Add perfdata support

This commit reworks condition and result handling code
to add support for perfdata and multi-line output.

Author: Robert Wikman

check_k8s.py

@@ -5,17 +5,13 @@
 import traceback
 import json
 
-from collections import namedtuple
 from urllib.error import URLError, HTTPError
 
 from k8s.components import MAPPINGS
 from k8s.cli import parse_cmdline
 from k8s.http import build_url, request
-from k8s.consts import NAGIOS_MSG, Severity
-from k8s.exceptions import PluginException
-
-
-Output = namedtuple("Output", ["state", "message", "channel"])
+from k8s.consts import NAGIOS_MSG, NaemonState
+from k8s.result import Output
 
 
 def main():
@@ -38,28 +34,27 @@ def main():
     # Request and check health data
     try:
         response, status = request(url, token=parsed.token, insecure=parsed.insecure)
-        result = health_check(response)
-        output = Output(Severity.OK, result, sys.stdout)
-    except PluginException as e:
-        output = Output(e.state, e.message, sys.stderr)
+        output = health_check(response).output
+        if not isinstance(output, Output):
+            raise TypeError("Unknown health check format")
     except HTTPError as e:
         body = json.loads(e.read().decode("utf8"))
         output = Output(
-            Severity.UNKNOWN,
+            NaemonState.UNKNOWN,
             "{0}: {1}".format(e.code, body.get("message")),
             sys.stderr
         )
     except URLError as e:
-        output = Output(Severity.UNKNOWN, e.reason, sys.stderr)
+        output = Output(NaemonState.UNKNOWN, e.reason, sys.stderr)
     except Exception as e:
         if parsed.debug:
             exc_type, exc_value, exc_traceback = sys.exc_info()
             traceback.print_tb(exc_traceback, file=sys.stdout)
 
-        output = Output(Severity.UNKNOWN, e, sys.stderr)
+        output = Output(NaemonState.UNKNOWN, e, sys.stderr)
 
     msg = NAGIOS_MSG.format(state=output.state.name, message=output.message)
-    print(msg, file=output.channel)
+    output.channel.write(msg)
     sys.exit(output.state.value)
 
 

k8s/components/deployment/check.py

@@ -1,4 +1,4 @@
-from k8s.exceptions import NagiosCritical, NagiosWarning
+from k8s.result import Result
 
 from .resource import Deployment
 
@@ -14,19 +14,4 @@ def check_deployments(items):
     :return: Deployments health summary
     """
 
-    for item in items:
-        deployment = Deployment(item)
-        reps = deployment.replicas
-
-        if deployment.alerts_critical:
-            raise NagiosCritical(deployment.alerts_critical[0])
-        elif deployment.alerts_warning:
-            raise NagiosWarning(deployment.alerts_warning[0])
-
-        if reps.available < reps.total or reps.updated < reps.total:
-            if reps.available != 0 and reps.updated != 0:
-                raise NagiosWarning("Deployment degraded", **deployment.meta)
-
-            raise NagiosCritical("Deployment unavailable", **deployment.meta)
-
-    return "Found {} healthy Deployments".format(len(items))
+    return Result(Deployment, items)

k8s/components/deployment/resource.py

@@ -1,22 +1,40 @@
 from collections import namedtuple
+from enum import Enum
+
+from k8s.consts import NaemonState
+
+from ..resource import Resource, NaemonStatus
 
-from k8s.resource import Resource
-from k8s.consts import Severity
 
 Replicas = namedtuple("Replicas", ["total", "ready", "updated", "available"])
 
 
 class Deployment(Resource):
-    def __init__(self, data):
-        super(Deployment, self).__init__(data)
+    class PerfMap(Enum):
+        AVAILABLE = "available"
+        UNAVAILABLE = "unavailable"
+        DEGRADED = "degraded"
+        NOREPS = "noreps"
+
+    def __init__(self, data, *args, **kwargs):
+        super(Deployment, self).__init__(data, *args, **kwargs)
 
         self.replicas = Replicas(
-            self._status["replicas"],
-            self._status["readyReplicas"],
-            self._status["updatedReplicas"],
-            self._status["availableReplicas"]
+            self._status.get("replicas", 0),
+            self._status.get("readyReplicas", 0),
+            self._status.get("updatedReplicas", 0),
+            self._status.get("availableReplicas", 0)
         )
 
-    def _condition_severity(self, _type, status):
-        if _type == "Available" and status != "True":
-            return Severity.CRITICAL
+    def _get_status(self, cnd_type, cnd_status):
+        reps = self.replicas
+
+        if cnd_type == "Available":
+            if cnd_status == "True":
+                return NaemonStatus(NaemonState.OK, self.perf.AVAILABLE)
+            else:
+                return NaemonStatus(NaemonState.CRITICAL, self.perf.UNAVAILABLE)
+        elif reps.available < reps.total or reps.updated < reps.total:
+            if reps.available != 0 and reps.updated != 0:
+                return NaemonStatus(NaemonState.WARNING, self.perf.DEGRADED)
+            return NaemonStatus(NaemonState.CRITICAL, self.perf.NOREPS)

k8s/components/node/check.py

@@ -1,4 +1,4 @@
-from k8s.exceptions import NagiosCritical, NagiosWarning
+from k8s.result import Result
 
 from .resource import Node
 
@@ -14,15 +14,4 @@ def check_nodes(items):
     :return: Nodes health summary
     """
 
-    for item in items:
-        node = Node(item)
-
-        if node.alerts_critical:
-            raise NagiosCritical(node.alerts_critical[0])
-        elif node.alerts_warning:
-            raise NagiosWarning(node.alerts_warning[0])
-
-        if node.unschedulable:
-            raise NagiosWarning("Node {} is ready, but unschedulable".format(node.meta["name"]))
-
-    return "Found {} healthy Nodes".format(len(items))
+    return Result(Node, items)

k8s/components/node/resource.py

@@ -1,16 +1,34 @@
-from k8s.resource import Resource
-from k8s.consts import Severity
+from enum import Enum
+
+from k8s.consts import NaemonState
+
+from ..resource import Resource, NaemonStatus
 
 
 class Node(Resource):
-    def __init__(self, data):
-        super(Node, self).__init__(data)
+    class PerfMap(Enum):
+        AVAILABLE = "available"
+        UNAVAILABLE = "unavailable"
+        DEGRADED = "degraded"
+        UNSCHEDULABLE = "unschedulable"
+
+    def __init__(self, data, *args, **kwargs):
+        super(Node, self).__init__(data, *args, **kwargs)
 
         # https://kubernetes.io/docs/concepts/architecture/nodes/#manual-node-administration
         self.unschedulable = data["spec"].get("unschedulable", False)
 
-    def _condition_severity(self, _type, status):
-        if _type == "Ready" and status != "True":
-            return Severity.CRITICAL
-        elif _type != "Ready" and status == "True":
-            return Severity.WARNING
+    def _get_status(self, cnd_type, cnd_status):
+        if self.unschedulable:
+            return NaemonStatus(
+                NaemonState.WARNING,
+                self.perf.UNSCHEDULABLE,
+                "Node {} is ready, but unschedulable".format(self.meta["name"])
+            )
+        elif cnd_type == "Ready":
+            if cnd_status == "True":
+                return NaemonStatus(NaemonState.OK, self.perf.AVAILABLE)
+            else:
+                return NaemonStatus(NaemonState.CRITICAL, self.perf.UNAVAILABLE)
+        elif cnd_type != "Ready" and cnd_status == "True":
+            return NaemonStatus(NaemonState.WARNING, self.perf.DEGRADED)

k8s/components/pod/check.py

@@ -1,7 +1,6 @@
-from k8s.exceptions import NagiosCritical, NagiosWarning
+from k8s.result import Result
 
 from .resource import Pod
-from .consts import Phase
 
 
 def check_pods(items):
@@ -15,17 +14,4 @@ def check_pods(items):
     :return: Pods health summary
     """
 
-    for item in items:
-        pod = Pod(item)
-
-        if pod.phase == Phase.pending:
-            raise NagiosWarning("{kind} {name} is {0}".format(pod.phase.value, **pod.meta))
-        elif pod.phase != Phase.running and pod.phase != Phase.succeeded:
-            raise NagiosCritical("Unexpected Phase for {kind} {name}: {0}".format(pod.phase.value, **pod.meta))
-
-        if pod.alerts_critical:
-            raise NagiosCritical(pod.alerts_critical[0])
-        elif pod.alerts_warning:
-            raise NagiosWarning(pod.alerts_warning[0])
-
-    return "Found {} healthy Pods".format(len(items))
+    return Result(Pod, items)

k8s/components/pod/consts.py

@@ -1,6 +1,6 @@
 from enum import Enum
 
-CONDITIONS_HEALTHY = ["Ready", "Initialized", "PodScheduled", "ContainersReady"]
+STATUSES = ["Ready", "Initialized", "PodScheduled", "ContainersReady"]
 
 
 class ContainerState(Enum):

k8s/components/pod/resource.py

@@ -1,7 +1,10 @@
-from k8s.resource import Resource
-from k8s.consts import Severity
+from enum import Enum
 
-from .consts import ContainerState, Phase, CONDITIONS_HEALTHY
+from k8s.consts import NaemonState
+
+from ..resource import Resource, NaemonStatus
+
+from .consts import ContainerState, Phase, STATUSES
 
 
 class Container:
@@ -12,21 +15,35 @@ def __init__(self, data):
         # Container State is a single-item dict, with a nested dict value.
         # https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.15/#containerstate-v1-core
         state = list(data["state"].keys())
-
-        # Ensure state is known
         self.state = ContainerState(state[0])
 
 
 class Pod(Resource):
-    def __init__(self, data):
-        super(Pod, self).__init__(data)
+    class PerfMap(Enum):
+        AVAILABLE = "available"
+        UNAVAILABLE = "unavailable"
+        DEGRADED = "degraded"
+        PENDING = "pending"
 
-        self.containers = [Container(c) for c in self._status["containerStatuses"]]
-        self.phase = Phase(self._status["phase"])
+    def __init__(self, data, *args, **kwargs):
+        super(Pod, self).__init__(data, *args, **kwargs)
 
-    def _condition_severity(self, _type, status):
-        if _type in CONDITIONS_HEALTHY and status != "True":
-            return Severity.CRITICAL
-        elif _type not in CONDITIONS_HEALTHY and status == "True":
-            return Severity.WARNING
+        self.containers = [Container(c) for c in self._status.get("containerStatuses", [])]
+        self.phase = Phase(self._status["phase"])
 
+    def _get_status(self, cnd_type, cnd_status):
+        if self.phase != Phase.running and self.phase != Phase.succeeded:
+            return NaemonStatus(
+                NaemonState.CRITICAL,
+                self.perf.UNAVAILABLE,
+                "Unexpected Phase for {kind} {name}: {0}".format(self.phase.value, **self.meta)
+            )
+        elif cnd_type in STATUSES:
+            if cnd_status == "True":
+                return NaemonStatus(NaemonState.OK, self.perf.AVAILABLE)
+            else:
+                return NaemonStatus(NaemonState.CRITICAL, self.perf.UNAVAILABLE)
+        elif cnd_type not in STATUSES and cnd_status == "True":
+            return NaemonStatus(NaemonState.WARNING, self.perf.DEGRADED)
+        elif self.phase == Phase.pending:
+            return NaemonStatus(NaemonState.WARNING, self.perf.PENDING)