionGet().getSnapshots().isEmpty(), equalTo(true));
// test that getting an unavailable snapshot status throws an exception if ignoreUnavailable is false on the request
SnapshotMissingException ex = expectThrows(SnapshotMissingException.class, () -> client.admin().cluster().prepareSnapshotStatus("test-repo").addSnapshots("test-snap-doesnt-exist").get());
assertEquals("[test-repo:test-snap-doesnt-exist] is missing", ex.getMessage());
// test that getting an unavailable snapshot status does not throw an exception if ignoreUnavailable is true on the request
response = client.admin().cluster().prepareSnapshotStatus("test-repo").addSnapshots("test-snap-doesnt-exist").setIgnoreUnavailable(true).get();
assertTrue(response.getSnapshots().isEmpty());
// test getting snapshot status for available and unavailable snapshots where ignoreUnavailable is true
// (available one should be returned)
response = client.admin().cluster().prepareSnapshotStatus("test-repo").addSnapshots("test-snap", "test-snap-doesnt-exist").setIgnoreUnavailable(true).get();
assertEquals(1, response.getSnapshots().size());
assertEquals("test-snap", response.getSnapshots().get(0).getSnapshot().getSnapshotId().getName());
}
Aggregations