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 #1023: DRILL-5922 Fixed Child Allocator Leak. DRILL-5926 ...
Date Fri, 10 Nov 2017 20:59:09 GMT
Github user paul-rogers commented on a diff in the pull request:

    https://github.com/apache/drill/pull/1023#discussion_r150336286
  
    --- Diff: pom.xml ---
    @@ -442,7 +442,7 @@
                   -Dorg.apache.drill.exec.server.Drillbit.system_options="org.apache.drill.exec.compile.ClassTransformer.scalar_replacement=on"
                   -Ddrill.test.query.printing.silent=true
                   -Ddrill.catastrophic_to_standard_out=true
    -              -XX:MaxPermSize=512M -XX:MaxDirectMemorySize=3072M
    +              -XX:MaxPermSize=512M -XX:MaxDirectMemorySize=4096M
    --- End diff --
    
    This is a fix for a test that we don't really even need. We are attempting to verify that
we can allocate a single value vector of 2 GB by doubling from a small amount. On the last
allocation, we have a 1 GB vector doubling to 2 GB, so we temporarily need 3 GB.
    
    But, vectors should never get this large. The batch size limitation project is doing work
to limit vectors to the 16 MB Netty slab size.
    
    For this reason, this is a workaround to a test for a feature that is actually a bug which
we will be fixing in a later release.


---

Mime
View raw message