Dan Hecht has posted comments on this change. ( http://gerrit.cloudera.org:8080/9155 )
Change subject: IMPALA-6450: fix EventSequence::Start() ...................................................................... Patch Set 2: (1 comment) http://gerrit.cloudera.org:8080/#/c/9155/2/be/src/util/runtime-profile-counters.h File be/src/util/runtime-profile-counters.h: http://gerrit.cloudera.org:8080/#/c/9155/2/be/src/util/runtime-profile-counters.h@304 PS2, Line 304: offset_ = MonotonicStopWatch::Now() - start_time_ns; : // TODO: IMPALA-4631: Occasionally we see MonotonicStopWatch::Now() return : // (start_time_ns - 1), even though 'start_time_ns' was obtained using : // MonotonicStopWatch::Now(). : DCHECK_GE(offset_, -1); : sw_.Start(); Not for this change, but I had to dig around to understand what this offset_ was really about (and look at the change that added it). Lars, I wonder if this would be clearer if either: a) We expose a sw_.StartTime(), and then offset can be computed from that, to make it more obvious that we're trying to retroactively start sw_. b) We make a way to set the start_ time in sw_ directly (i.e. pass start_time_ns to a new method sw_.StartSynchronized() that starts the clock using the given time, or something). And then we don't need this offset at at. -- To view, visit http://gerrit.cloudera.org:8080/9155 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I62317149cb8428f7d29e945a538f6cc8dd45342f Gerrit-Change-Number: 9155 Gerrit-PatchSet: 2 Gerrit-Owner: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Reviewer: Dan Hecht <dhe...@cloudera.com> Gerrit-Reviewer: Lars Volker <l...@cloudera.com> Gerrit-Reviewer: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Reviewer: anujphadke <apha...@cloudera.com> Gerrit-Comment-Date: Tue, 30 Jan 2018 17:37:42 +0000 Gerrit-HasComments: Yes