[PDS-10{3841,4895}] Part 5: Don't Reuse Streams In The Stream Store

atlasdb-client/src/main/java/com/palantir/atlasdb/stream/AbstractPersistentStreamStore.java

@@ -42,6 +42,7 @@
 import com.palantir.atlasdb.transaction.impl.TxTask;
 import com.palantir.common.base.Throwables;
 import com.palantir.common.compression.StreamCompression;
+import com.palantir.common.streams.KeyedStream;
 import com.palantir.util.Pair;
 import com.palantir.util.crypto.Sha256Hash;
 
@@ -134,16 +135,19 @@ public Map<Long, Sha256Hash> storeStreams(final Transaction tx, final Map<Long,
             return ImmutableMap.of();
         }
 
-        Map<Long, StreamMetadata> idsToEmptyMetadata = Maps.transformValues(streams,
-                Functions.constant(getEmptyMetadata()));
+        Map<Long, StreamMetadata> idsToEmptyMetadata = KeyedStream.stream(streams)
+                .map($ -> getEmptyMetadata())
+                .collectToMap();
         putMetadataAndHashIndexTask(tx, idsToEmptyMetadata);
 
-        Map<Long, StreamMetadata> idsToMetadata = Maps.transformEntries(streams,
-                (id, stream) -> storeBlocksAndGetFinalMetadata(tx, id, stream));
+        Map<Long, StreamMetadata> idsToMetadata = KeyedStream.stream(streams)
+                .map((id, stream) -> storeBlocksAndGetFinalMetadata(tx, id, stream))
+                .collectToMap();
         putMetadataAndHashIndexTask(tx, idsToMetadata);
 
-        Map<Long, Sha256Hash> hashes = Maps.transformValues(idsToMetadata,
-                metadata -> new Sha256Hash(metadata.getHash().toByteArray()));
+        Map<Long, Sha256Hash> hashes = KeyedStream.stream(idsToMetadata)
+                .map(metadata -> new Sha256Hash(metadata.getHash().toByteArray()))
+                .collectToMap();
         return hashes;
     }
 

atlasdb-tests-shared/build.gradle

@@ -9,6 +9,8 @@ dependencies {
   compile project(":atlasdb-config")
   testCompile project(":atlasdb-config")
 
+  testCompile project(":commons-api")
+
   compile project(path: ":atlasdb-feign", configuration: "shadow")
 
   compile group: 'com.palantir.tracing', name: 'tracing'

atlasdb-tests-shared/src/test/java/com/palantir/atlasdb/schema/stream/StreamTest.java

@@ -40,6 +40,7 @@
 import java.util.concurrent.CountDownLatch;
 import java.util.concurrent.ExecutorService;
 import java.util.concurrent.Future;
+import java.util.concurrent.atomic.AtomicBoolean;
 
 import org.apache.commons.io.FileUtils;
 import org.apache.commons.io.IOUtils;
@@ -90,6 +91,7 @@
 import com.palantir.atlasdb.transaction.api.TransactionConflictException;
 import com.palantir.atlasdb.transaction.api.TransactionTask;
 import com.palantir.common.concurrent.PTExecutors;
+import com.palantir.common.io.ForwardingInputStream;
 import com.palantir.util.Pair;
 import com.palantir.util.crypto.Sha256Hash;
 
@@ -581,6 +583,39 @@ id1, new ByteArrayInputStream(bytes1),
         assertNull(sha256HashLongMap.get(hash3));
     }
 
+    @Test
+    public void readingStreamIdsByHashInTheSameTransactionIsPermitted() throws IOException {
+        final byte[] bytes = new byte[2 * StreamTestStreamStore.BLOCK_SIZE_IN_BYTES];
+        long id = timestampService.getFreshTimestamp();
+
+        Random rand = new Random();
+        rand.nextBytes(bytes);
+        Sha256Hash hash = Sha256Hash.computeHash(bytes);
+
+        ImmutableMap<Long, InputStream> streams = ImmutableMap.of(id, new ByteArrayInputStream(bytes));
+
+        storeStreamAndCheckHash(id, hash, streams);
+        byte[] bytesInKvs = readBytesForSingleStream(id);
+        assertArrayEquals(bytesInKvs, bytes);
+    }
+
+    @Test
+    public void streamsAreNotReused() throws IOException {
+        final byte[] bytes = new byte[2 * StreamTestStreamStore.BLOCK_SIZE_IN_BYTES];
+        long id = timestampService.getFreshTimestamp();
+
+        Random rand = new Random();
+        rand.nextBytes(bytes);
+        Sha256Hash hash = Sha256Hash.computeHash(bytes);
+
+        ImmutableMap<Long, InputStream> streams = ImmutableMap.of(id,
+                new CloseEnforcingInputStream(new ByteArrayInputStream(bytes)));
+
+        storeStreamAndCheckHash(id, hash, streams);
+        byte[] bytesInKvs = readBytesForSingleStream(id);
+        assertArrayEquals(bytesInKvs, bytes);
+    }
+
     @Test
     public void testStoreCopy() {
         final byte[] bytes = new byte[2 * StreamTestStreamStore.BLOCK_SIZE_IN_BYTES];
@@ -682,6 +717,21 @@ private Optional<InputStream> getStream(long streamId) {
         });
     }
 
+    private void storeStreamAndCheckHash(long id, Sha256Hash hash, ImmutableMap<Long, InputStream> streams) {
+        txManager.runTaskWithRetry(t -> {
+            Map<Long, Sha256Hash> hashes = defaultStore.storeStreams(t, streams);
+            assertEquals(hash, hashes.get(id));
+            return null;
+        });
+    }
+
+    private byte[] readBytesForSingleStream(long id) throws IOException {
+        return txManager.runTaskWithRetry(t -> {
+            Optional<InputStream> inputStream = defaultStore.loadSingleStream(t, id);
+            return IOUtils.toByteArray(inputStream.get());
+        });
+    }
+
     private void assertStreamDoesNotExist(final long streamId) {
         assertFalse("This element should have been deleted", getStream(streamId).isPresent());
     }
@@ -745,4 +795,27 @@ private byte[] getIncompressibleBytes(int size) {
         new Random(0).nextBytes(data);
         return data;
     }
+
+    private static class CloseEnforcingInputStream extends ForwardingInputStream {
+        private final InputStream delegate;
+        private final AtomicBoolean closed = new AtomicBoolean(false);
+
+        private CloseEnforcingInputStream(InputStream delegate) {
+            this.delegate = delegate;
+        }
+
+        @Override
+        protected InputStream delegate() {
+            if (closed.get()) {
+                throw new UnsupportedOperationException("This byte array input stream has been closed");
+            }
+            return delegate;
+        }
+
+        @Override
+        public void close() throws IOException {
+            delegate.close();
+            closed.set(true);
+        }
+    }
 }

changelog/@unreleased/pr-4444.v2.yml

@@ -0,0 +1,9 @@
+type: fix
+fix:
+  description: We no longer read and physically store streams lazily in the Atlas
+    stream store when `storeStreams()` is called. This fixes a bug where users who
+    stored and marked streams in a single transaction and subsequently referenced
+    the returned `Sha256Hash` may throw when storing streams, and/or suffer from data
+    corruption.
+  links:
+  - https://github.com/palantir/atlasdb/pull/4444