Writebatch call crashes

[nav568]

Hi,
I have been trying WriteBatch method to do puts, I keep running into these two types of crashes when I try to use writeBatch (everything works fine when I do just puts instead of writebatch).

org.rocksdb.RocksDBException: Corruption: unknown WriteBatch tag
    at org.rocksdb.RocksDB.write(Native Method)
    at org.rocksdb.RocksDB.write(RocksDB.java:187)
    at org.apache.samza.storage.kv.RocksDbKeyValueStore.putAll(RocksDbKeyValueStore.scala:100)
    at org.apache.samza.storage.kv.SerializedKeyValueStore.putAll(SerializedKeyValueStore.scala:57)
    at org.apache.samza.storage.kv.CachedStore.flush(CachedStore.scala:162)
    at org.apache.samza.storage.kv.CachedStore.put(CachedStore.scala:146)
    at org.apache.samza.storage.kv.NullSafeKeyValueStore.put(NullSafeKeyValueStore.scala:42)
    at org.apache.samza.storage.kv.KeyValueStorageEngine.put(KeyValueStorageEngine.scala:49)
    at com.linkedin.samza.test.perf.TestStoreAllPerformance.process(TestStoreAllPerformance.java:41)
    at org.apache.samza.container.TaskInstance.process(TaskInstance.scala:133)
    at org.apache.samza.container.RunLoop$$anonfun$process$2.apply(RunLoop.scala:93)
    at org.apache.samza.util.TimerUtils$class.updateTimer(TimerUtils.scala:37)
    at org.apache.samza.container.RunLoop.updateTimer(RunLoop.scala:36)
    at org.apache.samza.container.RunLoop.process(RunLoop.scala:79)
    at org.apache.samza.container.RunLoop.run(RunLoop.scala:65)
    at org.apache.samza.container.SamzaContainer.run(SamzaContainer.scala:553)
    at org.apache.samza.container.SamzaContainer$.safeMain(SamzaContainer.scala:106)
    at org.apache.samza.container.SamzaContainer$.main(SamzaContainer.scala:85)
    at org.apache.samza.container.SamzaContainer.main(SamzaContainer.scala)

java version "1.8.0_05"
Java(TM) SE Runtime Environment (build 1.8.0_05-b13)
Java HotSpot(TM) 64-Bit Server VM (build 25.5-b02, mixed mode)
Exception in thread "main" org.rocksdb.RocksDBException: Corruption: malformed WriteBatch (too small)
    at org.rocksdb.RocksDB.write(Native Method)
    at org.rocksdb.RocksDB.write(RocksDB.java:187)
    at org.apache.samza.storage.kv.RocksDbKeyValueStore.putAll(RocksDbKeyValueStore.scala:100)
    at org.apache.samza.storage.kv.SerializedKeyValueStore.putAll(SerializedKeyValueStore.scala:57)
    at org.apache.samza.storage.kv.CachedStore.flush(CachedStore.scala:162)
    at org.apache.samza.storage.kv.NullSafeKeyValueStore.flush(NullSafeKeyValueStore.scala:69)
    at org.apache.samza.storage.kv.KeyValueStorageEngine.flush(KeyValueStorageEngine.scala:113)
    at org.apache.samza.storage.kv.KeyValueStorageEngine.close(KeyValueStorageEngine.scala:125)
    at org.apache.samza.storage.kv.KeyValueStorageEngine.stop(KeyValueStorageEngine.scala:119)
    at org.apache.samza.storage.TaskStorageManager$$anonfun$stop$2.apply(TaskStorageManager.scala:132)
    at org.apache.samza.storage.TaskStorageManager$$anonfun$stop$2.apply(TaskStorageManager.scala:132)
    at scala.collection.Iterator$class.foreach(Iterator.scala:727)
    at scala.collection.AbstractIterator.foreach(Iterator.scala:1157)

source code on how WriteBatch is used:

  def putAll(entries: java.util.List[Entry[Array[Byte], Array[Byte]]]) {
    val batch = new WriteBatch()
    val iter = entries.iterator
    var wrote = 0
    var deletes = 0
    while (iter.hasNext) {
      wrote += 1
      val curr = iter.next()
      if (curr.getValue == null) {
        deletes += 1
        batch.remove(curr.getKey)
      } else {
        val key = curr.getKey
        val value = curr.getValue
        metrics.bytesWritten.inc(key.size + value.size)
        batch.put(key, value)
      }
    }
    db.write(new WriteOptions(), batch)
    metrics.puts.inc(wrote)
    metrics.deletes.inc(deletes)
    deletesSinceLastCompaction += deletes
  }

[yhchiang]

Hey naveenatceg, I found I am not able to reproduce such problem when I use RocksDB Java directly (as in this under-reviewed diff https://reviews.facebook.net/D22983). Can I first know whether you are able to run the RocksDBSample.java in D22983 properly?

[dhruba]

Hi Naveen, any new findings on this task?

[dhruba]

Hi Naveen, any new information on this task?

[igorcanadi]

Closing due to inactivity. @naveenatceg feel free to reopen if you still need support from us.

[a574258039]

您好，我们发现了类似的问题，请问这个org.rocksdb.RocksDBException: unknown WriteBatch tag 是什么原因那？

[deshenglijun]

看着像是线程安全问题，数据量提现不出来，数据量写入的多了，写入不同目录下就会出现这个问题，老哥是咋解决的，加锁吗？

[polarker]

It might be that batch.dispose() is not called