[AGENTONB-2675] Remove deprecated deb/rpm installer and support SSI using install-ssi.sh script (#682)

* Remove legacy installer package and unit files

* Stop legacy installer systemd services and reload systemd graph

* Remove configuration and installation of datadog-installer

* Remove leftover from pr #624 installer/apm packages from pkg-redhat

* Remove air-gapped installer doc and some installer-specific tests

* First APM SSI iteration downloading and running the script

* fix env placement + correct yaml

* fix legacy installer removal task

* Remove installer tests

* Fix legacy installer removal conditionals

* Only create install.json file if apm_instrumentation enabled is set

* Use specific package built-in to avoid erroring out on AL2023

* Do not fail when trying to remove the package

* Change logic for APM SSI to determine installation instead of weird skipping logic

* Check apm status from installer + add .ansible to gitignore

Author: tbavelier

.circleci/config.yml

@@ -48,7 +48,6 @@ commands:
       - run: datadog-agent version
       - run: datadog-agent status || true
 
-
   test_agent_install_downgrade:
     parameters:
       version:
@@ -188,7 +187,7 @@ jobs:
       - checkout
       - when:
           condition:
-            equal: [ "6_macos", << parameters.agent_version >> ]
+            equal: ["6_macos", << parameters.agent_version >>]
           steps:
             - run:
                 name: Install Rosetta 2
@@ -224,6 +223,18 @@ jobs:
       - run: cd /tmp/system-tests/lib-injection/build/docker/python/dd-lib-python-init-test-django && sudo docker build -t system-tests/local .
       - run: sudo docker run -d --name test-app-python -p 5985:18080 system-tests/local:latest
       - run: curl --retry 10 --retry-max-time 30 --retry-all-errors localhost:5985
+      # Verify the instrumentation status from datadog-installer
+      - run: |
+            status_json=$(sudo datadog-installer status --json | jq .apm_injection_status)
+            for key in $(echo "$status_json" | jq -r 'keys[]'); do
+              value=$(echo "$status_json" | jq -r --arg k "$key" '.[$k]')
+              if [ "$value" != "true" ]; then
+                echo "❌ Check failed: $key is not true (value: $value)"
+                exit 1
+              fi
+              echo "✅ $key is true"
+            done
+            echo "🎉 All APM injection checks passed!"
       # verify that the emitted traces are received by checking agent status
       - run: |
           for i in {1..14}; do
@@ -256,147 +267,9 @@ jobs:
       - run: ps aux | grep -v grep | grep datadog-agent
       - run: >
           bash -c 'for tracer in ["java", "js", "dotnet", "python", "ruby"]; do
-            print "Checking $tracer";
             sudo datadog-installer is-installed datadog-apm-library-$tracer;
           done'
 
-  test_installer:
-    parameters:
-      ansible_version:
-        type: string
-      jinja2_native:
-        type: string
-        default: "false"
-      os:
-        type: string
-      inventory:
-        type: string
-        default: "ci.ini"
-      apm_enabled:
-        type: string
-    docker:
-      - image: datadog/docker-library:ansible_<<parameters.os>>_<<parameters.ansible_version>>
-    steps:
-      - checkout
-        # datadog-installer will bailout if there's no systemctl binary, and won't attempt to
-        # create the systemd folder to store its units
-        # Since we're running the tests in a docker container without systemd, we can "help" it
-        # proceed by pretending systemd is there
-      - run: printf "#!/bin/bash\n\nexit 0" > /usr/bin/systemctl && chmod +x /usr/bin/systemctl
-      - run: mkdir -p /etc/systemd/system/
-      - run: >
-          ANSIBLE_JINJA2_NATIVE="<<parameters.jinja2_native>>" ansible-playbook
-          -i ./ci_test/inventory/<<parameters.inventory>> "./ci_test/install_installer.yaml"
-          -e datadog_apm_instrumentation_enabled="<<parameters.apm_enabled>>"
-      - run: >
-          bash -c 'if [ -n "<<parameters.apm_enabled>>" ]; then
-              datadog-installer version;
-            elif [ -x "/opt/datadog-packages/datadog-installer" ]; then
-              echo datadog-installer should not be installed;
-              exit 2;
-            else
-              echo datadog-installer is not installed as expected;
-            fi'
-
-  test_installer_suse:
-    parameters:
-      ansible_version:
-        type: string
-      jinja2_native:
-        type: string
-        default: "false"
-      inventory:
-        type: string
-        default: "ci.ini"
-      apm_enabled:
-        type: string
-    docker:
-      - image: datadog/docker-library:ansible_suse_<<parameters.ansible_version>>
-    steps:
-      - checkout
-        # datadog-installer will bailout if there's no systemctl binary, and won't attempt to
-        # create the systemd folder to store its units
-        # Since we're running the tests in a docker container without systemd, we can "help" it
-        # proceed by pretending systemd is there
-      - run: printf "#!/bin/bash\n\nexit 0" > /usr/bin/systemctl && chmod +x /usr/bin/systemctl
-      - run: mkdir -p /etc/systemd/system/
-      - run: >
-          ANSIBLE_JINJA2_NATIVE="<<parameters.jinja2_native>>" ansible-playbook
-          -i ./ci_test/inventory/<<parameters.inventory>> "./ci_test/install_installer.yaml"
-          -e datadog_apm_instrumentation_enabled="<<parameters.apm_enabled>>"
-      - run: >
-          bash -c 'if [ -n "<<parameters.apm_enabled>>" ]; then
-              datadog-installer version;
-            elif [ -x "/opt/datadog-packages/datadog-installer" ]; then
-              echo datadog-installer should not be installed;
-              exit 2;
-            else
-              echo datadog-installer is not installed as expected;
-            fi'
-  
-  test_installer_air_gapped_rhel:
-    parameters:
-      ansible_version:
-        type: string
-      jinja2_native:
-        type: string
-        default: "false"
-      os:
-        type: string
-      inventory:
-        type: string
-        default: "ci.ini"
-      apm_enabled:
-        type: string
-    docker:
-      - image: datadog/docker-library:ansible_<<parameters.os>>_<<parameters.ansible_version>>
-    steps:
-      - checkout
-        # datadog-installer will bailout if there's no systemctl binary, and won't attempt to
-        # create the systemd folder to store its units
-        # Since we're running the tests in a docker container without systemd, we can "help" it
-        # proceed by pretending systemd is there
-      - run: printf "#!/bin/bash\n\nexit 0" > /usr/bin/systemctl && chmod +x /usr/bin/systemctl
-      - run: mkdir -p /etc/systemd/system/
-      - run: >
-          ANSIBLE_JINJA2_NATIVE="<<parameters.jinja2_native>>" ansible-playbook -v
-          -i ./ci_test/inventory/<<parameters.inventory>> "./ci_test/install_installer_air_gapped.yaml"
-          -e datadog_apm_instrumentation_enabled="<<parameters.apm_enabled>>"
-      - run: >
-          bash -c 'if [ -n "<<parameters.apm_enabled>>" ]; then
-              datadog-installer version;
-            elif [ -x "/opt/datadog-packages/datadog-installer" ]; then
-              echo datadog-installer should not be installed;
-              exit 2;
-            else
-              echo datadog-installer is not installed as expected;
-            fi
-
-            if [ -n "<<parameters.apm_enabled>>" ]; then
-              echo "Checking if datadog-apm-inject is installed...";
-              if ! datadog-installer is-installed datadog-apm-inject; then
-                echo "datadog-apm-inject is NOT installed as expected";
-                exit 3;
-              else
-                echo "datadog-apm-inject is installed";
-              fi
-            fi'
-
-  test_installer_over_pinned:
-    docker:
-      - image: datadog/docker-library:ansible_debian_2_10
-    steps:
-      - checkout
-        # datadog-installer will bailout if there's no systemctl binary, and won't attempt to
-        # create the systemd folder to store its units
-        # Since we're running the tests in a docker container without systemd, we can "help" it
-        # proceed by pretending systemd is there
-      - run: printf "#!/bin/bash\n\nexit 0" > /usr/bin/systemctl && chmod +x /usr/bin/systemctl
-      - run: mkdir -p /etc/systemd/system/
-      - run: DD_API_KEY=woof DD_SITE="datadoghq.com" DD_INSTALL_ONLY=true DD_AGENT_MINOR_VERSION=53 bash -c "$(curl -L https://install.datadoghq.com/scripts/install_script_agent7.sh)"
-      - run: >
-          ansible-playbook -v -i ./ci_test/inventory/ci.ini "./ci_test/install_installer_over_pinned.yaml"
-
   test_incorrect_rhel6_detect:
     # Ensure some RHEL derivatives aren't incorrectly detected as RHEL 6
     docker:
@@ -518,58 +391,16 @@ workflows:
             parameters:
               ansible_version: ["6.7.0", "12.0.0"]
 
-      - test_installer:
-         matrix:
-           parameters:
-              ansible_version: ["2_10", "3_4", "4_10"]
-              os: ["debian", "rocky8", "amazonlinux2023"]
-              apm_enabled: ["host", ""]
-
-      - test_installer:
-         matrix:
-           parameters:
-              ansible_version: ["12_0"]
-              os: ["debian", "rocky9", "amazonlinux2023"]
-              apm_enabled: ["host", ""]
-
-      - test_installer_suse:
-         matrix:
-           parameters:
-              ansible_version: ["2_10", "3_4", "4_10"]
-              apm_enabled: ["host", ""]
-
-      - test_installer_suse:
-         matrix:
-           parameters:
-              ansible_version: ["12_0"]
-              apm_enabled: ["host", ""]
-
-      - test_installer_air_gapped_rhel:
-         matrix:
-           parameters:
-              ansible_version: ["2_10", "3_4", "4_10"]
-              os: ["rocky8", "amazonlinux2023"]
-              apm_enabled: ["host", ""]
-
-      - test_installer_air_gapped_rhel:
-         matrix:
-           parameters:
-              ansible_version: ["12_0"]
-              os: ["rocky9", "amazonlinux2023"]
-              apm_enabled: ["host", ""]
-
-      - test_installer_over_pinned
-
       - test_incorrect_rhel6_detect:
-         matrix:
-           parameters:
+          matrix:
+            parameters:
               ansible_version: ["4_10"]
-              os: [ "amazonlinux2023"]
+              os: ["amazonlinux2023"]
               python: ["python3"]
 
       - test_incorrect_rhel6_detect:
-         matrix:
-           parameters:
+          matrix:
+            parameters:
               ansible_version: ["12_0"]
-              os: [ "amazonlinux2023"]
+              os: ["amazonlinux2023"]
               python: ["python3.11"]

.gitignore

@@ -1,6 +1,7 @@
 .idea/
 *.retry
 .venv
+.ansible
 
 # Vagrant
 .vagrant

README.md

@@ -46,27 +46,6 @@ To deploy the Datadog Agent on hosts, add the Datadog role and your API key to y
 
 The API key is required and its absence causes the role to fail. If you want to provide it through another way, outside of Ansible's control, specify a placeholder key and substitute the key at a later point.
 
-### Air-gapped environments
-
-To install Datadog in an air-gapped environment using a specific registry and mirrored images, use the Datadog Ansible collection along with the `datadog_installer_registry`, `datadog_installer_auth`, and `agent_datadog_config` variables. While Datadog recommends fetching `latest`, `datadog_installer_version` and `datadog_apm_inject_version` can be provided to pin versions.
-
-**Note**: `agent_datadog_config` overrides the `installer_registry_config` setting.
-
-For example:
-
-```yaml
-name: Datadog Agent Install
-  include_role:
-    name: datadog.dd.agent
-  vars:
-    datadog_installer_registry: "my.local.registry"
-    datadog_installer_version: 7.63
-    datadog_apm_inject_version: 0.38
-    datadog_yum_repo: "my.local.repo"
-    datadog_api_key: "MY_DATADOG_API_KEY"
-    datadog_site: "MY_DATADOG_SITE"
-```
-
 ## Role variables
 
 These variables provide additional configuration during the installation of the Datadog Agent. They should be specified in the `vars` section of your playbook.

ci_test/install_installer.yaml

@@ -64,8 +64,6 @@ datadog_agent_version: ""
 # Default Package name for APT and RPM installs - can override in playbook for IOT Agent
 datadog_agent_flavor: "datadog-agent"
 
-datadog_installer_flavor: "datadog-installer"
-
 # Default apt repo and keyserver
 
 # By default, the role uses the official apt Datadog repository for the chosen major version
@@ -181,33 +179,27 @@ datadog_windows_config_root: "{{ ansible_facts.env['ProgramData'] }}\\Datadog"
 # arguments to supply to the windows installer.
 win_install_args: " "
 
+# Advanced users - Configure datadog-installer to support advanced scenarios
+datadog_installer_registry: ""
+datadog_installer_auth: ""
+datadog_installer_version: ""
+datadog_apm_inject_version: ""
+
 # Configure APM host injection. Possible values are: "all", "host" or "docker".
-datadog_apm_instrumentation_enabled: ""
+# datadog_apm_instrumentation_enabled: ""
 
 # List of APM libraries with pinned versions to install if host or Docker injection is enabled.
 # Example: `["java:1.23.0", "python:2.0.9"]`
 # You can see the available values in our official docs:
 # https://docs.datadoghq.com/tracing/trace_collection/library_injection_local
 datadog_apm_instrumentation_libraries: []
 
-# Used to send telemetry data to datadog on installation
-datadog_apm_telemetry_endpoint: "https://instrumentation-telemetry-intake.{{ datadog_site | default('datadoghq.com') }}/api/v2/apmtelemetry"
-
 # Enable remote updates through datadog-installer
 datadog_remote_updates: false
 
 # Set infrastructure monitoring mode (optional)
 # datadog_infrastructure_mode: ""
 
-# Registry, auth, and version for the datadog installer
-datadog_installer_registry: ""
-datadog_installer_auth: ""
-datadog_installer_version: ""
-installer_registry_config: {}
-
-# Version for the datadog apm inject package
-datadog_apm_inject_version: ""
-
 #
 # Internal variables
 # The following variables are for internal use only, do not modify them.
@@ -266,8 +258,6 @@ datadog_macos_etc_dir: "/opt/datadog-agent/etc"
 datadog_macos_logs_dir: "/opt/datadog-agent/logs"
 datadog_macos_run_dir: "/opt/datadog-agent/run"
 
-datadog_installer_enabled: false
-
 datadog_ansible_facts_subset:
   - '!all'
   - '!any'

ci_test/install_installer_air_gapped.yaml

@@ -19,15 +19,6 @@
   when: datadog_enabled and not ansible_check_mode and not ansible_facts.os_family == "Windows" and not ansible_facts.os_family == "Darwin"
     and (ansible_facts.services['datadog-agent-exp.service'] is not defined or ansible_facts.services['datadog-agent-exp.service'].state != "running")
 
-- name: restart datadog-installer  # noqa name[casing]
-  ansible.builtin.service:
-    name: datadog-installer
-    state: restarted
-    use: service
-  # The installer currently only setup its systemd unit when APM instrumentation is enabled
-  when: datadog_enabled and datadog_installer_enabled and
-    not ansible_check_mode and not ansible_facts.os_family == "Windows" and not ansible_facts.os_family == "Darwin"
-
 # We can't add the Windows Agent service restart handler directly here because that makes the role require
 # the ansible.windows collection on all platforms. We only want it to be needed on Windows.
 # Therefore, what we do is the following: when needed, our Windows tasks call this handler to require a