/** Disable the automatic refresh. */
public void disableRefresh() {
m_refresh_disabled = true;
m_refresh_thread.setInterval(0);
}
public static void main(String[] args) throws Exception {
Natives.tryMlockall();
Settings settings =
settingsBuilder()
.put("refresh_interval", "-1")
.put(SETTING_NUMBER_OF_SHARDS, 1)
.put(SETTING_NUMBER_OF_REPLICAS, 0)
.build();
String clusterName = TermsAggregationSearchAndIndexingBenchmark.class.getSimpleName();
nodes = new InternalNode[1];
for (int i = 0; i < nodes.length; i++) {
nodes[i] =
(InternalNode)
nodeBuilder()
.settings(settingsBuilder().put(settings).put("name", "node1"))
.clusterName(clusterName)
.node();
}
Client client = nodes[0].client();
client
.admin()
.cluster()
.prepareHealth(indexName)
.setWaitForGreenStatus()
.setTimeout("10s")
.execute()
.actionGet();
try {
client
.admin()
.indices()
.prepareCreate(indexName)
.addMapping(typeName, generateMapping("eager", "lazy"))
.get();
Thread.sleep(5000);
long startTime = System.currentTimeMillis();
ObjectOpenHashSet<String> uniqueTerms = ObjectOpenHashSet.newInstance();
for (int i = 0; i < NUMBER_OF_TERMS; i++) {
boolean added;
do {
added = uniqueTerms.add(RandomStrings.randomAsciiOfLength(random, STRING_TERM_SIZE));
} while (!added);
}
String[] sValues = uniqueTerms.toArray(String.class);
long ITERS = COUNT / BATCH;
long i = 1;
int counter = 0;
for (; i <= ITERS; i++) {
BulkRequestBuilder request = client.prepareBulk();
for (int j = 0; j < BATCH; j++) {
counter++;
XContentBuilder builder = jsonBuilder().startObject();
builder.field("id", Integer.toString(counter));
final String sValue = sValues[counter % sValues.length];
builder.field("s_value", sValue);
builder.field("s_value_dv", sValue);
for (String field : new String[] {"sm_value", "sm_value_dv"}) {
builder.startArray(field);
for (int k = 0; k < NUMBER_OF_MULTI_VALUE_TERMS; k++) {
builder.value(sValues[ThreadLocalRandom.current().nextInt(sValues.length)]);
}
builder.endArray();
}
request.add(
Requests.indexRequest(indexName)
.type("type1")
.id(Integer.toString(counter))
.source(builder));
}
BulkResponse response = request.execute().actionGet();
if (response.hasFailures()) {
System.err.println("--> failures...");
}
if (((i * BATCH) % 10000) == 0) {
System.out.println("--> Indexed " + (i * BATCH));
}
}
System.out.println(
"--> Indexing took " + ((System.currentTimeMillis() - startTime) / 1000) + " seconds.");
} catch (IndexAlreadyExistsException e) {
System.out.println("--> Index already exists, ignoring indexing phase, waiting for green");
ClusterHealthResponse clusterHealthResponse =
client
.admin()
.cluster()
.prepareHealth(indexName)
.setWaitForGreenStatus()
.setTimeout("10m")
.execute()
.actionGet();
if (clusterHealthResponse.isTimedOut()) {
System.err.println("--> Timed out waiting for cluster health");
}
}
client
.admin()
.indices()
.preparePutMapping(indexName)
.setType(typeName)
.setSource(generateMapping("lazy", "lazy"))
.get();
client.admin().indices().prepareRefresh().execute().actionGet();
System.out.println(
"--> Number of docs in index: "
+ client.prepareCount().setQuery(matchAllQuery()).execute().actionGet().getCount());
String[] nodeIds = new String[nodes.length];
for (int i = 0; i < nodeIds.length; i++) {
nodeIds[i] = nodes[i].injector().getInstance(Discovery.class).localNode().getId();
}
List<TestRun> testRuns = new ArrayList<>();
testRuns.add(new TestRun("Regular field ordinals", "eager", "lazy", "s_value", "ordinals"));
testRuns.add(
new TestRun("Docvalues field ordinals", "lazy", "eager", "s_value_dv", "ordinals"));
testRuns.add(
new TestRun(
"Regular field global ordinals", "eager_global_ordinals", "lazy", "s_value", null));
testRuns.add(
new TestRun("Docvalues field global", "lazy", "eager_global_ordinals", "s_value_dv", null));
List<TestResult> testResults = new ArrayList<>();
for (TestRun testRun : testRuns) {
client
.admin()
.indices()
.preparePutMapping(indexName)
.setType(typeName)
.setSource(
generateMapping(testRun.indexedFieldEagerLoading, testRun.docValuesEagerLoading))
.get();
client.admin().indices().prepareClearCache(indexName).setFieldDataCache(true).get();
SearchThread searchThread =
new SearchThread(client, testRun.termsAggsField, testRun.termsAggsExecutionHint);
RefreshThread refreshThread = new RefreshThread(client);
System.out.println("--> Running '" + testRun.name + "' round...");
new Thread(refreshThread).start();
new Thread(searchThread).start();
Thread.sleep(2 * 60 * 1000);
refreshThread.stop();
searchThread.stop();
System.out.println("--> Avg refresh time: " + refreshThread.avgRefreshTime + " ms");
System.out.println("--> Avg query time: " + searchThread.avgQueryTime + " ms");
ClusterStatsResponse clusterStateResponse =
client.admin().cluster().prepareClusterStats().setNodesIds(nodeIds).get();
System.out.println(
"--> Heap used: " + clusterStateResponse.getNodesStats().getJvm().getHeapUsed());
ByteSizeValue fieldDataMemoryUsed =
clusterStateResponse.getIndicesStats().getFieldData().getMemorySize();
System.out.println("--> Fielddata memory size: " + fieldDataMemoryUsed);
testResults.add(
new TestResult(
testRun.name,
refreshThread.avgRefreshTime,
searchThread.avgQueryTime,
fieldDataMemoryUsed));
}
System.out.println(
"----------------------------------------- SUMMARY ----------------------------------------------");
System.out.format(
Locale.ENGLISH,
"%30s%18s%15s%15s\n",
"name",
"avg refresh time",
"avg query time",
"fieldata size");
for (TestResult testResult : testResults) {
System.out.format(
Locale.ENGLISH,
"%30s%18s%15s%15s\n",
testResult.name,
testResult.avgRefreshTime,
testResult.avgQueryTime,
testResult.fieldDataSizeInMemory);
}
System.out.println(
"----------------------------------------- SUMMARY ----------------------------------------------");
client.close();
for (InternalNode node : nodes) {
node.close();
}
}
/**
* Set the interval by which the internal data is refreshed.
*
* @param interval refresh interval in ms.
*/
public void setRefreshInterval(int interval) {
m_current_refresh_interval = interval;
m_refresh_disabled = false;
m_refresh_thread.setInterval(interval);
}