refactor(runner): ability to opt-out of using --ephemeral / opt-in to legacy --once for GHES older than 3.3 (#1384)

* runner: Remove the ability to use the deprecated `--once` flag

Ref #1196

* runner: Ability to opt-out of using --ephemeral

Although we are going to eventually remove the ability to use the legacy --once flag as proposed in #1196, there might be folks still using legacy GHES versions 3.2 or earlier.

This commit removes the existing feature flag to opt-in for --ephemeral, while adding another feature flag RUNNER_FEATURE_FLAG_ONCE to opt-in for --once so that folks stuck in legacy GHES versions
can still use ARC.

Since this change every user starts using --ephemeral by default. If they see any issues on legacy GHES instance, RUNNER_FEATURE_FLAG_ONCE=true can be set to opt-in to keep using --once, which gives one more ARC release until they upgrade their GHES instance.

But beware, we won't support legacy GHES instances forever as it's going to be a maintenance nightmare. Please upgrade!

Ref #1196

README.md

@@ -1311,12 +1311,11 @@ spec:
         # Disables automatic runner updates
         - name: DISABLE_RUNNER_UPDATE
           value: "true"
-        # Configure runner with --ephemeral instead of --once flag
+        # Configure runner with legacy --once instead of --ephemeral flag
         # WARNING | THIS ENV VAR IS DEPRECATED AND WILL BE REMOVED
-        # IN A FUTURE VERSION OF ARC. IN 0.22.0 ARC SETS --ephemeral VIA 
+        # IN A FUTURE VERSION OF ARC.
-        # THE CONTROLLER SETTING THIS ENV VAR ON POD CREATION.
         # THIS ENV VAR WILL BE REMOVED, SEE ISSUE #1196 FOR DETAILS
-        - name: RUNNER_FEATURE_FLAG_EPHEMERAL
+        - name: RUNNER_FEATURE_FLAG_ONCE
           value: "true"
 ```
 

runner/entrypoint.sh

@@ -76,7 +76,7 @@ cd ${RUNNER_HOME}
 # past that point, it's all relative pathes from /runner
 
 config_args=()
-if [ "${RUNNER_FEATURE_FLAG_EPHEMERAL:-}" == "true" -a "${RUNNER_EPHEMERAL}" == "true" ]; then
+if [ "${RUNNER_FEATURE_FLAG_ONCE:-}" != "true" -a "${RUNNER_EPHEMERAL}" == "true" ]; then
   config_args+=(--ephemeral)
   log.debug 'Passing --ephemeral to config.sh to enable the ephemeral runner.'
 fi
@@ -141,10 +141,10 @@ if [ -z "${UNITTEST:-}" ]; then
 fi
 
 args=()
-if [ "${RUNNER_FEATURE_FLAG_EPHEMERAL:-}" != "true" -a "${RUNNER_EPHEMERAL}" == "true" ]; then
+if [ "${RUNNER_FEATURE_FLAG_ONCE:-}" == "true" -a "${RUNNER_EPHEMERAL}" == "true" ]; then
   args+=(--once)
   log.warning 'Passing --once is deprecated and will be removed as an option' \
-    'from the image and actions-runner-controller at the release of 0.24.0.' \
+    'from the image and actions-runner-controller at the release of 0.25.0.' \
     'Upgrade to GHES => 3.3 to continue using actions-runner-controller. If' \
     'you are using github.com ignore this warning.'
 fi

test/entrypoint/should_work_use_once_switch/test.sh

@@ -1,90 +0,0 @@
-#!/usr/bin/env bash
-
-# UNITTEST: should work legacy once switch set
-# Will simulate a scenario where RUNNER_FEATURE_FLAG_EPHEMERAL=false. expects:
-# - the configuration step to be run exactly once
-# - the entrypoint script to exit with no error
-# - the run.sh script to run with the --once flag
-
-source ../assets/logging.sh
-
-entrypoint_log() {
-  while read I; do
-    printf "\tentrypoint.sh: $I\n"
-  done
-}
-
-log "Setting up test area"
-export RUNNER_HOME=testarea
-mkdir -p ${RUNNER_HOME}
-
-log "Setting up the test"
-export UNITTEST=true
-export RUNNER_NAME="example_runner_name"
-export RUNNER_REPO="myorg/myrepo"
-export RUNNER_TOKEN="xxxxxxxxxxxxx"
-export RUNNER_FEATURE_FLAG_EPHEMERAL="false"
-export RUNNER_EPHEMERAL="true"
-
-# run.sh and config.sh get used by the runner's real entrypoint.sh and are part of actions/runner.
-# We change symlink dummy versions so the entrypoint.sh can run allowing us to test the real entrypoint.sh
-log "Symlink dummy config.sh and run.sh"
-ln -s ../../assets/config.sh ${RUNNER_HOME}/config.sh
-ln -s ../../assets/run.sh ${RUNNER_HOME}/run.sh
-
-cleanup() {
-  rm -rf ${RUNNER_HOME}
-  unset UNITTEST
-  unset RUNNERHOME
-  unset RUNNER_NAME
-  unset RUNNER_REPO
-  unset RUNNER_TOKEN
-  unset RUNNER_EPHEMERAL
-  unset RUNNER_FEATURE_FLAG_EPHEMERAL
-}
-
-# Always run cleanup when test ends regardless of how it ends
-trap cleanup SIGINT SIGTERM SIGQUIT EXIT
-
-log "Running the entrypoint"
-log ""
-
-# run.sh and config.sh get used by the runner's real entrypoint.sh and are part of actions/runner.
-# We change symlink dummy versions so the entrypoint.sh can run allowing us to test the real entrypoint.sh
-../../../runner/entrypoint.sh 2> >(entrypoint_log)
-
-if [ "$?" != "0" ]; then
-  error "==========================================="
-  error "FAIL | Entrypoint script did not exit successfully"
-  exit 1
-fi
-
-log "Testing if we went through the configuration step only once"
-count=`cat ${RUNNER_HOME}/counter || echo "not_found"`
-if [ ${count} != "1" ]; then
-  error "==============================================="
-  error "FAIL | The configuration step was not run exactly once"
-  exit 1
-fi
-
-success "PASS | The configuration ran ${count} time(s)"
-
-log "Testing if the configuration included the --once flag"
-if ! grep -q -- '--once' ${RUNNER_HOME}/runner_args; then
-  error "==============================================="
-  error "FAIL | The configuration did not include the --once flag, config printed below:"
-  exit 1
-fi
-
-success "PASS | The --once argument was passed in"
-
-log "Testing if run.sh ran"
-if [ ! -f "${RUNNER_HOME}/run_sh_ran" ]; then
-  error "=============================="
-  error "FAIL | The runner service has not run"
-  exit 1
-fi
-success "PASS | run.sh ran"
-success ""
-success "==========================="
-success "Test completed successfully"