Question

我一直在尝试使用Hazelcast PartitionPredicate，并注意到同一个密钥可以出现在多个分区中。有人可以解释一下这是为什么吗？或者如果你知道一篇解释这种行为的好文章？

示例代码：

创建测试缓存：

   def hzCreateTestCache(): Unit = {
        val hzClient: HazelcastInstance = getClient
        val testCache = hzClient.getMap[HzMapKey, HzMapValue]("Test")
        // delete all entries.
        testCache.destroy()
        for (i <- 1 to 100000) {
          var d = 20171224
          if (i > 50 * 1000) d = 20171225
          val key = new HzMapKey {_date = d; _Id = "id_" + f"$i%010d"}
          val value = new HzMapValue {_body = "body_" + key}
          testCache.put(key, value)
        }
      }

从缓存中读取：

  def test_read_all_entries_by_partition(): Unit = {
    val client = getClient
    val testsMap = client.getMap[HzMapKey, HzMapValue]("Test")
    println("Size: " + testsMap.size())
    val datePredicate = new PredicateBuilder().getEntryObject.get("__key#_date").equal(20171224)
    var c = 0
    var partValues = Array[String]()
    // foreach Hz partition
    client.getPartitionService.getPartitions.toArray().foreach(p => {
      // create partitionPredicate with PartitionId
      val partPredicate = new PartitionPredicate(p.asInstanceOf[Partition].getPartitionId, datePredicate)
      val values = testsMap.keySet(partPredicate)
      // Increment counter
      c = c + values.size()
      // Insert into partValues for later...
      values.toArray().foreach(i => partValues = partValues :+ i.asInstanceOf[HzMapKey].toString)
    })
    println("Count: " + partValues.length)
    // find duplicates
    partValues.groupBy(i => i.toString).map(t => (t._1, t._2.length)).foreach(e => if (e._2 > 1) println(e._1 + ": " + e._2))
  }

返回：

尺寸：100000 数：50295 ...

Hazelcast：相同的条目出现在不同的分区中

0 个答案: