Surya Hebbar has posted comments on this change. ( http://gerrit.cloudera.org:8080/21683 )
Change subject: IMPALA-13304: Include aggregate instance-level metrics in JSON profile ...................................................................... Patch Set 18: (9 comments) http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile-test.cc File be/src/util/runtime-profile-test.cc: http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile-test.cc@1961 PS17, Line 1961: std::uniform_int_distribution<> uni_di > VLOG(1) or cout the randomized NUM_PROFILES and BUCKET_SIZE. Done http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile-test.cc@1967 PS17, Line 1967: ze) { > greater than or equal to? uniform_int_distribution range is inclusive, righ Done. Mentioned equal. http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile-test.cc@1970 PS17, Line 1970: } else { : // Generate a bucket size or equal to number of instances : std::uniform_int_distribution<> uni_dis_granularity(NUM_PROFILES > I don't really like this. The randomization may never pick BUCKET_SIZE == N Done. I have added a third parameter. http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile-test.cc@2015 PS17, Line 2015: _event_probability(g > Go ahead and fix it in next patch set. Done http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile-test.cc@2015 PS17, Line 2015: _event_probability(g > This should have been 1ms. Do not want to trigger another unnecessary build Done http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile-test.cc@2251 PS17, Line 2251: > NonAggregatedCase? I thought "Grouped" would be more meaningful than NonAggregated. http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile-test.cc@2253 PS17, Line 2253: arDown is implicitly > Should this be "bucket_size >= number"? Done. http://gerrit.cloudera.org:8080/#/c/21683/14/be/src/util/runtime-profile.cc File be/src/util/runtime-profile.cc: http://gerrit.cloudera.org:8080/#/c/21683/14/be/src/util/runtime-profile.cc@66 PS14, Line 66: json_profile_event_timestamp_limit, 5, > Query option might not fit well for this. Imagine that I ran a query using Thank you for the suggestion, it would certainly be more helpful to use parsed args. http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile.cc File be/src/util/runtime-profile.cc: http://gerrit.cloudera.org:8080/#/c/21683/17/be/src/util/runtime-profile.cc@67 PS17, Line 67: "Sets the number of spans / buckets for grouping of event timestamps within" : " the aggregated JSON profile. For example, if the number of fragment" : " instances(N) reporting an event is more than(N > M) the set > Description can be better here. Example can also help. How about: The aggregated profile was always being included in the traditional JSON profile irrespective of the '--gen_experimental_profile'. This was told to be the expected behavior. Also, when M=0 or M>=N, grouping of event sequence timestamps for each event is still done, but no aggregates are calculated. This is required to support query timeline display for experimental profile. Accordingly, I have added a readable yet accurate description with an example now. -- To view, visit http://gerrit.cloudera.org:8080/21683 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I49e18a7a7e1288e3e674e15b6fc86aad60a08214 Gerrit-Change-Number: 21683 Gerrit-PatchSet: 18 Gerrit-Owner: Surya Hebbar <[email protected]> Gerrit-Reviewer: Impala Public Jenkins <[email protected]> Gerrit-Reviewer: Kurt Deschler <[email protected]> Gerrit-Reviewer: Riza Suminto <[email protected]> Gerrit-Reviewer: Surya Hebbar <[email protected]> Gerrit-Comment-Date: Wed, 29 Jan 2025 08:55:19 +0000 Gerrit-HasComments: Yes
