Come non "sparare e dimenticare" quando si fa un trigger remoto di una build Jenkins?

Sto tentando di innescare una build Jenkins con parametri da Bamboo, facendo quanto segue:

POST - http://jenkins-url.com/job/jobname/buildWithParameters?ENVIRONMENT=dev&APPLICATION=hello-world

Ma otterrò immediatamente una 201 che mi dice che la build è stata creata. Come faccio ad attendere questa richiesta e a restituire lo stato di successo della build invece del fuoco e dimenticare?

Apparentemente è possibile come da Parameterized-Remote-Trigger-Plugin :

Modifica: creato alla fine se ne hai bisogno. https://github.com/owenmorgan/jenkins-remote-builder

jenkins jenkins-plugins bamboo

— osmorgan
fonte

Quando si imposta il trigger remoto, è necessario fare clic su "Blocca fino a quando i progetti con trigger remoto non completano le loro build".

— Travis Thompson
fonte

grazie, potrei avere la parte sbagliata dello stick .. che plug-in sembra essere per innescare build remote da jenkins stesso? Nel mio scenario, la build viene attivata da una richiesta di Bamboo, ma vorremmo che la richiesta rimanesse in attesa fino a quando la build non viene completata e restituisce lo stato della build.

— Osmorgan,

Ah, adesso capisco. Sfortunatamente non penso che l'API funzioni in questo modo, se osservi come viene sviluppato quel plug-in per Jenkins vedrai che innesca il lavoro remoto e quindi interroga lo stato fino al termine.

— Travis Thompson,

Grazie, ho finito per realizzare questo github.com/owenmorgan/jenkins-remote-builder

— osmorgan

@osmorgan: dovresti descriverlo in una risposta (con un po 'di riepilogo, non solo il link) e accettarlo.

— Dan Cornilescu,

sheesh .. è nella domanda .. e c'è una descrizione nel link. ci vediamo su StackOverflow.

— Osmorgan,

Ho creato lo script jenkins-remote-builder che seguirà il tuo build remoto fino al completamento.

Alcuni maggiori dettagli al riguardo (dal suo README.md ):

Esempio

jenkins=https://user:pass@jenkins.mydomain.com:8080
jenkins_job=MyApp-Deploy

environment=dev
application=myapp
revision=9fd71f63b351b8208264daf86d292ced580a2f60

./jenkins_remote_trigger.sh \
            -h ${jenkins} \
            -j ${jenkins_job} \
            -p "ENVIRONMENT=${environment}&APPLICATION=${application}&REVISION=${revision}"

Uso:

-h HOST     | --host=HOST                       Jenkins host
-j JOBNAME  | --jobname=test-build-job          The name of the jenkins job to trigger
-p JOBPARAM | --jobparam=environment=uat&test=1 Jenkins job paramiters
-q          | --quiet                           Don't output any status messages

— osmorgan
fonte

Un po 'un riassunto su come funziona sarebbe meglio;)

— Tensibai

Ti preghiamo di rivedere la modifica che ho applicato alla tua (interessante!) Risposta, ma che penso abbia avuto il rischio di essere cancellata (tramite moderazione) perché era praticamente una risposta solo link (come indicato in qualche modo anche dal commento di @Tensibai ).

— Pierre.Vriens

Se è necessario attivare un processo dalla CLI e attendere il suo completamento, è possibile utilizzare "CLI Jenkins" (vedere qui ).

Tuttavia, la CLI di jenkins non supporta le promozioni, quindi per loro ho ideato il seguente script:

#!/bin/bash
# Trigger a promotion and wait for its completion
#
# For triggering jobs jenkins cli is sufficient: https://support.cloudbees.com/hc/en-us/articles/228392127-How-to-wait-for-build-to-finish-when-triggering-from-CLI-
#
# The script is dependent on the current jenkins implementation of:
# - the promotion web page (links for triggering/re-executing the promotions)
# - the behaviour of the XML of the promotion status
#
# The behaviour of the job run status XML is:
#   - if the the promotion is not yet been triggered than the response is 404 not found
#   - is the the promotion has been triggered
#     - ... but it's still waiting for an executor:  the response is 404 not found or <duration> is empty
#     - ... and has started:                         <duration> is empty or 0
#     - ... and it's finished:                       <duration> is a non-zero number
#
#
#  run syntax:
#    ./trigger_promotion_and_wait_for_completion.sh \
#       <job_name> \
#       <job_run_number_to_promote> \
#       <promotion_name> \
#       <jenkins_user> \
#       <jenkins_pwd> \
#       <jenkins_url> \
#       <script_workspace_folder> \
#       '{"name": "prom_param_1", "value": "stringvalue" } , {"name": "prom_param_2", "value": true }'
#
#   example:
#    ./trigger_promotion_and_wait_for_completion.sh \
#       job1 \
#       22 \
#       promotion1 \
#       admin \
#       password \
#       http://localhost:8080/jenkins/ \
#       . \
#       '{"name": "prom_param_1", "value": "stringvalue" } , {"name": "prom_param_2", "value": true }'



set -uexo pipefail

#other debug options:
#PS4='+\t '
#set -v

JOB_NAME="${1}"
JOB_RUN_NUMBER="${2}"
DEPLOY_PROMOTION_NAME="${3}"
BUILDER_USER="${4}"
BUILDER_PASSWORD="${5}"
JENKINS_URL="${6}"
WORKSPACE_FOLDER="${7}"
PROMOTION_ARGUMENTS="${8}"

TIMEOUT=900


echo "retrieving the promotion nextBuildNumber (so we can poll the promotion status and check if it's finished)..."
PROMOTION_RUN_NUMBER="$( curl -s  -u${BUILDER_USER}:${BUILDER_PASSWORD} "${JENKINS_URL}job/${JOB_NAME}/promotion/process/${DEPLOY_PROMOTION_NAME}/api/xml"  | grep -P  '<nextBuildNumber>(.*)</nextBuildNumber'   | sed -re 's/.*<nextBuildNumber>(.*)<\/nextBuildNumber.*/\1/g' )"




echo "running the promotion..."
echo 'only the first promotion can be triggered with "Approve", while subsequent promotions are triggered with "Re-execute promotion"'
echo 'so we check in the web page if the approve link is present'

#the link to search in the web page
PROMOTION_APPROVE_STRING="promotionProcess/${DEPLOY_PROMOTION_NAME}/promotionCondition/hudson.plugins.promoted_builds.conditions.ManualCondition/approve"

PROM_STATUS_URL="${JENKINS_URL}job/${JOB_NAME}/${JOB_RUN_NUMBER}/promotion/"

if curl -s -vvv -u${BUILDER_USER}:${BUILDER_PASSWORD}  "${PROM_STATUS_URL}" | grep "${PROMOTION_APPROVE_STRING}" ; then
    echo "The job has not yet been promoted, triggering it with 'Approve'"
    WEB_PATH="job/${JOB_NAME}/${JOB_RUN_NUMBER}/promotion/promotionProcess/${DEPLOY_PROMOTION_NAME}/promotionCondition/hudson.plugins.promoted_builds.conditions.ManualCondition/approve"
    SUBMIT="Approve"
else
    echo "The job has already been promoted, triggering it with 'Re-execute promotion'"
    WEB_PATH="job/${JOB_NAME}/${JOB_RUN_NUMBER}/promotion/${DEPLOY_PROMOTION_NAME}/build"
    SUBMIT="Re-execute+promotion"
fi


#note for the troubleshooting: in case the following curl fails then the error cause can be found near the string "stack trace"
CURL_OUTPUT="$(  curl -s -vvv -XPOST -u${BUILDER_USER}:${BUILDER_PASSWORD} "${JENKINS_URL}${WEB_PATH}"  \
                       --data 'json={
                                 "parameter": [
                                                '"${PROMOTION_ARGUMENTS}"'
                                              ]
                                    }&Submit='"${SUBMIT}" 2>&1 )"

if ( echo "${CURL_OUTPUT}" |  grep -P "< HTTP/1.1 5\d\d" ) || ( echo "${CURL_OUTPUT}" |  grep -P "< HTTP/1.1 4\d\d" ) ; then
  echo  'error in triggering the job/promotion! exiting...'
  exit 1
else
  echo  'curl good'
fi



echo "checking promotion status until promotion is finished"
FINISHED=no


INITIAL_TIME="$(date +%s)"
while [ "${FINISHED}" != "ok" ]
do
    sleep 2


    #checking if promotion is finished (we check the value of <duration> XML element in the job run status)
    ERROR="" ; DURATION="$(curl -s  -u${BUILDER_USER}:${BUILDER_PASSWORD} "${JENKINS_URL}job/${JOB_NAME}/${JOB_RUN_NUMBER}/promotion/${DEPLOY_PROMOTION_NAME}/promotionBuild/${PROMOTION_RUN_NUMBER}/api/xml"    | grep -Po  '<duration>.*</duration>'   | sed -re  's/<duration>(.*)<\/duration>/\1/g' )"   || ERROR="yes"
    if [[ $ERROR == "yes" ]] ; then
      echo " the promotion has been queued but not yet started, waiting for it to start..."
      curl -s  -u${BUILDER_USER}:${BUILDER_PASSWORD} "${JENKINS_URL}job/${JOB_NAME}/${JOB_RUN_NUMBER}/promotion/${DEPLOY_PROMOTION_NAME}/promotionBuild/${PROMOTION_RUN_NUMBER}/api/xml"
      ERROR=""
      continue
    fi  ;   ERROR=""




    #we interrupt the polling of the job/promotion status if the promotion
    #  - is terminated
    #  - is taking too long (there is some problem)
    #(in the XML of the job run status the <duration> XML element value is initially empty, than it is 0, and eventually is the number of seconds of the run duration )

    POLLING_TIME="$(date +%s)"
    let "ELAPSED_TIME=POLLING_TIME-INITIAL_TIME"
    echo "ELAPSED_TIME=${ELAPSED_TIME}"

    if  (( ${ELAPSED_TIME} \> $TIMEOUT ))  ; then
      echo "error: the promotion has taken too long... exiting"
      exit 1
    fi

    if  [[ "${DURATION}" != "" ]] ; then
      re='^[0-9]+$'
      if [[ $DURATION =~ $re ]] ; then
        if (( "${DURATION}" \> "0" )) ; then
          FINISHED=ok
        else
          : #do nothing (the value of <duration> is 0 , that is the job/promotion has been started in a slave and is still running)
        fi
      else
        echo "error: the promotion duration is not a number. exiting..."
        exit 1
      fi
    else
      :  # the job/promotion has not yet started
    fi

    echo "waiting for the promotion to finish..."
done

echo "Promotion finished"



echo "Promotion output:"
curl -s  -u${BUILDER_USER}:${BUILDER_PASSWORD} "${JENKINS_URL}job/${JOB_NAME}/${JOB_RUN_NUMBER}/promotion/${DEPLOY_PROMOTION_NAME}/promotionBuild/${PROMOTION_RUN_NUMBER}/consoleText" > ${WORKSPACE_FOLDER}/promotionOutput

cat ${WORKSPACE_FOLDER}/promotionOutput

if [[ ! "$(tail -n1 ${WORKSPACE_FOLDER}/promotionOutput)" =~ "SUCCESS"  ]] ; then
      echo "Promotion did not successfully terminate"
      exit 1
else
      echo "Promotion successfully terminated"
fi

— Giuliano
fonte