spark-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From mateiz <>
Subject [GitHub] spark pull request: SPARK-1145: Memory mapping with many small blo...
Date Tue, 04 Mar 2014 05:38:29 GMT
Github user mateiz commented on a diff in the pull request:
    --- Diff: core/src/main/scala/org/apache/spark/storage/BlockFetcherIterator.scala ---
    @@ -146,6 +146,12 @@ object BlockFetcherIterator {
         protected def splitLocalRemoteBlocks(): ArrayBuffer[FetchRequest] = {
    +      // Make remote requests at most maxBytesInFlight / 5 in length; the reason to keep
    +      // smaller than maxBytesInFlight is to allow multiple, parallel fetches from up
to 5
    +      // nodes, rather than blocking on reading output from one node.
    +      val maxRequestSize = math.max(maxBytesInFlight / 5, 1L)
    --- End diff --
    Yeah I agree, note that the if statement below is `if (curRequestSize >= maxRequestSize)`.
In general you might have some requests bigger than this of one map's output is just super
large. This threshold is just used to batch together small map outputs so we can grab more
than one at once.

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 or file a JIRA ticket
with INFRA.

View raw message