Tim Armstrong has posted comments on this change. ( http://gerrit.cloudera.org:8080/10977 )
Change subject: IMPALA-7296: bytes limit for row batch queue ...................................................................... Patch Set 5: Mainly to avoid potentially making things worse for narrow rows. I'm open to changing it but this was the easiest way I could convince myself that it wouldn't make anything worse. There are some weird cases like count(*) where we may end up with row batches that have very little memory - just the tuple_ptrs_ which aren't counted in the byte size. I did some rough calculations and in those cases, even if we include the tuple_ptrs_ size, we could potentially end up with 1000s of batches in the queue, which would mean more untracked memory. An alternative is to add some constant number of bytes per row batch for overhead. -- To view, visit http://gerrit.cloudera.org:8080/10977 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Iaa06d1d8da2a6d101efda08f620c0bf84a71e681 Gerrit-Change-Number: 10977 Gerrit-PatchSet: 5 Gerrit-Owner: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Reviewer: Michael Ho <k...@cloudera.com> Gerrit-Reviewer: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Comment-Date: Fri, 20 Jul 2018 18:22:13 +0000 Gerrit-HasComments: No