Dan Hecht has posted comments on this change.

Change subject: IMPALA-3201: reservation implementation for new buffer pool
......................................................................


Patch Set 10:

(4 comments)

http://gerrit.cloudera.org:8080/#/c/3993/10/be/src/bufferpool/reservation-tracker.cc
File be/src/bufferpool/reservation-tracker.cc:

PS10, Line 100: MemLimitExceeded
> I was deferring it until I can solve it more holistically in the context of
Having the caller directly use IncreaseReservation() to acquire its initial 
reservation is good with me.  Fragment-instances/exec-nodes still have a 
concept of initial reservation but I agree, no need to carry it all the way 
down to this level in the code.


http://gerrit.cloudera.org:8080/#/c/3993/9/be/src/bufferpool/reservation-tracker.h
File be/src/bufferpool/reservation-tracker.h:

Line 121:   bool IncreaseReservation(int64_t bytes);
> It will be useful in setting up initial reservations: we can request the en
Adding the TODO and revisiting later is okay with me.


http://gerrit.cloudera.org:8080/#/c/3993/10/be/src/bufferpool/reservation-tracker.h
File be/src/bufferpool/reservation-tracker.h:

Line 178:       int64_t bytes, bool use_existing_reservation, bool 
is_child_reservation);
> Removed IncreaseReservationInternal() by just having all callsites acquire 
yeah, didn't suggest getting rid of the unlocked one cause now the child 
acquires the parent lock which seems a little messy.  But I don't feel too 
strongly about that.


http://gerrit.cloudera.org:8080/#/c/3993/11/be/src/bufferpool/reservation-tracker.h
File be/src/bufferpool/reservation-tracker.h:

Line 120:   /// made to the state of any ReservationTrackers.
formatting glitch


-- 
To view, visit http://gerrit.cloudera.org:8080/3993
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I35cc89e863efb4cc506657bfdaaaf633a10bbab6
Gerrit-PatchSet: 10
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Tim Armstrong <[email protected]>
Gerrit-Reviewer: Dan Hecht <[email protected]>
Gerrit-Reviewer: Marcel Kornacker <[email protected]>
Gerrit-Reviewer: Michael Ho
Gerrit-Reviewer: Tim Armstrong <[email protected]>
Gerrit-HasComments: Yes

Reply via email to