drill-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From paul-rogers <...@git.apache.org>
Subject [GitHub] drill pull request #822: DRILL-5457: Spill implementation for Hash Aggregate
Date Sat, 27 May 2017 05:38:12 GMT
Github user paul-rogers commented on a diff in the pull request:

    https://github.com/apache/drill/pull/822#discussion_r118812251
  
    --- Diff: exec/java-exec/src/main/java/org/apache/drill/exec/physical/impl/aggregate/HashAggTemplate.java
---
    @@ -266,17 +508,138 @@ public void setup(HashAggregate hashAggrConfig, HashTableConfig
htConfig, Fragme
           }
         }
     
    -    ChainedHashTable ht =
    +    spillSet = new SpillSet(context,hashAggrConfig, UserBitShared.CoreOperatorType.HASH_AGGREGATE);
    +    baseHashTable =
             new ChainedHashTable(htConfig, context, allocator, incoming, null /* no incoming
probe */, outgoing);
    -    this.htable = ht.createAndSetupHashTable(groupByOutFieldIds);
    -
    +    this.groupByOutFieldIds = groupByOutFieldIds; // retain these for delayedSetup, and
to allow recreating hash tables (after a spill)
         numGroupByOutFields = groupByOutFieldIds.length;
    -    batchHolders = new ArrayList<BatchHolder>();
    -    // First BatchHolder is created when the first put request is received.
     
         doSetup(incoming);
       }
     
    +  /**
    +   *  Delayed setup are the parts from setup() that can only be set after actual data
arrives in incoming
    +   *  This data is used to compute the number of partitions.
    +   */
    +  private void delayedSetup() {
    +
    +    // Set the number of partitions from the configuration (raise to a power of two,
if needed)
    +    numPartitions = context.getConfig().getInt(ExecConstants.HASHAGG_NUM_PARTITIONS_KEY);
    +    if ( numPartitions == 1 ) {
    +      canSpill = false;
    +      logger.warn("Spilling was disabled");
    +    }
    +    while (Integer.bitCount(numPartitions) > 1) { // in case not a power of 2
    +      numPartitions++;
    +    }
    --- End diff --
    
    `BaseAllocator.nextPowerOfTwo()`. I've seen other implementations as well. Maybe pick
one and put it in a utilities class somewhere so we don't have to reinvent it multiple times?


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

Mime
View raw message