Enable netchecker for CI

* Enable netchecker app for CI postinstall tests
* Rework outputs and better coverage to the ping between pods post
intall test case. With netchecker deployed, the test covers hostnet
to hostnet and standard to standrad pods ping check as well.

Signed-off-by: Bogdan Dobrelya <bdobrelia@mirantis.com>

.travis.yml

@@ -127,7 +127,7 @@ script:
     -e inventory_path=${PWD}/inventory/inventory.ini
     -e cloud_region=${CLOUD_REGION}
 
-    # Create cluster
+    # Create cluster with netchecker app deployed
   - >
     $HOME/.local/bin/ansible-playbook -i inventory/inventory.ini -u $SSH_USER -e ansible_ssh_user=$SSH_USER $SSH_ARGS
     -b --become-user=root -e cloud_provider=gce  $LOG_LEVEL -e kube_network_plugin=${KUBE_NETWORK_PLUGIN}
@@ -136,15 +136,16 @@ script:
     -e download_run_once=true
     -e download_localhost=true
     -e local_release_dir=/var/tmp/releases
+    -e deploy_netchecker=true
     cluster.yml
 
     # Tests Cases
     ## Test Master API
-  - $HOME/.local/bin/ansible-playbook -i inventory/inventory.ini -e ansible_python_interpreter=${PYPATH} tests/testcases/010_check-apiserver.yml $LOG_LEVEL
-    ## Create a POD
-  - $HOME/.local/bin/ansible-playbook -i inventory/inventory.ini -e ansible_python_interpreter=${PYPATH} -u $SSH_USER -e ansible_ssh_user=$SSH_USER $SSH_ARGS -b --become-user=root tests/testcases/020_check-create-pod.yml $LOG_LEVEL
+  - $HOME/.local/bin/ansible-playbook -i inventory/inventory.ini -e ansible_python_interpreter=${PYPATH} -u $SSH_USER -e ansible_ssh_user=$SSH_USER $SSH_ARGS -b --become-user=root tests/testcases/010_check-apiserver.yml $LOG_LEVEL
     ## Ping the between 2 pod
   - $HOME/.local/bin/ansible-playbook -i inventory/inventory.ini -e ansible_python_interpreter=${PYPATH} -u $SSH_USER -e ansible_ssh_user=$SSH_USER $SSH_ARGS -b --become-user=root tests/testcases/030_check-network.yml $LOG_LEVEL
+    ## Advanced DNS checks
+  - $HOME/.local/bin/ansible-playbook -i inventory/inventory.ini -e ansible_python_interpreter=${PYPATH} -u $SSH_USER -e ansible_ssh_user=$SSH_USER $SSH_ARGS -b --become-user=root tests/testcases/040_check-network-adv.yml $LOG_LEVEL
 
 after_script:
   - >

tests/scripts/ansibl8s_test.sh

@@ -39,14 +39,14 @@ should_api_server_respond() {
   assertion__status_code_is_success $?
 }
 
-should_create_pod() {
-  ansible-playbook -i inventory.ini -s ${private_key} testcases/020_check-create-pod.yml -vv
-
-  assertion__status_code_is_success $?
-}
-
 should_pod_be_in_expected_subnet() {
   ansible-playbook -i inventory.ini -s ${private_key} testcases/030_check-network.yml -vv
 
   assertion__status_code_is_success $?
 }
+
+should_resolve_cluster_dns() {
+  ansible-playbook -i inventory.ini -s ${private_key} testcases/040_check-network-adv.yml -vv
+
+  assertion__status_code_is_success $?
+}

tests/testcases/030_check-network.yml

@@ -15,16 +15,44 @@
   - name: Get pod names
     shell: "{{bin_dir}}/kubectl get pods -o json"
     register: pods
+    no_log: true
+
+  - name: Get hostnet pods
+    command: "{{bin_dir}}/kubectl get pods -o
+             jsonpath='{range .items[?(.spec.hostNetwork)]}{.metadata.name} {.status.podIP} {end}'"
+    register: hostnet_pods
+
+  - name: Get running pods
+    command: "{{bin_dir}}/kubectl get pods -o
+             jsonpath='{range .items[?(.status.phase==\"Running\")]}{.metadata.name} {.status.podIP} {end}'"
+    register: running_pods
 
   - set_fact:
       pod_names: "{{ (pods.stdout | from_json)['items'] | map(attribute = 'metadata.name') | list }}"
-      pod_ips: "{{ (pods.stdout | from_json)['items'] | map(attribute = 'status.podIP') | list }}"
+      pod_ips: "{{ (pods.stdout | from_json)['items'] | selectattr('status.podIP', 'defined') | map(attribute = 'status.podIP') | list }}"
+      pods_hostnet: |
+        {% set list = hostnet_pods.stdout.split(" ") %}
+        {{list}}
+      pods_running: |
+        {% set list = running_pods.stdout.split(" ") %}
+        {{list}}
 
   - name: Check pods IP are in correct network
     assert:
       that: item | ipaddr(kube_pods_subnet)
+    when: not item in pods_hostnet and item in pods_running
     with_items: "{{pod_ips}}"
 
-
   - name: Ping between pods is working
-    shell: "{{bin_dir}}/kubectl exec {{pod_names[0]}} -- ping -c 4 {{ pod_ips[1] }}"
+    shell: "{{bin_dir}}/kubectl exec {{item[0]}} -- ping -c 4 {{ item[1] }}"
+    when: not item[0] in pods_hostnet and not item[1] in pods_hostnet
+    with_nested:
+      - "{{pod_names}}"
+      - "{{pod_ips}}"
+
+  - name: Ping between hostnet pods is working
+    shell: "{{bin_dir}}/kubectl exec {{item[0]}} -- ping -c 4 {{ item[1] }}"
+    when: item[0] in pods_hostnet and item[1] in pods_hostnet
+    with_nested:
+      - "{{pod_names}}"
+      - "{{pod_ips}}"

tests/testcases/040_check-network-adv.yml

@@ -0,0 +1,71 @@
+---
+- hosts: kube-node
+  tasks:
+    - name: Test tunl0 routes
+      shell: "! /sbin/ip ro | grep '/26 via' | grep -v tunl0"
+      when: (ipip|default(false) or cloud_provider is defined) and (kube_network_plugin == 'calico')
+
+- hosts: k8s-cluster
+  vars:
+    agent_report_interval: 10
+    netcheck_namespace: default
+    netchecker_port: 31081
+
+  tasks:
+    - name: Force binaries directory for CoreOS
+      set_fact:
+        bin_dir: "/opt/bin"
+      when: ansible_os_family == "CoreOS"
+
+    - set_fact:
+        bin_dir: "/usr/local/bin"
+      when: ansible_os_family != "CoreOS"
+
+    - name: Wait for netchecker server
+      shell: "{{ bin_dir }}/kubectl get pods --namespace {{netcheck_namespace}} | grep ^netchecker-server"
+      delegate_to: "{{groups['kube-master'][0]}}"
+      run_once: true
+      register: ncs_pod
+      until: ncs_pod.stdout.find('Running') != -1
+      retries: 10
+      delay: 10
+
+    - name: Wait for netchecker agents
+      shell: "{{ bin_dir }}/kubectl get pods --namespace {{netcheck_namespace}} | grep '^netchecker-agent-.*Running'"
+      run_once: true
+      delegate_to: "{{groups['kube-master'][0]}}"
+      register: nca_pod
+      until: "{{ nca_pod.stdout_lines|length }} >= {{ groups['kube-node']|length * 2 }}"
+      retries: 10
+      delay: 10
+
+    - name: Get netchecker agents
+      uri: url=http://localhost:{{netchecker_port}}/api/v1/agents/ return_content=yes
+      run_once: true
+      delegate_to: "{{groups['kube-node'][0]}}"
+      register: agents
+      retries: 20
+      delay: "{{ agent_report_interval }}"
+      until: "{{ agents.content != 'No JSON object could be decoded' and agents.content != '{}' and (agents.content|default('{}'))|from_json|length >= groups['kube-node']|length * 2 }}"
+      ignore_errors: true
+      no_log: true
+
+    - debug: var=agents.content|from_json
+      failed_when: not agents|success
+      delegate_to: "{{groups['kube-node'][0]}}"
+      run_once: true
+
+    - name: Check netchecker status
+      uri: url=http://localhost:{{netchecker_port}}/api/v1/connectivity_check status_code=200 return_content=yes
+      delegate_to: "{{groups['kube-node'][0]}}"
+      run_once: true
+      register: result
+      retries: 5
+      delay: "{{ agent_report_interval }}"
+      no_log: true
+      ignore_errors: true
+
+    - debug: var=result.content|from_json
+      failed_when: not result|success
+      delegate_to: "{{groups['kube-node'][0]}}"
+      run_once: true