This is an automated email from the ASF dual-hosted git repository. dkulp pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/avro.git
commit 008b19051b0de6b81a57b93442b341916a6acf8a Author: rstata <[email protected]> AuthorDate: Sun Nov 25 16:06:20 2018 -0800 Added build.sh flag to pass extra docker-run args, updated perf-doc to explain how to use. --- build.sh | 12 ++++++-- doc/src/content/htmldocs/performance-testing.html | 34 +++++++++++++++++++++++ 2 files changed, 44 insertions(+), 2 deletions(-) diff --git a/build.sh b/build.sh index 0dc4788..10544df 100755 --- a/build.sh +++ b/build.sh @@ -20,9 +20,10 @@ set -e # exit on error cd `dirname "$0"` # connect to root VERSION=`cat share/VERSION.txt` +DOCKER_XTRA_ARGS="" function usage { - echo "Usage: $0 {test|dist|sign|clean|docker|rat|githooks|docker-test}" + echo "Usage: $0 {test|dist|sign|clean|docker [--args \"docker-args\"]|rat|githooks|docker-test}" exit 1 } @@ -33,8 +34,10 @@ fi set -x # echo commands -for target in "$@" +while (( "$#" )) do + target="$1" + shift case "$target" in test) @@ -200,6 +203,10 @@ do ;; docker) + if [[ $1 =~ ^--args ]]; then + DOCKER_XTRA_ARGS=$2 + shift 2 + fi docker build -t avro-build -f share/docker/Dockerfile . if [ "$(uname -s)" == "Linux" ]; then USER_NAME=${SUDO_USER:=$USER} @@ -226,6 +233,7 @@ UserSpecificDocker -v ${HOME}/.m2:/home/${USER_NAME}/.m2 \ -v ${HOME}/.gnupg:/home/${USER_NAME}/.gnupg \ -u ${USER_NAME} \ + ${DOCKER_XTRA_ARGS} \ avro-build-${USER_NAME} bash ;; diff --git a/doc/src/content/htmldocs/performance-testing.html b/doc/src/content/htmldocs/performance-testing.html index f01c36a..5cd8026 100644 --- a/doc/src/content/htmldocs/performance-testing.html +++ b/doc/src/content/htmldocs/performance-testing.html @@ -81,10 +81,44 @@ As mentioned in the introduction, we tried a number of different mechanisms to r <p> <li> Modified the code slightly, for example: starting the timer of a cycle after, rather than before, encoders or decoders are constructed; cacheing encoders and decoders; and reusing record objects during read tests rather than construct new ones for each record being read. +<p> <li> Using Docker's <code>--cpuset-cpus</code> flag to force the tests onto a single core. + <p> <li> Using a dedicated EC2 instance (<code>c5d.2xlarge</code>). </ul> Of the above, the only change that made a significant difference was the last: in going from a laptop and desktop computer to a dedicated EC2 instances, we went from over 70 tests (out of 200) with a variance of 5% or more between runs to 35. As mentioned in the introduction, we should switch to a framework like <a href="https://java-performance.info/jmh/";>JMH</a> to attack this problem more fundamentally. +<p> If you want to setup your own EC2 instance for testing, here's how we did it. We launched a dedicated EC2 <code>c5d.2xlarge</code> instance from the AWS console, using the "Amazon Linux 64-bit HVM GP2" AMI. We logged into this instance and ran the following commands to install Docker and Git (we did all our Avro build and testing inside the Docker image): +<pre> + sudo yum update + sudo yum install -y git-all + git config --global user.name "Your Name" + git config --global user.email [email protected] + git config --global core.editor emacs + sudo install -y docker + sudo usermod -aG docker ec2-user ## Need to log back in for this to take effect + sudo service docker start +</pre> +At this point you can checkout Avro and launch your Docker container: +<pre> + git clone https://github.com/apache/avro.git + cd avro + ./build.sh docker --args "--cpuset-cpus 2,6" +</pre> +The <code>--args</code> flag in the last command deserves some explanation. In general, the <code>--args</code> allows you to pass additional arguments to the <code>docker run</code> command executed inside <code>build.sh</code>. In this case, the <code>--cpuset-cpus</code> flag for <code>docker</code> tells docker to schedule the contianer exclusively on the listed (virtual) CPUs. We identified vCPUs 2 and 6 using the <code>lscpu</code> Linux command: +<pre> + [ec2-user@ip-0-0-0-0 avro]$ lscpu --extended + CPU NODE SOCKET CORE L1d:L1i:L2:L3 ONLINE + 0 0 0 0 0:0:0:0 yes + 1 0 0 1 1:1:1:0 yes + 2 0 0 2 2:2:2:0 yes + 3 0 0 3 3:3:3:0 yes + 4 0 0 0 0:0:0:0 yes + 5 0 0 1 1:1:1:0 yes + 6 0 0 2 2:2:2:0 yes + 7 0 0 3 3:3:3:0 yes +</pre> +Notice that (v)CPUs 2 and 6 are both on core 2: it's sufficient to schedule the container on the same core, vs a single vCPU. + <h1>Appendix A: Sample uses of run-perf.sh</h1>
