[ 
https://issues.apache.org/jira/browse/HBASE-24936?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17418189#comment-17418189
 ] 

Duo Zhang commented on HBASE-24936:
-----------------------------------

We were listed again by the INFRA team to eat too much space and they mentioned 
nightlies.apache.org again...

This is what I got from the jenkins file of Cassandra team:

{code}
    stage('Summary') {
      steps {
          sh "rm -fR cassandra-builds"
          sh "git clone --depth 1 --single-branch 
https://gitbox.apache.org/repos/asf/cassandra-builds.git";
          sh "./cassandra-builds/build-scripts/cassandra-test-report.sh"
          junit testResults: 
'**/build/test/**/TEST*.xml,**/cqlshlib.xml,**/nosetests.xml', 
testDataPublishers: [[$class: 'StabilityTestDataPublisher']]

          // the following should fail on any installation other than 
ci-cassandra.apache.org
          //  TODO: keep jenkins infrastructure related settings in 
`cassandra_job_dsl_seed.groovy`
          warnError('cannot send notifications') {
              script {
                changes = formatChanges(currentBuild.changeSets)
                echo "changes: ${changes}"
              }
              slackSend channel: '#cassandra-builds', message: ":apache: 
<${env.BUILD_URL}|${currentBuild.fullDisplayName}> completed: 
${currentBuild.result}. 
<https://github.com/apache/cassandra/commit/${env.GIT_COMMIT}|${env.GIT_COMMIT}>\n${changes}"
              emailext to: '[email protected]', subject: "Build 
complete: ${currentBuild.fullDisplayName} [${currentBuild.result}] 
${env.GIT_COMMIT}", presendScript: 
'${FILE,path="cassandra-builds/jenkins-dsl/cassandra_email_presend.groovy"}', 
body: '''
-------------------------------------------------------------------------------
Build ${ENV,var="JOB_NAME"} #${BUILD_NUMBER} ${BUILD_STATUS}
URL: ${BUILD_URL}
-------------------------------------------------------------------------------
Changes:
${CHANGES}
-------------------------------------------------------------------------------
Failed Tests:
${FAILED_TESTS,maxTests=500,showMessage=false,showStack=false}
-------------------------------------------------------------------------------
For complete test report and logs see 
https://nightlies.apache.org/cassandra/${JOB_NAME}/${BUILD_NUMBER}/
'''
          }
          sh "echo \"cassandra-builds at: `git -C cassandra-builds log -1 
--pretty=format:'%h %an %ad %s'`\" > builds.head"
          sh "find . -type f -name \\*.head -exec cat {} \\;"
          sh "xz TESTS-TestSuites.xml"
          sh "echo \"For test report and logs see 
https://nightlies.apache.org/cassandra/${JOB_NAME}/${BUILD_NUMBER}/\"";
      }
      post {
          always {
              sshPublisher(publishers: [sshPublisherDesc(configName: 
'Nightlies', transfers: [sshTransfer(remoteDirectory: 
'cassandra/${JOB_NAME}/${BUILD_NUMBER}/', sourceFiles: 
'TESTS-TestSuites.xml.xz')])])
          }
      }
    }
{code}

Let's see if we could use these scripts too...

> review Jenkins build artifacts
> ------------------------------
>
>                 Key: HBASE-24936
>                 URL: https://issues.apache.org/jira/browse/HBASE-24936
>             Project: HBase
>          Issue Type: Task
>            Reporter: Sean Busbey
>            Assignee: Sean Busbey
>            Priority: Critical
>         Attachments: usage-hbase-find-flaky.zip, usage-hbase-full.zip
>
>
> Post move to the ci-hadoop build servers we are now the biggest user of 
> space. That is not a problem in and of itself, but the master node has run 
> out of disk space twice now. As of this snapshot we are using 125GB of 
> storage and the next largest project is only using 20GB.
> https://paste.apache.org/kyrds
> We should review our builds for any issues and come up with expectations for 
> what our steady-state disk usage should look like
> * we are supposed to compress any test logs (usually this gets us 90-99% 
> space savings)
> * we are supposed to clean up workspaces when jobs are done
> * we are supposed to keep a fixed window of prior builds (either by days or 
> number of runs)
> If all of our jobs are currently following these guidelines, another 
> possibility is to push the artifacts we need over to 
> [nightlies.a.o|https://nightlies.apache.org/authoring.html]. Barring that, we 
> should formally request asf infra set up [a plugin for storing artifact on 
> s3|https://plugins.jenkins.io/artifact-manager-s3/].



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to