Todd Lipcon has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/12613 )

Change subject: KUDU-2711 (part 1): add a benchmark for GetTableLocations
......................................................................


Patch Set 2:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/12613/1/src/kudu/integration-tests/table_locations-itest.cc
File src/kudu/integration-tests/table_locations-itest.cc:

http://gerrit.cloudera.org:8080/#/c/12613/1/src/kudu/integration-tests/table_locations-itest.cc@326
PS1, Line 326:   const int kNumSplits = FLAGS_benchmark_num_tablets - 1;
> I'm not sure what Todd's reason was, but I tested using a single messenger
I think I went with multiple whole messengers so that we end up with multiple 
TCP connections. A single TCP connection ends up on a single reactor regardless 
of the number of reactors available. Watching 'top -H' while the benchmark was 
running I saw that one of the reactors was at 100% CPU, meaning we were 
benchmarking the serialization/network code and not realistic to a normal 
workload with many inbound connections



--
To view, visit http://gerrit.cloudera.org:8080/12613
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Ic1643638d82b7697d20ea960592c8b70c30ac7bd
Gerrit-Change-Number: 12613
Gerrit-PatchSet: 2
Gerrit-Owner: Todd Lipcon <[email protected]>
Gerrit-Reviewer: Adar Dembo <[email protected]>
Gerrit-Reviewer: Kudu Jenkins (120)
Gerrit-Reviewer: Tidy Bot (241)
Gerrit-Reviewer: Todd Lipcon <[email protected]>
Gerrit-Reviewer: Will Berkeley <[email protected]>
Gerrit-Comment-Date: Wed, 06 Mar 2019 17:50:51 +0000
Gerrit-HasComments: Yes

Reply via email to