Set initial capacity on HashSet/HashMap when known

core/src/main/java/org/apache/accumulo/core/client/MutationsRejectedException.java

@@ -43,7 +43,7 @@ public class MutationsRejectedException extends AccumuloException {
   private static final long serialVersionUID = 1L;
 
   private final ArrayList<ConstraintViolationSummary> cvsl = new ArrayList<>();
-  private final HashMap<TabletId,Set<SecurityErrorCode>> af = new HashMap<>();
+  private final HashMap<TabletId,Set<SecurityErrorCode>> af;
   private final HashSet<String> es = new HashSet<>();
   private final int unknownErrors;
 
@@ -69,7 +69,7 @@ public MutationsRejectedException(Instance instance, List<ConstraintViolationSum
         + "  security codes: " + hashMap.toString() + "  # server errors " + serverSideErrors.size()
         + " # exceptions " + unknownErrors, cause);
     this.cvsl.addAll(cvsList);
-    this.af.putAll(hashMap);
+    this.af = new HashMap<>(hashMap);
     this.es.addAll(serverSideErrors);
     this.unknownErrors = unknownErrors;
   }
@@ -95,7 +95,7 @@ public MutationsRejectedException(AccumuloClient client, List<ConstraintViolatio
         + "  security codes: " + format(hashMap, (ClientContext) client) + "  # server errors "
         + serverSideErrors.size() + " # exceptions " + unknownErrors, cause);
     this.cvsl.addAll(cvsList);
-    this.af.putAll(hashMap);
+    this.af = new HashMap<>(hashMap);
     this.es.addAll(serverSideErrors);
     this.unknownErrors = unknownErrors;
   }

core/src/main/java/org/apache/accumulo/core/client/rfile/RFileWriter.java

@@ -123,7 +123,7 @@ private void _startNewLocalityGroup(String name, Set<ByteSequence> columnFamilie
    * @throws IllegalStateException When default locality group already started.
    */
   public void startNewLocalityGroup(String name, List<byte[]> families) throws IOException {
-    HashSet<ByteSequence> fams = new HashSet<>();
+    HashSet<ByteSequence> fams = new HashSet<>(families.size(), 1.0f);
     for (byte[] family : families) {
       fams.add(new ArrayByteSequence(family));
     }
@@ -147,7 +147,7 @@ public void startNewLocalityGroup(String name, byte[]... families) throws IOExce
    * @throws IllegalStateException When default locality group already started.
    */
   public void startNewLocalityGroup(String name, Set<String> families) throws IOException {
-    HashSet<ByteSequence> fams = new HashSet<>();
+    HashSet<ByteSequence> fams = new HashSet<>(families.size(), 1.0f);
     for (String family : families) {
       fams.add(new ArrayByteSequence(family));
     }
@@ -162,7 +162,7 @@ public void startNewLocalityGroup(String name, Set<String> families) throws IOEx
    * @throws IllegalStateException When default locality group already started.
    */
   public void startNewLocalityGroup(String name, String... families) throws IOException {
-    HashSet<ByteSequence> fams = new HashSet<>();
+    HashSet<ByteSequence> fams = new HashSet<>(families.length, 1.0f);
     for (String family : families) {
       fams.add(new ArrayByteSequence(family));
     }

core/src/main/java/org/apache/accumulo/core/clientImpl/InstanceOperationsImpl.java

@@ -31,6 +31,7 @@
 import static org.apache.accumulo.core.util.threads.ThreadPoolNames.INSTANCE_OPS_COMPACTIONS_FINDER_POOL;
 
 import java.util.ArrayList;
+import java.util.Collection;
 import java.util.Collections;
 import java.util.ConcurrentModificationException;
 import java.util.HashSet;
@@ -214,8 +215,11 @@ public List<String> getManagerLocations() {
 
   @Override
   public Set<String> getCompactors() {
-    Set<String> compactors = new HashSet<>();
-    ExternalCompactionUtil.getCompactorAddrs(context).values().forEach(addrs -> {
+    Collection<List<HostAndPort>> compactorGroups =
+        ExternalCompactionUtil.getCompactorAddrs(context).values();
+    Set<String> compactors =
+        new HashSet<>(compactorGroups.stream().mapToInt(List::size).sum(), 1.0f);
+    compactorGroups.forEach(addrs -> {
       addrs.forEach(hp -> compactors.add(hp.toString()));
     });
     return compactors;

core/src/main/java/org/apache/accumulo/core/clientImpl/TableOperationsImpl.java

@@ -1217,11 +1217,11 @@ public Map<String,Set<Text>> getLocalityGroups(String tableName)
 
     AccumuloConfiguration conf = new ConfigurationCopy(this.getProperties(tableName));
     Map<String,Set<ByteSequence>> groups = LocalityGroupUtil.getLocalityGroups(conf);
-    Map<String,Set<Text>> groups2 = new HashMap<>();
+    Map<String,Set<Text>> groups2 = new HashMap<>(groups.size(), 1.0f);
 
     for (Entry<String,Set<ByteSequence>> entry : groups.entrySet()) {
 
-      HashSet<Text> colFams = new HashSet<>();
+      HashSet<Text> colFams = new HashSet<>(entry.getValue().size(), 1.0f);
 
       for (ByteSequence bs : entry.getValue()) {
         colFams.add(new Text(bs.toArray()));
@@ -1291,7 +1291,7 @@ public Set<Range> splitRangeByTablets(String tableName, Range range, int maxSpli
 
     mergedExtents.addAll(unmergedExtents);
 
-    Set<Range> ranges = new HashSet<>();
+    Set<Range> ranges = new HashSet<>(mergedExtents.size(), 1.0f);
     for (KeyExtent k : mergedExtents) {
       ranges.add(k.toDataRange().clip(range));
     }
@@ -1671,7 +1671,7 @@ public void importTable(String tableName, Set<String> importDirs, ImportConfigur
     boolean keepOffline = ic.isKeepOffline();
     boolean keepMapping = ic.isKeepMappings();
 
-    Set<String> checkedImportDirs = new HashSet<>();
+    Set<String> checkedImportDirs = new HashSet<>(importDirs.size(), 1.0f);
     try {
       for (String s : importDirs) {
         checkedImportDirs.add(checkPath(s, "Table", "").toString());

core/src/main/java/org/apache/accumulo/core/data/Condition.java

@@ -256,8 +256,8 @@ public Condition setIterators(IteratorSetting... iterators) {
     checkArgument(iterators != null, "iterators is null");
 
     if (iterators.length > 1) {
-      HashSet<String> names = new HashSet<>();
-      HashSet<Integer> prios = new HashSet<>();
+      HashSet<String> names = new HashSet<>(iterators.length, 1.0f);
+      HashSet<Integer> prios = new HashSet<>(iterators.length, 1.0f);
 
       for (IteratorSetting iteratorSetting : iterators) {
         if (!names.add(iteratorSetting.getName())) {

core/src/main/java/org/apache/accumulo/core/data/Mutation.java

@@ -1454,7 +1454,7 @@ public void readFields(DataInput in) throws IOException {
 
     if ((first & 0x02) == 0x02) {
       int numMutations = WritableUtils.readVInt(in);
-      this.replicationSources = new HashSet<>();
+      this.replicationSources = new HashSet<>(numMutations, 1.0f);
       for (int i = 0; i < numMutations; i++) {
         replicationSources.add(WritableUtils.readString(in));
       }

core/src/main/java/org/apache/accumulo/core/data/constraints/VisibilityConstraint.java

@@ -64,7 +64,7 @@ public List<Short> check(Environment env, Mutation mutation) {
 
     HashSet<String> ok = null;
     if (updates.size() > 1) {
-      ok = new HashSet<>();
+      ok = new HashSet<>(updates.size());
     }
 
     VisibilityEvaluator ve = null;

core/src/main/java/org/apache/accumulo/core/file/rfile/RFile.java

@@ -1449,7 +1449,7 @@ public void init(SortedKeyValueIterator<Key,Value> source, Map<String,String> op
      * @see LocalityGroupUtil#seek(FileSKVIterator, Range, String, Map)
      */
     public Map<String,ArrayList<ByteSequence>> getLocalityGroupCF() {
-      Map<String,ArrayList<ByteSequence>> cf = new HashMap<>();
+      Map<String,ArrayList<ByteSequence>> cf = new HashMap<>(localityGroups.size(), 1.0f);
 
       for (LocalityGroupMetadata lcg : localityGroups) {
         ArrayList<ByteSequence> setCF;

core/src/main/java/org/apache/accumulo/core/iteratorsImpl/system/ColumnFamilySkippingIterator.java

@@ -117,8 +117,7 @@ public void seek(Range range, Collection<ByteSequence> columnFamilies, boolean i
     if (columnFamilies instanceof Set<?>) {
       colFamSet = (Set<ByteSequence>) columnFamilies;
     } else {
-      colFamSet = new HashSet<>();
-      colFamSet.addAll(columnFamilies);
+      colFamSet = new HashSet<>(columnFamilies);
     }
 
     if (inclusive) {

core/src/main/java/org/apache/accumulo/core/metadata/schema/TabletMetadata.java

@@ -535,11 +535,12 @@ public static TabletMetadata create(String id, String prevEndRow, String endRow)
    * pulled from org.apache.accumulo.server.manager.LiveTServerSet
    */
   public static synchronized Set<TServerInstance> getLiveTServers(ClientContext context) {
-    final Set<TServerInstance> liveServers = new HashSet<>();
 
     final String path = context.getZooKeeperRoot() + Constants.ZTSERVERS;
+    final List<String> children = context.getZooCache().getChildren(path);
+    final Set<TServerInstance> liveServers = new HashSet<>(children.size());
 
-    for (String child : context.getZooCache().getChildren(path)) {
+    for (String child : children) {
       checkServer(context, path, child).ifPresent(liveServers::add);
     }
     log.trace("Found {} live tservers at ZK path: {}", liveServers.size(), path);

core/src/main/java/org/apache/accumulo/core/util/LocalityGroupUtil.java

@@ -111,8 +111,8 @@ public static void checkLocalityGroups(Map<String,String> config)
 
   public static Map<String,Set<ByteSequence>> getLocalityGroups(AccumuloConfiguration acuconf)
       throws LocalityGroupConfigurationException {
-    Map<String,Set<ByteSequence>> result = new HashMap<>();
     String[] groups = acuconf.get(Property.TABLE_LOCALITY_GROUPS).split(",");
+    Map<String,Set<ByteSequence>> result = new HashMap<>(groups.length);
     for (String group : groups) {
       if (!group.isEmpty()) {
         result.put(group, new HashSet<>());

core/src/main/java/org/apache/accumulo/core/util/compaction/ExternalCompactionUtil.java

@@ -116,10 +116,10 @@ public static Optional<HostAndPort> findCompactionCoordinator(ClientContext cont
    */
   public static Map<String,List<HostAndPort>> getCompactorAddrs(ClientContext context) {
     try {
-      final Map<String,List<HostAndPort>> queuesAndAddresses = new HashMap<>();
       final String compactorQueuesPath = context.getZooKeeperRoot() + Constants.ZCOMPACTORS;
       ZooReader zooReader = context.getZooReader();
       List<String> queues = zooReader.getChildren(compactorQueuesPath);
+      final Map<String,List<HostAndPort>> queuesAndAddresses = new HashMap<>(queues.size(), 1.0f);
       for (String queue : queues) {
         queuesAndAddresses.putIfAbsent(queue, new ArrayList<>());
         try {
@@ -254,16 +254,18 @@ public static List<RunningCompaction> getCompactionsRunningOnCompactors(ClientCo
       getCompactionIdsRunningOnCompactors(ClientContext context) {
     final ExecutorService executor = ThreadPools.getServerThreadPools()
         .getPoolBuilder(COMPACTOR_RUNNING_COMPACTION_IDS_POOL).numCoreThreads(16).build();
-    List<Future<ExternalCompactionId>> futures = new ArrayList<>();
 
-    getCompactorAddrs(context).forEach((q, hp) -> {
+    var compactors = getCompactorAddrs(context);
+    List<Future<ExternalCompactionId>> futures = new ArrayList<>(compactors.size());
+
+    compactors.forEach((q, hp) -> {
       hp.forEach(hostAndPort -> {
         futures.add(executor.submit(() -> getRunningCompactionId(hostAndPort, context)));
       });
     });
     executor.shutdown();
 
-    HashSet<ExternalCompactionId> runningIds = new HashSet<>();
+    HashSet<ExternalCompactionId> runningIds = new HashSet<>(compactors.size() / 4);
 
     futures.forEach(future -> {
       try {

server/base/src/main/java/org/apache/accumulo/server/manager/state/TabletStateChangeIterator.java

@@ -129,7 +129,7 @@ static Set<KeyExtent> decodeMigrations(DataInputStream input, int count) throws
     if (input == null) {
       return Collections.emptySet();
     }
-    Set<KeyExtent> result = new HashSet<>();
+    Set<KeyExtent> result = new HashSet<>(count, 1.0f);
     for (int i = 0; i < count; i++) {
       // need a count and cannot use InputStream.available() because its behavior is not reliable
       // across InputStream impls
@@ -142,8 +142,9 @@ private Set<TableId> parseTableIDs(String tableIDs) {
     if (tableIDs == null) {
       return null;
     }
-    Set<TableId> result = new HashSet<>();
-    for (String tableID : tableIDs.split(",")) {
+    String[] ids = tableIDs.split(",");
+    Set<TableId> result = new HashSet<>(ids.length, 1.0f);
+    for (String tableID : ids) {
       result.add(TableId.of(tableID));
     }
     return result;
@@ -154,9 +155,10 @@ private Set<TServerInstance> parseServers(String servers) {
       return null;
     }
     // parse "host:port[INSTANCE]"
-    Set<TServerInstance> result = new HashSet<>();
     if (!servers.isEmpty()) {
-      for (String part : servers.split(",")) {
+      String[] s = servers.split(",");
+      Set<TServerInstance> result = new HashSet<>(s.length, 1.0f);
+      for (String part : s) {
         String[] parts = part.split("\\[", 2);
         String hostport = parts[0];
         String instance = parts[1];
@@ -165,8 +167,9 @@ private Set<TServerInstance> parseServers(String servers) {
         }
         result.add(new TServerInstance(AddressUtil.parseAddress(hostport, false), instance));
       }
+      return result;
     }
-    return result;
+    return null;
   }
 
   private Map<TableId,MergeInfo> parseMerges(String merges) {

server/gc/src/main/java/org/apache/accumulo/gc/GCRun.java

@@ -244,8 +244,9 @@ public Map<TableId,TableState> getTableIDs() throws InterruptedException {
     while (retries <= 10) {
       try {
         zr.sync(tablesPath);
-        final Map<TableId,TableState> tids = new HashMap<>();
-        for (String table : zr.getChildren(tablesPath)) {
+        var tables = zr.getChildren(tablesPath);
+        final Map<TableId,TableState> tids = new HashMap<>(tables.size(), 1.0f);
+        for (String table : tables) {
           TableId tableId = TableId.of(table);
           TableState tableState = null;
           String statePath = context.getZooKeeperRoot() + Constants.ZTABLES + "/"
@@ -607,8 +608,9 @@ public Set<TableId> getCandidateTableIDs() throws InterruptedException {
     } else if (level == DataLevel.METADATA) {
       return Set.of(MetadataTable.ID);
     } else if (level == DataLevel.USER) {
-      Set<TableId> tableIds = new HashSet<>();
-      getTableIDs().forEach((k, v) -> {
+      var tids = getTableIDs();
+      Set<TableId> tableIds = new HashSet<>(tids.size(), 1.0f);
+      tids.forEach((k, v) -> {
         if (v == TableState.ONLINE || v == TableState.OFFLINE) {
           // Don't return tables that are NEW, DELETING, or in an
           // UNKNOWN state.

server/gc/src/main/java/org/apache/accumulo/gc/GarbageCollectionAlgorithm.java

@@ -145,7 +145,7 @@ private void removeCandidatesInUse(GarbageCollectionEnvironment gce,
 
     List<GcCandidate> candidateEntriesToBeDeleted = new ArrayList<>();
     Set<TableId> tableIdsBefore = gce.getCandidateTableIDs();
-    Set<TableId> tableIdsSeen = new HashSet<>();
+    Set<TableId> tableIdsSeen = new HashSet<>(tableIdsBefore.size());
     Iterator<Reference> iter = gce.getReferences().iterator();
     while (iter.hasNext()) {
       Reference ref = iter.next();

server/tserver/src/main/java/org/apache/accumulo/tserver/TabletClientHandler.java

@@ -177,7 +177,7 @@ public List<TKeyExtent> bulkImport(TInfo tinfo, TCredentials credentials, final
         for (Entry<TKeyExtent,Map<String,MapFileInfo>> entry : files.entrySet()) {
           TKeyExtent tke = entry.getKey();
           Map<String,MapFileInfo> fileMap = entry.getValue();
-          Map<TabletFile,MapFileInfo> fileRefMap = new HashMap<>();
+          Map<TabletFile,MapFileInfo> fileRefMap = new HashMap<>(fileMap.size(), 1.0f);
           for (Entry<String,MapFileInfo> mapping : fileMap.entrySet()) {
             Path path = new Path(mapping.getKey());
             FileSystem ns = context.getVolumeManager().getFileSystemByPath(path);
@@ -219,7 +219,7 @@ public void loadFiles(TInfo tinfo, TCredentials credentials, long tid, String di
 
     watcher.runQuietly(Constants.BULK_ARBITRATOR_TYPE, tid, () -> {
       tabletImports.forEach((tke, fileMap) -> {
-        Map<TabletFile,MapFileInfo> newFileMap = new HashMap<>();
+        Map<TabletFile,MapFileInfo> newFileMap = new HashMap<>(fileMap.size(), 1.0f);
 
         for (Entry<String,MapFileInfo> mapping : fileMap.entrySet()) {
           Path path = new Path(dir, mapping.getKey());
@@ -1209,7 +1209,9 @@ public void loadTablet(TInfo tinfo, TCredentials credentials, String lock,
           Set<KeyExtent> onlineOverlapping =
               KeyExtent.findOverlapping(extent, server.getOnlineTablets());
 
-          Set<KeyExtent> all = new HashSet<>();
+          Set<KeyExtent> all = new HashSet<>(
+              unopenedOverlapping.size() + openingOverlapping.size() + onlineOverlapping.size(),
+              1.0f);
           all.addAll(unopenedOverlapping);
           all.addAll(openingOverlapping);
           all.addAll(onlineOverlapping);

server/tserver/src/main/java/org/apache/accumulo/tserver/ThriftScanClientHandler.java

@@ -200,7 +200,7 @@ public InitialScan startScan(TInfo tinfo, TCredentials credentials, KeyExtent ex
       throw new NotServingTabletException(extent.toThrift());
     }
 
-    HashSet<Column> columnSet = new HashSet<>();
+    HashSet<Column> columnSet = new HashSet<>(columns.size(), 1.0f);
     for (TColumn tcolumn : columns) {
       columnSet.add(new Column(tcolumn));
     }
@@ -356,7 +356,7 @@ public InitialMultiScan startMultiScan(TInfo tinfo, TCredentials credentials,
       Map<String,String> executionHints, long busyTimeout)
       throws ThriftSecurityException, TSampleNotPresentException, ScanServerBusyException {
 
-    final Map<KeyExtent,List<TRange>> batch = new HashMap<>();
+    final Map<KeyExtent,List<TRange>> batch = new HashMap<>(tbatch.size(), 1.0f);
     tbatch.forEach((k, v) -> {
       batch.put(KeyExtent.fromThrift(k), v);
     });

server/tserver/src/main/java/org/apache/accumulo/tserver/session/SessionManager.java

@@ -406,7 +406,7 @@ public void run() {
   public Map<TableId,MapCounter<ScanRunState>> getActiveScansPerTable() {
     Map<TableId,MapCounter<ScanRunState>> counts = new HashMap<>();
 
-    Set<Entry<Long,Session>> copiedIdleSessions = new HashSet<>();
+    Set<Entry<Long,Session>> copiedIdleSessions = new HashSet<>(deferredCleanupQueue.size(), 1.0f);
 
     /**
      * Add sessions so that get the list returned in the active scans call
@@ -446,7 +446,8 @@ public List<ActiveScan> getActiveScans() {
 
     final List<ActiveScan> activeScans = new ArrayList<>();
     final long ct = System.currentTimeMillis();
-    final Set<Entry<Long,Session>> copiedIdleSessions = new HashSet<>();
+    final Set<Entry<Long,Session>> copiedIdleSessions =
+        new HashSet<>(deferredCleanupQueue.size(), 1.0f);
 
     /**
      * Add sessions that get the list returned in the active scans call

server/tserver/src/main/java/org/apache/accumulo/tserver/tablet/CompactableUtils.java

@@ -112,7 +112,7 @@ public class CompactableUtils {
 
   public static Map<StoredTabletFile,Pair<Key,Key>> getFirstAndLastKeys(Tablet tablet,
       Set<StoredTabletFile> allFiles) throws IOException {
-    final Map<StoredTabletFile,Pair<Key,Key>> result = new HashMap<>();
+    final Map<StoredTabletFile,Pair<Key,Key>> result = new HashMap<>(allFiles.size(), 1.0f);
     final FileOperations fileFactory = FileOperations.getInstance();
     final VolumeManager fs = tablet.getTabletServer().getVolumeManager();
     final TableConfiguration tableConf = tablet.getTableConfiguration();

server/tserver/src/main/java/org/apache/accumulo/tserver/tablet/Tablet.java

@@ -341,7 +341,7 @@ public Tablet(final TabletServer tabletServer, final KeyExtent extent,
       final AtomicLong maxTime = new AtomicLong(Long.MIN_VALUE);
       final CommitSession commitSession = getTabletMemory().getCommitSession();
       try {
-        Set<String> absPaths = new HashSet<>();
+        Set<String> absPaths = new HashSet<>(datafiles.size());
         for (TabletFile ref : datafiles.keySet()) {
           absPaths.add(ref.getPathStr());
         }
@@ -395,7 +395,7 @@ public Tablet(final TabletServer tabletServer, final KeyExtent extent,
         }
       }
       // make some closed references that represent the recovered logs
-      currentLogs = new HashSet<>();
+      currentLogs = new HashSet<>(logEntries.size(), 1.0f);
       for (LogEntry logEntry : logEntries) {
         currentLogs.add(new DfsLogger(tabletServer.getContext(), tabletServer.getServerConfig(),
             logEntry.filename, logEntry.getColumnQualifier().toString()));