[JIRA] (JENKINS-57990) Avoid filtering contents without sensible data and do per-content anonymization

[manuelramonleonjimenez@gmail.com (JIRA)]

Ramon Leon created an issue

Jenkins / JENKINS-57990 Avoid filtering contents without sensible data and do per-content anonymization Issue Type: Bug Assignee: Ramon Leon Components: support-core-plugin Created: 2019-06-12 14:34 Priority: Minor Reporter: Ramon Leon Change the behavior of the Contents to avoid filtering when it's not needed. And change the way it's anonymized by delegating it to the component to avoid losing the structure of the information. Should be filtered NodesContent: include labels and filesystem path AboutUser AgentsConfigFile (secrets encrypted but maybe other sensitive info) BuildQueue: name of the job ConfigFileComponent (secrets encrypted but maybe other sensitive info) EnvironmentVariables JenkinsLogs JVMProcessSystemMetricsContent (environment variables) LoggerManager NetworInterfaces (ips) NodeMonitors (computer name) OtherConfigFilesComponent RoocCAs SlaveLaunchLogs SlaveLogs SystemConfiguration Agents and Master (user running) UpdateCenter (proxy information: urls) SystemProperties Shouldn’t be filtered AboutContent AboutBrowser ActivePlugins DisabledPlugins FailedPlugins Dockerfile MasterChecksumsContent NodeChecksumsContent AdministrativeMonitors DeadlockRequestComponent DumpExportTable FileDescriptorLimit GCLogs Out of scope, it needs some refactor of classes. To tackle on another ticket HeapUsageHistogram ItemsContent LoadStats Metrics RemotingDiagnostics ReverseProxy SlaveCommandStatistics SlowRequestComponent ThreadDumps Avoid anonymizing the labels, the structure of the files under the control of the plugin (.md files, and information files). Analyze what values should be anonymized. The contents generated on the fly are going to extend the PrefilteredContent by creating a new PrefilteredPrintedContent and they are going to filter the information by themselves. The ones that are being filtered and shouldn't to will override the shouldBeFiltered method to avoid that. Acceptance criteria The support-core plugin is changed by creating a new PrefilteredPrintedContent that extends PrefilteredContent The contents that don't need to be filtered don't do that by overriding shouldBeFiltered (false) The contents that need to be filtered extends from a PrefilteredContent, most likely PrefilteredPrintedContent and do the filtering by themselves implementing the new writeTo method. Look at: https://github.com/jenkinsci/support-core-plugin/blob/e2d79249818d41111ea96979a710ac3404018f78/src/main/java/com/cloudbees/jenkins/support/api/StringContent.java#L58 avoiding to filter the structure of the markdown or whatever is going to affect the comprehension of the content. Some tests are implemented for: A content that was filtered and now it's not A content that was filtered affecting the structure (markdown) and now it's not affected The PR in support-core is merged Add Comment

This message was sent by Atlassian Jira (v7.11.2#711002-sha1:fdc329d)

[manuelramonleonjimenez@gmail.com (JIRA)]

Ramon Leon started work on JENKINS-57990

Change By: Ramon Leon Status: Open In Progress

Add Comment

[manuelramonleonjimenez@gmail.com (JIRA)]

Ramon Leon updated JENKINS-57990

Jenkins / JENKINS-57990 Avoid filtering contents without sensible data and do per-content anonymization

Change By: Ramon Leon Status: In Progress Review

Add Comment

[manuelramonleonjimenez@gmail.com (JIRA)]

Ramon Leon commented on JENKINS-57990

Re: Avoid filtering contents without sensible data and do per-content anonymization PR: https://github.com/jenkinsci/support-core-plugin/pull/174/files

Add Comment

[manuelramonleonjimenez@gmail.com (JIRA)]

Ramon Leon updated an issue

Jenkins / JENKINS-57990

Avoid filtering contents without sensible data and do per-content anonymization

Change By: Ramon Leon

Change the behavior of the {{Contents}} to avoid filtering when it's not needed. And change the way it's anonymized by delegating it to the component to avoid losing the structure of the information.

*Should be filtered* * NodesContent: include labels and filesystem path * AboutUser * AgentsConfigFile (secrets encrypted but maybe other sensitive info) * BuildQueue: name of the job * ConfigFileComponent (secrets encrypted but maybe other sensitive info) * EnvironmentVariables * JenkinsLogs * JVMProcessSystemMetricsContent (environment variables) * LoggerManager * NetworInterfaces (ips) * NodeMonitors (computer name) * OtherConfigFilesComponent * RoocCAs * SlaveLaunchLogs * SlaveLogs * SystemConfiguration Agents and Master (user running) * UpdateCenter (proxy information: urls) * SystemProperties *Shouldn’t be filtered* * AboutContent * AboutBrowser * ActivePlugins * DisabledPlugins * FailedPlugins * Dockerfile * MasterChecksumsContent * NodeChecksumsContent * AdministrativeMonitors * DeadlockRequestComponent * DumpExportTable * FileDescriptorLimit * GCLogs - Out of scope, it needs some refactor of classes. To tackle on another ticket - Done because i needed the infra for other components. * HeapUsageHistogram * ItemsContent * LoadStats * Metrics * RemotingDiagnostics * ReverseProxy * SlaveCommandStatistics * SlowRequestComponent * ThreadDumps

Avoid anonymizing the labels, the structure of the files under the control of the plugin (.md files, and information files). Analyze what values should be anonymized.

The contents generated on the fly are going to extend the PrefilteredContent by creating a new *PrefilteredPrintedContent* and they are going to filter the information by themselves.

The ones that are being filtered and shouldn't to will override the {{shouldBeFiltered}} method to avoid that.

h3. Acceptance criteria * The support-core plugin is changed by creating a new {{PrefilteredPrintedContent}} that extends {{PrefilteredContent}} * The contents that don't need to be filtered don't do that by overriding {{shouldBeFiltered}} (false) * The contents that need to be filtered extends from a {{PrefilteredContent}}, most likely {{PrefilteredPrintedContent}} and do the filtering by themselves implementing the new {{writeTo}} method. Look at: [https://github.com/jenkinsci/support-core-plugin/blob/e2d79249818d41111ea96979a710ac3404018f78/src/main/java/com/cloudbees/jenkins/support/api/StringContent.java#L58] avoiding to filter the structure of the markdown or whatever is going to affect the comprehension of the content. * Some tests are implemented for: * * * A content that was filtered and now it's not * * * A content that was filtered affecting the structure (markdown) and now it's not affected * The PR in support-core is merged

Add Comment

[hchen@marqeta.com (JIRA)]

Henry Chen commented on JENKINS-57990

Re: Avoid filtering contents without sensible data and do per-content anonymization I've recently (last week) been seeing my console data being masked as "FILTERED". Is this related to the bug/issue? {:timestamp=>15616234449694, :message=>"[FILTERED]", :ingestion_time=>15616234449694} should see: {:timestamp=>15616234449694, :message=>"[ACTUAL DATA]", :ingestion_time=>15616234449694}

Add Comment

[hchen@marqeta.com (JIRA)]

Henry Chen edited a comment on JENKINS-57990

Re: Avoid filtering contents without sensible data and do per-content anonymization

I've recently (last week) been seeing my console data being masked as "FILTERED". Is this related to the bug/issue?

Jenkins console displays:

{:timestamp=>15616234449694, :message=>"[FILTERED]", :ingestion_time=>15616234449694}

Local displays ( should see ) :

{:timestamp=>15616234449694, :message=>"[ACTUAL DATA]", :ingestion_time=>15616234449694}

Add Comment

[manuelramonleonjimenez@gmail.com (JIRA)]

Ramon Leon commented on JENKINS-57990

Re: Avoid filtering contents without sensible data and do per-content anonymization

I don't think so Henry Chen, it's only related to the logs included in a zip generated by the support-core-plugin. A plugin to create zips with information about the instance to help to diagnose problems on it.

Add Comment

[hchen@marqeta.com (JIRA)]

Henry Chen commented on JENKINS-57990

Re: Avoid filtering contents without sensible data and do per-content anonymization

Thank you Ramon Leon, turns out to be an ruby version issue on my end.

Add Comment

[manuelramonleonjimenez@gmail.com (JIRA)]

Ramon Leon commented on JENKINS-57990

Re: Avoid filtering contents without sensible data and do per-content anonymization

Add Comment

[manuelramonleonjimenez@gmail.com (JIRA)]

Ramon Leon updated JENKINS-57990

Jenkins / JENKINS-57990

Avoid filtering contents without sensible data and do per-content anonymization

Change By: Ramon Leon Status: In Review Resolved Resolution: Done Released As: 2.57

Add Comment