Replace multi-role checks with action plugin

This approach should make it easier to add new checks without having to
write lots of YAML and doing things against Ansible (e.g.
ignore_errors).

A single action plugin determines what checks to run per each host,
including arguments to the check. A check is implemented as a class with
a run method, with the same signature as an action plugin and module,
and is normally backed by a regular Ansible module.

Each check is implemented as a separate Python file. This allows whoever
adds a new check to focus solely in a single Python module, and
potentially an Ansible module within library/ too.

All checks are automatically loaded, and only active checks that are
requested by the playbook get executed.

5 files changed, 188 insertions, 0 deletions

diff --git a/roles/openshift_health_checker/openshift_checks/__init__.py b/roles/openshift_health_checker/openshift_checks/__init__.py
new file mode 100644
index 000000000..d893ba591
--- /dev/null
+++ b/roles/openshift_health_checker/openshift_checks/__init__.py
@@ -0,0 +1,57 @@
+"""
+Health checks for OpenShift clusters.
+"""
+
+import os
+from abc import ABCMeta, abstractmethod, abstractproperty
+from importlib import import_module
+
+import six
+
+
+class OpenShiftCheckException(Exception):
+    """Raised when a check cannot proceed."""
+    pass
+
+
+@six.add_metaclass(ABCMeta)
+class OpenShiftCheck(object):
+    """A base class for defining checks for an OpenShift cluster environment."""
+
+    def __init__(self, module_executor):
+        self.module_executor = module_executor
+
+    @abstractproperty
+    def name(self):
+        """The name of this check, usually derived from the class name."""
+        return "openshift_check"
+
+    @classmethod
+    def is_active(cls, task_vars):  # pylint: disable=unused-argument
+        """Returns true if this check applies to the ansible-playbook run."""
+        return True
+
+    @abstractmethod
+    def run(self, tmp, task_vars):
+        """Executes a check, normally implemented as a module."""
+        return {}
+
+    @classmethod
+    def subclasses(cls):
+        """Returns a generator of subclasses of this class and its subclasses."""
+        for subclass in cls.__subclasses__():  # pylint: disable=no-member
+            yield subclass
+            for subclass in subclass.subclasses():
+                yield subclass
+
+
+# Dynamically import all submodules for the side effect of loading checks.
+
+EXCLUDES = (
+    "__init__.py",
+    "mixins.py",
+)
+
+for name in os.listdir(os.path.dirname(__file__)):
+    if name.endswith(".py") and name not in EXCLUDES:
+        import_module(__package__ + "." + name[:-3])
diff --git a/roles/openshift_health_checker/openshift_checks/mixins.py b/roles/openshift_health_checker/openshift_checks/mixins.py
new file mode 100644
index 000000000..4e0415944
--- /dev/null
+++ b/roles/openshift_health_checker/openshift_checks/mixins.py
@@ -0,0 +1,24 @@
+# pylint: disable=missing-docstring
+from openshift_checks import OpenShiftCheckException
+
+
+class NotContainerized(object):
+    """Mixin for checks that are only active when not in containerized mode."""
+
+    @classmethod
+    def is_active(cls, task_vars):
+        return (
+            # This mixin is meant to be used with subclasses of
+            # OpenShiftCheck. Pylint disables this by default on mixins,
+            # though it relies on the class name ending in 'mixin'.
+            # pylint: disable=no-member
+            super(NotContainerized, cls).is_active(task_vars) and
+            not cls.is_containerized(task_vars)
+        )
+
+    @staticmethod
+    def is_containerized(task_vars):
+        try:
+            return task_vars["openshift"]["common"]["is_containerized"]
+        except (KeyError, TypeError):
+            raise OpenShiftCheckException("'openshift.common.is_containerized' is undefined")
diff --git a/roles/openshift_health_checker/openshift_checks/package_availability.py b/roles/openshift_health_checker/openshift_checks/package_availability.py
new file mode 100644
index 000000000..4260cbf7c
--- /dev/null
+++ b/roles/openshift_health_checker/openshift_checks/package_availability.py
@@ -0,0 +1,69 @@
+# pylint: disable=missing-docstring
+from openshift_checks import OpenShiftCheck, OpenShiftCheckException
+from openshift_checks.mixins import NotContainerized
+
+
+class PackageAvailability(NotContainerized, OpenShiftCheck):
+    """Check that required RPM packages are available."""
+
+    name = "package_availability"
+
+    def run(self, tmp, task_vars):
+        try:
+            rpm_prefix = task_vars["openshift"]["common"]["service_type"]
+        except (KeyError, TypeError):
+            raise OpenShiftCheckException("'openshift.common.service_type' is undefined")
+
+        group_names = task_vars.get("group_names", [])
+
+        packages = set()
+
+        if "masters" in group_names:
+            packages.update(self.master_packages(rpm_prefix))
+        if "nodes" in group_names:
+            packages.update(self.node_packages(rpm_prefix))
+
+        args = {"packages": sorted(set(packages))}
+        return self.module_executor("check_yum_update", args, tmp, task_vars)
+
+    @staticmethod
+    def master_packages(rpm_prefix):
+        return [
+            "{rpm_prefix}".format(rpm_prefix=rpm_prefix),
+            "{rpm_prefix}-clients".format(rpm_prefix=rpm_prefix),
+            "{rpm_prefix}-master".format(rpm_prefix=rpm_prefix),
+            "bash-completion",
+            "cockpit-bridge",
+            "cockpit-docker",
+            "cockpit-kubernetes",
+            "cockpit-shell",
+            "cockpit-ws",
+            "etcd",
+            "httpd-tools",
+        ]
+
+    @staticmethod
+    def node_packages(rpm_prefix):
+        return [
+            "{rpm_prefix}".format(rpm_prefix=rpm_prefix),
+            "{rpm_prefix}-node".format(rpm_prefix=rpm_prefix),
+            "{rpm_prefix}-sdn-ovs".format(rpm_prefix=rpm_prefix),
+            "bind",
+            "ceph-common",
+            "dnsmasq",
+            "docker",
+            "firewalld",
+            "flannel",
+            "glusterfs-fuse",
+            "iptables-services",
+            "iptables",
+            "iscsi-initiator-utils",
+            "libselinux-python",
+            "nfs-utils",
+            "ntp",
+            "openssl",
+            "pyparted",
+            "python-httplib2",
+            "PyYAML",
+            "yum-utils",
+        ]
diff --git a/roles/openshift_health_checker/openshift_checks/package_update.py b/roles/openshift_health_checker/openshift_checks/package_update.py
new file mode 100644
index 000000000..316a776f5
--- /dev/null
+++ b/roles/openshift_health_checker/openshift_checks/package_update.py
@@ -0,0 +1,13 @@
+# pylint: disable=missing-docstring
+from openshift_checks import OpenShiftCheck
+from openshift_checks.mixins import NotContainerized
+
+
+class PackageUpdate(NotContainerized, OpenShiftCheck):
+    """Check that there are no conflicts in RPM packages."""
+
+    name = "package_update"
+
+    def run(self, tmp, task_vars):
+        args = {"packages": []}
+        return self.module_executor("check_yum_update", args, tmp, task_vars)
diff --git a/roles/openshift_health_checker/openshift_checks/package_version.py b/roles/openshift_health_checker/openshift_checks/package_version.py
new file mode 100644
index 000000000..a473119f3
--- /dev/null
+++ b/roles/openshift_health_checker/openshift_checks/package_version.py
@@ -0,0 +1,25 @@
+# pylint: disable=missing-docstring
+from openshift_checks import OpenShiftCheck, OpenShiftCheckException
+from openshift_checks.mixins import NotContainerized
+
+
+class PackageVersion(NotContainerized, OpenShiftCheck):
+    """Check that available RPM packages match the required versions."""
+
+    name = "package_version"
+
+    @classmethod
+    def is_active(cls, task_vars):
+        return (
+            super(PackageVersion, cls).is_active(task_vars)
+            and task_vars.get("deployment_type") == "openshift-enterprise"
+        )
+
+    def run(self, tmp, task_vars):
+        try:
+            openshift_release = task_vars["openshift_release"]
+        except (KeyError, TypeError):
+            raise OpenShiftCheckException("'openshift_release' is undefined")
+
+        args = {"version": openshift_release}
+        return self.module_executor("aos_version", args, tmp, task_vars)