qaprosoft / qps-pipeline Goto Github PK

runCron method should scan all sub-projects inside the repo

in order to unify way of cloning and get rid of code duplication we should pass parameters dynamically

Subj

setupGoalsForiOS and setupGoalsForAndroid

http://demo.qaprosoft.com/zafira-ws/swagger-ui.html#!/test-runs-api-controller/sendTestRunResultsEmail

all kind of integrations between carina and zafira for emailing should be removed

To reproduce:

1. select job
2. click 'build with parameters'
3. click at 'retry count' parameter
   Actual: in appeared drop down list '0' is mentioned twice

To be able to generate unified runner/scanner for the repo it would be nice to have single Jenkinsfile with ability to detect is it pr checker or merge/push startup and execute appropriate logic...

for each run browser value is empty string and it is ignored on child job level somehow

...to avoid following errors:

org.openqa.selenium.WebDriverException: cannot create session QPS-HUB-* on any hosts after 1 attempt(s) (WARNING: The server did not provide any stacktrace information)
Command duration or timeout: 300.00 seconds
Build info: version: '3.12.0', revision: '7c6e0b3', time: '2018-05-08T14:04:26.12Z'
System info: host: '00953bdc2828', ip: '172.18.0.28', os.name: 'Linux', os.arch: 'amd64', os.version: '4.15.0-24-generic', java.version: '1.8.0_151'
Driver info: driver.version: AndroidDriver

com.qaprosoft.carina.core.foundation.webdriver.DriverPool.createDriver(DriverPool.java:370)
com.qaprosoft.carina.core.foundation.webdriver.DriverPool.getDriver(DriverPool.java:146)
com.qaprosoft.carina.core.foundation.webdriver.DriverPool.getDriver(DriverPool.java:98)
com.qaprosoft.carina.core.foundation.webdriver.DriverPool.getDriver(DriverPool.java:67)
com.qaprosoft.carina.core.foundation.utils.factory.CustomTypePageFactory.initPage(CustomTypePageFactory.java:59)

To be able to manage extra actions for cron jobs it is supposed not to re-create them during scanner.
It will allow to adjust those items manually including:

• custom cron trigger
• custom project specific parameters

Currently we get default parameter build=.* for all jobs. We don't need this parameter.

Or figure out how to configure them manually. For now only DefaultPool or Any is working.

Every Groovy class initially is either of jobDSL or pipeline type and they both have their own characteristics, so it seems reasonable to store them accordingly in separate directories

In additions to default set of statndard CI parameters we have to provide possibility to declare and create custom arguments: hidden or visible. Using overrideFields declaration is not enough

for now we write into the log and in UI zafira general host like "demo.qaprosoft.com" and port 8000.
this functionality should be improved in such way:

1. pipeline Runner should detect current ip address

2. this ip and port should be written in log

3. wrap this functionality into separate protected method to be able to override on children level

For the case when cron pipeline job has browser parameter declared we should limit children job startup byt his browser only.
If no param["browser"] declared then we should start regression for all supported browsers based on

1. add new capability for mobile devices:
   emulated = true [false]

2. adjust selenium-hub to be able to track new capability (by default is should be supported without any changes)

3. adjust pipeline Runner for iOS mobile to detect build artifact extension.
   if it is ".ipa" then add capabilities.emulated=false
   if it is ".app" then capabilities.emulated=true
   also pay attantion for the zip like "...app.zip"

Currently we are able to override retry_count parameter only using override fields. This flow doesn't reflect actual value while starting a job.
We need some flow to reflect a correct value in job's extensible choice parameter.

Current logic for generating local eTAF_report into dedicated location and replacing it by zafira has a lot of custom logic.
We should simplify this process a lot and incorporate into the pipeline runner steps for getting zafira results based on id/ci_run_id

Old logic to find and publish local report could be saved as well but using single findFiles logic and publish with different name: for example: eTAF_Local_Report

let's enable both enableVNC and enableVideo by default where possible:
enbleVNC:

• web
• android

enableVideo
-web
-android
-iOS (still there is a problem in Appium...)

we should support debug mode in order not to comment/uncomment console outputs

https://nerdwallet.qaprosoft.com/jenkins/job/Automation/job/Android_goal_tests/13/

The given job was started from Jenkins with parameter: thread_count = 3
but 4 phones were occupied.

1. add new boolean field: onlyUpdated - true by default
2. modify ignoreExisting functionality removing dependency onto the isUpdated(filter)
3. if we execute job with onlyUpdated = true then
   a) skip execution when no suites updated (as previous)
   b) during registerObject call for job/cron do a verification that's appropriate suite was updated in latest commits

for now browser_version was removed completely from job arguments.
I suggest:

1. return it back as separate field (maybe hidden for the beginning
2. adjust jobdsl for web use-case to insert appropriate hiddent fields
3. improve cron parser supporting new notation with browser_version, i.e.:
   old one:
   
   new one
   

-> So if browsername contain space in jenkinsPipelibebrowsers then it is supposed to be a version. So we have to parse it correctly and generate 5 jobs during startup for:
-chrome *

• firefox *
• ie 9.0
  -ie *
  -edge 16.0

We use hardcoded URL in the scripts which cause automation fails. File for the reference : https://github.com/qaprosoft/qps-pipeline/blob/master/src/com/qaprosoft/jenkins/repository/jobdsl/Job.groovy
line 281

Some workaround 9dacd54

Pipeline jobs should have such params:
priority - 5 as default and global choice list
env - empty by default as string param (think about global choice usage if possible)
browser - TDB. maybe empty again as default value

The param already exists in Configurator as ci_parent_build. Rename upstream_job_id as well.

for now all scanner jobs are started when push/merge to central master happen. Investigate what we can do with creating personal Automation jobs basde on forked repo and scanner functionality

Summary:
After setting of android pipeline job on specific environment (e.g. MASTER_BETA, QA_REMOTE_PROD) all specific jobs (e.g. SANITY, SMOKE etc.) are triggered on 2 envs: 1) which was determined in pipeline job 2) general env (INT | BETA | PROD)

take a look to commented piece of code
https://github.com/qaprosoft/qps-pipeline/blob/a8b7313a1fb5e81cd232cdea1374fe6c9dbbd62a/src/com/qaprosoft/selenium/grid/ProxyInfo.
groovy#L13

Sometimes we test/debug new features on CI and update job manually adding extra arguments, changing existing one etc.
But as only push happens scanner recreate this job forcibly and we have to declare new fields again.

It would be nice to have possibility to avoid job re-creation via scanner.
Proposed solution:

1. reuse description field for that putting into it some keyword like:
   DO NOT RECREATE, SCANNER-OUT etc :)
2. scanner job before starting re-creation should just analyze description content and do not force re-creation until this description exists
3. but to be informed that some jobs are not in sync with latest jobdsl/pipeline it is recommended t inform suite owners and swat admin team about it

need adjust process and provide valid git args:
ci.setGitBranch(config.getString("git_branch", null));
ci.setGitCommit(config.getString("git_commit", null));
ci.setGitUrl(config.getString("git_url", null));

Subj

by default we configure scanner to start on each push/merge. It would be nice to analyze changed files and do not recreate jobs if xml were not updated

QA jobs should be forcibly stopped if jenkins is being restarted. So seems like there is an appropriate checkbox in one of the latest Jenkns build.
Please update jobdsl for test job only to make it enabled by default

1. remove com.qaprosoft.jenkins.repository package and move sub packages up
2. remove com.qaprosoft.jenkins.repository.pipeline.v2 as we agreed do not support several versions of pipelines in the single bundle

Details:

1. within thw suite I want to run following parameter is set:
   

2. following pipeline job is created on Jenkins:
   https://ua.qaprosoft.com/jenkins/job/Pipeline/job/MMF-Andoid-Minimal-Acceptance-Regression-Pipeline/

in wich following checkbox is checked: Build periodically
and the schedule is set according to the same parameter from within suite:

3. there're no any errors in pipeline job or PipelineDSL job
   https://ua.qaprosoft.com/jenkins/job/Pipeline/job/MMF-Andoid-Minimal-Acceptance-Regression-Pipeline/65/console

https://ua.qaprosoft.com/jenkins/job/Management_Jobs/job/PipelineDSL/39/consoleFull

And apply OO approach for this script