(feat)(cr) define the chaos custom resource specifications (litmuscha…

…os#3)

* (feat)(cr) define the chaos custom resource specifications

Signed-off-by: ksatchit <[email protected]>

deploy/crds/chaosengine.yaml

@@ -0,0 +1,101 @@
+---
+## This is the chaos engine resource requested by dev for the nginx 
+## app, i.e., the user facing custom resource. Mapped to a dedicated
+## controller which triggers the actual chaos experiments as per 
+## defined schedule. The app developer is expected to annotate the
+## application with litmuschaos.io/engine: <engine-name>
+
+apiVersion: litmuschaos.io/v1alpha1
+kind: ChaosEngine
+metadata:
+  ## Eventually launched chaos (cron) jobs will bear <engine>-<exp></reference>-<hash>
+  name: engine-nginx
+  namespace: litmus 
+spec:
+
+  ## This will be used to identify AUT (Application-Under-Test) 
+  ## This app will be value of probed for annotation "litmuschaos.io/chaos"
+  ## to check whether it can be subjected to chaos. 
+
+  appinfo: 
+    appns: default
+    applabel: "app=nginx"
+
+  ## The chaos experiments themselves will be "pulled" via
+  ## downloadable litmus charts and versioned against it
+  ## The chaos operator comes with a default set of experiments
+
+  experiments:
+    - name: pod-delete 
+      spec:
+
+        # The experiments can be ranked on execution priority
+        rank: 1
+
+        # An experiment may need additional data points
+        # "apart" from the "chaos-specific params" defined
+        # in the template. See spec of flaky-networks for
+        # case-in-point
+
+        # The same will be available in experiment spec as
+        # well, this definition here will override that.
+
+        components: 
+
+    - name: network-delay
+      spec:
+        rank: 2 
+        component:
+          nwinterface: ""
+
+    - name: disk-fill
+      spec: 
+        rank: 4
+        component:
+          disk: ""
+
+        # Individual experiments can also have schedule
+
+        schedule: 
+          interval: ""
+          excludedTimes: ""
+          excludedDays: ""
+          concurrencyPolicy: "" 
+
+    - name: container-kill
+      spec: 
+        ranks: 3
+        component:
+          container: ""
+
+  schedule:
+    # quarter-hourly, half-hourly, hourly, bi-hourly, trihoral, daily
+    interval: "half-hourly"
+    excludedTimes: ""
+    excludedDays: ""
+    concurrencyPolicy: "" 
+
+  ## Hold status of experiments - traceback from chaos experiment 
+  ## results. In case of scheduled experiments, the status, verdict 
+  ## and lastUpdateTime keep getting updated  
+  status:
+    experiments: 
+      - name: pod-delete 
+        status: completed
+        verdict: pass
+        lastUpdateTime: "2019-04-23T09:10:50Z"
+
+      - name: network-delay
+        status: completed
+        verdict: pass
+        lastUpdateTime: "2019-04-23T09:13:50Z"
+
+      - name: disk-fill
+        status: completed
+        verdict: fail
+        lastUpdateTime: "2019-04-23T09:16:50Z"
+
+      - name: container-kill
+        status: progressing
+        verdict: ""
+        lastUpdateTime: "2019-04-23T09:23:50Z"

deploy/crds/chaosexperiment.yaml

@@ -0,0 +1,48 @@
+---
+## An experiment is the definition of a chaos test and is listed as an item 
+## in the chaos engine to be run against a given app. Consists of reference
+## to chaosGraph (which has chaos param details and executor spec) and also
+## component information for the Object-Under-Test. The results upon running
+## this experiment is relayed back to the chaos engine which invoked it.
+
+## Experiments can be pulled from litmus charts and are versioned based on 
+## chart.
+
+## Charts should (not?) be auto-upgraded
+
+apiVersion: litmuschaos.io/v1alpha1
+kind: ChaosExperiment
+metadata:
+
+  ## Eventually launched chaos litmusbook/job will bear <name>-<hash>
+  name: pod-delete
+  namespace: nginx
+  labels:
+    chart/type: kubernetes
+    chart/version: 0.9
+
+description:
+  message: |
+    Deletes a pod belonging to a deployment/statefulset/daemonset
+    a specified number of times 
+spec:
+
+  ## A predefined Chaos template type. Analogy: *Similar* in function to storage 
+  ## classes. Will contain params "specific" to that chaos operation w/ default values.
+  ## By default, the experiment is mapped to a chaosgraph of the same name, unless 
+  ## explicitly specified
+
+  # chaosgraph: 
+
+  ## Some experiments need more info on the "object of chaos". For example, in case 
+  ## of container crash tests in a multi-container app, it is necessary to know both 
+  ## general app info (namespace, labels) as well as container name which has to undergo
+  ## failures. Component list can be kept to a minimum. 
+
+  components:
+    container: ""    
+    nwinterface: ""
+    node: ""        
+    pvc: ""         
+    disk: ""        
+

deploy/crds/chaosgraph.yaml

@@ -0,0 +1,41 @@
+---
+
+## The resource that contains the actual/low-level chaos params.
+## Consists of executor definition in its spec, with ENV default
+## values and entrypoint scripts (playbooks). The chaosGraph is
+## referenced in the chaostype parameter of the chaosExperiment
+## resource. Consists of default values for chaos-specific params
+## which are expected to be overridden from the chaosExperiment  
+
+apiVersion: litmuschaos.io/v1alpha1
+kind: ChaosGraph
+metadata: 
+  name: pod-delete
+  namespace: nginx
+  labels:
+    chart/type: kubernetes
+    chart/version: "1.0"
+spec:
+  ## The chaos operator can use different execution frameworks
+  ## to achieve a certain chaos operation. It could use Litmus
+  ## or chaostoolkit or pumba etc.., Default: litmus
+  executor: litmus
+
+  ## A low-level definition of chaos parameters which is fed
+  ## to the executor for running the experiment
+  definition:
+    labels: 
+      name: pod-delete
+    env:
+      - name: ANSIBLE_STDOUT_CALLBACK
+        value: default
+      - name: APP_NAMESPACE     # override from c.engine
+        value: ""
+      - name: APP_LABEL         # override from c.engine 
+        value: ""
+      - name: CHAOS_TYPE
+        value: "pod-delete"
+      - name: TARGET_CONTAINER
+        value: "nginx"          # override from c.experiment
+    command: ["/bin/bash"]
+    args: ["-c", "ansible-playbook ./experiments/chaos/simple_pod_failure/test.yml -i /etc/ansible/hosts -vv; exit 0"]

deploy/crds/litmuschaos_v1alpha1_chaostemplate_crd.yaml → deploy/crds/litmuschaos_v1alpha1_chaosgraph_crd.yaml

@@ -1,14 +1,14 @@
 apiVersion: apiextensions.k8s.io/v1beta1
 kind: CustomResourceDefinition
 metadata:
-  name: chaostemplates.litmuschaos.io
+  name: chaosgraph.litmuschaos.io
 spec:
   group: litmuschaos.io
   names:
-    kind: ChaosTemplate
-    listKind: ChaosTemplateList
-    plural: chaostemplates
-    singular: chaostemplate
+    kind: ChaosGraph
+    listKind: ChaosGraphList
+    plural: chaosgraphs
+    singular: chaosgraphs
   scope: Namespaced
   subresources:
     status: {}