use of org.elasticsearch.action.admin.cluster.snapshots.status.SnapshotIndexStatus 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.SnapshotIndexStatus 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