Marcel Kornacker has posted comments on this change.

Change subject: IMPALA-4348 / IMPALA-4333: Improve coordinator fragment 
cancellation
......................................................................


Patch Set 1:

(6 comments)

http://gerrit.cloudera.org:8080/#/c/4865/1/be/src/exec/plan-root-sink.cc
File be/src/exec/plan-root-sink.cc:

Line 156:   if (consumer_done_) return state->CheckQueryState();
if someone called CloseConsumer(), does it even make sense to keep executing 
this function? shouldn't you just return eos at this point?


Line 163:     consumer_cv_.wait(l);
> Agree it's not needed for correctness. consumer_cv_wait_ can wake up spurio
make that part of the declared interface (if you call CloseConsumer() then 
GetNext() will immediately exit with eos).


http://gerrit.cloudera.org:8080/#/c/4865/1/be/src/runtime/coordinator.cc
File be/src/runtime/coordinator.cc:

Line 1550
leave a comment in place explaining why the cancel call is done despite a 
possible error status


Line 166:   void SetInitialStatus(const Status& status, bool rpc_sent = true) {
remove default.


Line 506:   // If there was an error, but the coordinator fragment was 
successfully started, it
since they're all started at the same time now, why is the coordinator fragment 
still special?


Line 522:     executor_ = root_fragment_instance_->executor();
get rid of executor_ and root_sink_ and add executor() and root_sink() that go 
through root_fragment_instance_, to make clear that those are only valid as 
long as root_fragment_instance_ is valid.


-- 
To view, visit http://gerrit.cloudera.org:8080/4865
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I5b4e25c1d658b3929182ba5e56b5c5e881dd394a
Gerrit-PatchSet: 1
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Henry Robinson <[email protected]>
Gerrit-Reviewer: Dan Hecht <[email protected]>
Gerrit-Reviewer: Henry Robinson <[email protected]>
Gerrit-Reviewer: Marcel Kornacker <[email protected]>
Gerrit-Reviewer: Tim Armstrong <[email protected]>
Gerrit-HasComments: Yes

Reply via email to