Add Initial support for CAPG Provider

README.md

@@ -110,6 +110,7 @@ Crashd utilizes the concept of a provider to enumerate compute resources. Each i
 * *Kubernetes Nodes Provider* - extracts host information from a Kubernetes API node objects
 * *CAPV Provider* - uses Cluster-API to discover machines in vSphere cluster
 * *CAPA Provider* - uses Cluster-API to discover machines running on AWS
+* *CAPG Provider* - uses Cluster-API to discover machines running on Google Cloud
 * More providers coming!
 
 

examples/capg_provider.crsh

@@ -0,0 +1,40 @@
+ssh_conf = ssh_config(username=args.username, jump_user=args.username, jump_host=args.jump_host, private_key_path=args.private_key_path)
+kube_conf = kube_config(path=args.mc_config)
+
+#list out management and workload cluster nodes
+wc_provider=capg_provider(
+    workload_cluster=args.cluster_name,
+    namespace=args.cluster_ns,
+    ssh_config=ssh_conf,
+    mgmt_kube_config=kube_conf
+)
+# This is broken. I need to look in to why.
+# nodes = resources(provider=wc_provider)
+print("BEFORE IT BREAKS")
+print(wc_provider)
+print("THIS IS BROKEN")
+print(resources(provider=wc_provider))
+print("THIS IS NOT BROKEN")
+nodes = resources(provider = host_list_provider(hosts=wc_provider.hosts, ssh_config=ssh_conf))
+print(nodes)
+
+capture(cmd="sudo df -i", resources=nodes)
+capture(cmd="sudo crictl info", resources=nodes)
+capture(cmd="df -h /var/lib/containerd", resources=nodes)
+capture(cmd="sudo systemctl status kubelet", resources=nodes)
+capture(cmd="sudo systemctl status containerd", resources=nodes)
+capture(cmd="sudo journalctl -xeu kubelet", resources=nodes)
+
+capture(cmd="sudo cat /var/log/cloud-init-output.log", resources=nodes)
+capture(cmd="sudo cat /var/log/cloud-init.log", resources=nodes)
+
+#add code to collect pod info from cluster
+set_defaults(kube_config(capi_provider = wc_provider))
+
+pod_ns=["default", "kube-system"]
+
+kube_capture(what="logs", namespaces=pod_ns)
+kube_capture(what="objects", kinds=["pods", "services"], namespaces=pod_ns)
+kube_capture(what="objects", kinds=["deployments", "replicasets"], groups=["apps"], namespaces=pod_ns)
+
+# archive(output_file="diagnostics.tar.gz", source_paths=[conf.workdir])

examples/capv_provider.crsh

@@ -29,4 +29,4 @@ kube_capture(what="logs", namespaces=pod_ns)
 kube_capture(what="objects", kinds=["pods", "services"], namespaces=pod_ns)
 kube_capture(what="objects", kinds=["deployments", "replicasets"], groups=["apps"], namespaces=pod_ns)
 
-archive(output_file="diagnostics.tar.gz", source_paths=[conf.workdir])
+archive(output_file="diagnostics.tar.gz", source_paths=[conf.workdir])

examples/kind-api-objects.crsh

@@ -10,12 +10,11 @@ nspaces=[
     "cert-manager", "tkg-system",
 ]
 
-set_defaults(kube_config(path=args.kubecfg))
 
 # capture Kubernetes API object and store in files (under working dir)
 kube_capture(what="objects", kinds=["services", "pods"], namespaces=nspaces)
 kube_capture(what="objects", kinds=["deployments", "replicasets"], namespaces=nspaces)
 kube_capture(what="objects", kinds=["clusters", "machines", "machinesets", "machinedeployments"], namespaces=["tkg-system"])
 
 # bundle files stored in working dir
-archive(output_file="/tmp/crashobjs.tar.gz", source_paths=[conf.workdir])
+archive(output_file="/tmp/crashobjs.tar.gz", source_paths=[conf.workdir])

k8s/client.go

@@ -116,11 +116,11 @@ func (k8sc *Client) Search(ctx context.Context, params SearchParams) ([]SearchRe
 }
 
 // Search does a drill-down search from group, version, resourceList, to resources.  The following rules are applied
-// 1) Legacy core group can be specified as "core" instead of empty string.
-// 2) All specified search params will use AND operator for match (i.e. groups=core AND kinds=pods AND versions=v1 AND ... etc)
-// 3) kinds (resources) will match resource.Kind or resource.Name
-// 4) All search params are passed as comma- or space-separated sets that are matched using OR (i.e. kinds=pods services
-//    will match resouces of type pods or services)
+//  1. Legacy core group can be specified as "core" instead of empty string.
+//  2. All specified search params will use AND operator for match (i.e. groups=core AND kinds=pods AND versions=v1 AND ... etc)
+//  3. kinds (resources) will match resource.Kind or resource.Name
+//  4. All search params are passed as comma- or space-separated sets that are matched using OR (i.e. kinds=pods services
+//     will match resouces of type pods or services)
 func (k8sc *Client) _search(ctx context.Context, groups, categories, kinds, namespaces, versions, names, labels, containers string) ([]SearchResult, error) {
 
 	// normalize params
@@ -277,7 +277,7 @@ func (k8sc *Client) _search(ctx context.Context, groups, categories, kinds, name
 					results = append(results, result)
 				}
 			} else {
-				logrus.Debugf("searching for %s objects in [group=%s; non-namespced; labels=%v]",
+				logrus.Debugf("searching for %s objects in [group=%s; non-namespaced; labels=%v]",
 					resource.Name, groupVer, listOptions.LabelSelector,
 				)
 

starlark/capg_provider.go

@@ -0,0 +1,99 @@
+// Copyright (c) 2020 VMware, Inc. All Rights Reserved.
+// SPDX-License-Identifier: Apache-2.0
+
+package starlark
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/pkg/errors"
+	"github.com/vmware-tanzu/crash-diagnostics/k8s"
+	"github.com/vmware-tanzu/crash-diagnostics/provider"
+	"go.starlark.net/starlark"
+	"go.starlark.net/starlarkstruct"
+)
+
+// CapaProviderFn is a built-in starlark function that collects compute resources from a k8s cluster
+// Starlark format: capg_provider(kube_config=kube_config(), ssh_config=ssh_config()[workload_cluster=<name>, namespace=<namespace>, nodes=["foo", "bar], labels=["bar", "baz"]])
+func CapgProviderFn(thread *starlark.Thread, _ *starlark.Builtin, args starlark.Tuple, kwargs []starlark.Tuple) (starlark.Value, error) {
+
+	var (
+		workloadCluster, namespace string
+		names, labels              *starlark.List
+		sshConfig, mgmtKubeConfig  *starlarkstruct.Struct
+	)
+
+	err := starlark.UnpackArgs("capg_provider", args, kwargs,
+		"ssh_config", &sshConfig,
+		"mgmt_kube_config", &mgmtKubeConfig,
+		"workload_cluster?", &workloadCluster,
+		"namespace?", &namespace,
+		"labels?", &labels,
+		"nodes?", &names)
+	if err != nil {
+		return starlark.None, errors.Wrap(err, "failed to unpack input arguments")
+	}
+
+	ctx, ok := thread.Local(identifiers.scriptCtx).(context.Context)
+	if !ok || ctx == nil {
+		return starlark.None, fmt.Errorf("script context not found")
+	}
+
+	if sshConfig == nil || mgmtKubeConfig == nil {
+		return starlark.None, errors.New("capg_provider requires the name of the management cluster, the ssh configuration and the management cluster kubeconfig")
+	}
+
+	if mgmtKubeConfig == nil {
+		mgmtKubeConfig = thread.Local(identifiers.kubeCfg).(*starlarkstruct.Struct)
+	}
+	mgmtKubeConfigPath, err := getKubeConfigPathFromStruct(mgmtKubeConfig)
+	if err != nil {
+		return starlark.None, errors.Wrap(err, "failed to extract management kubeconfig")
+	}
+
+	// if workload cluster is not supplied, then the resources for the management cluster
+	// should be enumerated
+	clusterName := workloadCluster
+	if clusterName == "" {
+		config, err := k8s.LoadKubeCfg(mgmtKubeConfigPath)
+		if err != nil {
+			return starlark.None, errors.Wrap(err, "failed to load kube config")
+		}
+		clusterName, err = config.GetClusterName()
+		if err != nil {
+			return starlark.None, errors.Wrap(err, "cannot find cluster with name "+workloadCluster)
+		}
+	}
+
+	providerConfigPath, err := provider.KubeConfig(mgmtKubeConfigPath, clusterName, namespace)
+	if err != nil {
+		return starlark.None, err
+	}
+
+	nodeAddresses, err := k8s.GetNodeAddresses(ctx, providerConfigPath, toSlice(names), toSlice(labels))
+	if err != nil {
+		return starlark.None, errors.Wrap(err, "could not fetch host addresses")
+	}
+
+	// dictionary for capg provider struct
+	capgProviderDict := starlark.StringDict{
+		"kind":        starlark.String(identifiers.capgProvider),
+		"transport":   starlark.String("ssh"),
+		"kube_config": starlark.String(providerConfigPath),
+	}
+
+	// add node info to dictionary
+	var nodeIps []starlark.Value
+	for _, node := range nodeAddresses {
+		nodeIps = append(nodeIps, starlark.String(node))
+	}
+	capgProviderDict["hosts"] = starlark.NewList(nodeIps)
+
+	sshConfigDict := starlark.StringDict{}
+	sshConfig.ToStringDict(sshConfigDict)
+
+	capgProviderDict[identifiers.sshCfg] = starlarkstruct.FromStringDict(starlark.String(identifiers.sshCfg), sshConfigDict)
+
+	return starlarkstruct.FromStringDict(starlark.String(identifiers.capgProvider), capgProviderDict), nil
+}

starlark/kube_config.go

@@ -37,7 +37,7 @@ func KubeConfigFn(_ *starlark.Thread, _ *starlark.Builtin, args starlark.Tuple,
 		val := provider.Constructor()
 		if constructor, ok := val.(starlark.String); ok {
 			constStr := constructor.GoString()
-			if constStr != identifiers.capvProvider && constStr != identifiers.capaProvider {
+			if constStr != identifiers.capvProvider && constStr != identifiers.capgProvider && constStr != identifiers.capaProvider {
 				return starlark.None, errors.New("unknown capi provider")
 			}
 		}

starlark/starlark_exec.go

@@ -138,6 +138,7 @@ func newPredeclareds() starlark.StringDict {
 		identifiers.kubeNodesProvider: starlark.NewBuiltin(identifiers.kubeNodesProvider, KubeNodesProviderFn),
 		identifiers.capvProvider:      starlark.NewBuiltin(identifiers.capvProvider, CapvProviderFn),
 		identifiers.capaProvider:      starlark.NewBuiltin(identifiers.capaProvider, CapaProviderFn),
+		identifiers.capgProvider:      starlark.NewBuiltin(identifiers.capgProvider, CapgProviderFn),
 		identifiers.setDefaults:       starlark.NewBuiltin(identifiers.setDefaults, SetDefaultsFunc),
 		identifiers.log:               starlark.NewBuiltin(identifiers.log, logFunc),
 	}

starlark/support.go

@@ -54,6 +54,7 @@ var (
 		kubeNodesProvider string
 		capvProvider      string
 		capaProvider      string
+		capgProvider      string
 
 		sshAgent string
 	}{
@@ -91,6 +92,7 @@ var (
 		kubeNodesProvider: "kube_nodes_provider",
 		capvProvider:      "capv_provider",
 		capaProvider:      "capa_provider",
+		capgProvider:      "capg_provider",
 
 		sshAgent: "crashd_ssh_agent",
 	}