Skip to content

A CLI to nicely display resources/helm charts deployed in CAPI Cluster by Sveltos. Collect tech-support from managed Kubernetes clusters.

License

Notifications You must be signed in to change notification settings

projectsveltos/sveltosctl

Repository files navigation

CI Go Report Card Slack License Twitter Follow

sveltosctl

Please refere to sveltos documentation.

sveltosctl is the command line client for Sveltos. sveltosctl nicely displays resources and helm charts info in custer deployed using ClusterProfile/Profile. It also provides the ability to generate configuration snapshots and rollback system to a previously taken configuration snapshot.

It assumes:

  1. ClusterProfile/Profile is used to programmatically define which resources/helm charts need to be deployed in which CAPI Clusters;
  2. management cluster can be accessed

Note: sveltosctl can run as binary though it is advised to run it as pod in a management cluster to get access to all of its features.

Quick start

Run sveltosctl as a binary

If you decide to run it as a binary:

  1. make sure management cluster can be accessed;
  2. runmake build
  3. Use ./bin/sveltosctl --help to see help message

Run sveltosctl as a pod

If you decide to run it as a pod in the management cluster, YAML is in manifest subdirectory. This assumes you have already installed Sveltos.

kubectl create -f  https://raw.githubusercontent.com/projectsveltos/sveltosctl/main/manifest/manifest.yaml

Please keep in mind it requires a PersistentVolume. So modify this section accordingly before posting the YAML.

  volumeClaimTemplates:
  - metadata:
      name: snapshot
    spec:
      accessModes: [ "ReadWriteOnce" ]
      storageClassName: "standard"
      resources:
        requests:
          storage: 1Gi

Once the pod is running,

 kubectl exec -it -n projectsveltos sveltosctl-0   -- ./sveltosctl

You might also want to change the timezone of sveltosctl pod by using specific timezone config and hostPath volume to set specific timezone. Currently:

  volumes:
  - hostPath:
      path: /usr/share/zoneinfo/America/Los_Angeles
      type: File
    name: tz-config

Display deployed add-ons

show addons can be used to display list of Kubernetes addons (resources/helm) releases deployed in CAPI clusters. Displayed information contains:

  1. the CAPI Cluster in the form /
  2. resource/helm chart information
  3. list of ClusterProfiles/Profiles currently (at the time the command is run) having resource/helm release deployed in the CAPI cluster.
./bin/sveltosctl show addons
+-------------------------------------+---------------+-----------+----------------+---------+-------------------------------+------------------+
|               CLUSTER               | RESOURCE TYPE | NAMESPACE |      NAME      | VERSION |             TIME              | CLUSTER PROFILE |
+-------------------------------------+---------------+-----------+----------------+---------+-------------------------------+------------------+
| default/sveltos-management-workload | helm chart    | kyverno   | kyverno-latest | v2.5.0  | 2022-09-30 11:48:45 -0700 PDT | clusterfeature1  |
| default/sveltos-management-workload | :Pod          | default   | nginx          | N/A     | 2022-09-30 13:41:05 -0700 PDT | clusterfeature2  |
+-------------------------------------+---------------+-----------+----------------+---------+-------------------------------+------------------+

show addons command has some argurments which allow filtering by:

  1. clusters' namespace
  2. clusters' name
  3. ClusterProfile/Profile
./bin/sveltosctl show addons --help
Usage:
  sveltosctl show addons [options] [--namespace=<name>] [--cluster=<name>] [--profile=<name>] [--verbose]

     --namespace=<name>     Show addons deployed in clusters in this namespace. If not specified all namespaces are considered.
     --cluster=<name>       Show addons deployed in cluster with name. If not specified all cluster names are considered.
     --profile=<kind/name>  Show addons deployed because of this clusterprofile/profile. If not specified all clusterprofiles/profiles are considered.

Register a cluster

If there is kubeconfig with multiple contexts, the option fleet-cluster-context allows to specify the context for the cluster to be managed.

So with default context pointing to the management cluster, following command will:

  1. create a ServiceAccount in the managed cluster (using cluster-1 context)
  2. grant this ServiceAccount cluster-admin permission
  3. create a TokenRequest for such account and a Kubeconfig with bearer token from the TokenRequest
  4. create a SveltosCluster in the management cluster (so using default context) and a Secret with kubeconfig generated in the step above
sveltosctl register cluster --namespace=gcp --cluster=cluster-1 --fleet-cluster-context=cluster-1 --labels=k1=v1,k2=v2

Display information about resources in managed cluster

show resources looks at all the HealthCheckReport instances and display information about those. Defining ClusterHealthCheck/HealthCheck you can define which information to collect from which managed clusters. Please see documentation

For instance:

+-------------------------------------+--------------------------+----------------+-------------------------+----------------------------+
|               CLUSTER               |           GVK            |   NAMESPACE    |          NAME           |          MESSAGE           |
+-------------------------------------+--------------------------+----------------+-------------------------+----------------------------+
| default/sveltos-management-workload | apps/v1, Kind=Deployment | kube-system    | calico-kube-controllers | All replicas 1 are healthy |
|                                     |                          | kube-system    | coredns                 | All replicas 2 are healthy |
|                                     |                          | projectsveltos | sveltos-agent-manager   | All replicas 1 are healthy |
| gke/production                      | apps/v1, Kind=Deployment | kube-system    | calico-kube-controllers | All replicas 1 are healthy |
|                                     |                          | kube-system    | coredns                 | All replicas 2 are healthy |
|                                     |                          | projectsveltos | sveltos-agent-manager   | All replicas 1 are healthy |
+-------------------------------------+--------------------------+----------------+-------------------------+----------------------------+

Display usage

show usage displays following information:

  1. which CAPI clusters are currently a match for a ClusterProfile
  2. for ConfigMap/Secret referenced by at least by ClusterProfile, in which CAPI clusters their content is currently deployed.

Such information is useful to see what CAPI clusters would be affected by a change before making such a change.

./bin/sveltosctl show usage
----------------+--------------------+----------------------------+-------------------------------------+
| RESOURCE KIND  | RESOURCE NAMESPACE |       RESOURCE NAME        |              CLUSTERS               |
+----------------+--------------------+----------------------------+-------------------------------------+
| ClusterProfile |                    | mgianluc                   | default/sveltos-management-workload |
| ConfigMap      | default            | kyverno-disallow-gateway-2 | default/sveltos-management-workload |
+----------------+--------------------+----------------------------+-------------------------------------+

Multi-tenancy: display admin permissions

show admin-rbac can be used to display permissions granted to tenant admins in each managed clusters. If we have two clusters, a ClusterAPI powered one and a SveltosCluster, both matching label selector env=internal and we post RoleRequests, we get:

./bin/sveltosctl show admin-rbac       
+---------------------------------------------+-------+----------------+------------+-----------+----------------+-------+
|                   CLUSTER                   | ADMIN |   NAMESPACE    | API GROUPS | RESOURCES | RESOURCE NAMES | VERBS |
+---------------------------------------------+-------+----------------+------------+-----------+----------------+-------+
| Cluster:default/sveltos-management-workload | eng   | build          | *          | *         | *              | *     |
| Cluster:default/sveltos-management-workload | eng   | ci-cd          | *          | *         | *              | *     |
| Cluster:default/sveltos-management-workload | hr    | human-resource | *          | *         | *              | *     |
| SveltosCluster:gke/prod-cluster             | eng   | build          | *          | *         | *              | *     |
| SveltosCluster:gke/prod-cluster             | eng   | ci-cd          | *          | *         | *              | *     |
| SveltosCluster:gke/prod-cluster             | hr    | human-resource | *          | *         | *              | *     |
+---------------------------------------------+-------+----------------+------------+-----------+----------------+-------+

Log severity settings

log-level used to display and change log severity in Sveltos PODs without restarting them.

Following for instance change log severity for the Classifier POD to debug

./bin/sveltosctl log-level set --component=Classifier --debug

Show can be used to display current log severity settings

./bin/sveltosctl log-level show                              
+------------+---------------+
| COMPONENT  |   VERBOSITY   |
+------------+---------------+
| Classifier | LogLevelDebug |

Display outcome of ClusterProfile in DryRun mode

See video. A ClusterProfile can be set in DryRun mode. While in DryRun mode, nothing gets deployed/withdrawn to/from matching CAPI clusters. A report is instead generated listing what would happen if ClusterProfile sync mode would be changed from DryRun to Continuous.

Here is an example of outcome

./bin/sveltosctl show dryrun
+-------------------------------------+--------------------------+-----------+----------------+-----------+--------------------------------+------------------+
|               CLUSTER               |      RESOURCE TYPE       | NAMESPACE |      NAME      |  ACTION   |            MESSAGE             | CLUSTER PROFILE |
+-------------------------------------+--------------------------+-----------+----------------+-----------+--------------------------------+------------------+
| default/sveltos-management-workload | helm release             | kyverno   | kyverno-latest | Install   |                                | dryrun           |
| default/sveltos-management-workload | helm release             | nginx     | nginx-latest   | Install   |                                | dryrun           |
| default/sveltos-management-workload | :Pod                     | default   | nginx          | No Action | Object already deployed.       | dryrun           |
|                                     |                          |           |                |           | And policy referenced by       |                  |
|                                     |                          |           |                |           | ClusterProfile has not changed |                  |
|                                     |                          |           |                |           | since last deployment.         |                  |
| default/sveltos-management-workload | kyverno.io:ClusterPolicy |           | no-gateway     | Create    |                                | dryrun           |
+-------------------------------------+--------------------------+-----------+----------------+-----------+--------------------------------+------------------+

show dryrun command has some argurments which allow filtering by:

  1. clusters' namespace
  2. clusters' name
  3. ClusterProfile
./bin/sveltosctl show dryrun --help  
Usage:
  sveltosctl show dryrun [options] [--namespace=<name>] [--cluster=<name>] [--profile=<name>] [--verbose]

     --namespace=<name> Show which Kubernetes addons would change in clusters in this namespace. If not specified all namespaces are considered.
     --cluster=<name>   Show which Kubernetes addons would change in cluster with name. If not specified all cluster names are considered.
     --profile=<name>   Show which Kubernetes addons would change because of this clusterprofile/profile. If not specified all clusterprofiles/profiles are considered.

Techsupport

When running sveltosctl as pod in the management cluster, it can take collect techsupports (both logs and resources).

Define a Techsupport instance, following for instance will collect a techsupport every hour, collecting:

  1. logs for all pods in kube-system namespace (last 10 minutes,i.e, 600 seconds, of logs);
  2. All Secrets and Deployments

from all managed clusters matching cluster selectors env=fv

apiVersion: utils.projectsveltos.io/v1beta1
kind: Techsupport
metadata:
 name: hourly
spec:
 clusterSelector:
   matchLabels:
     env: fv
 schedule: “00 * * * *”
 storage: /techsupport
 logs:
 - namespace: kube-system
   sinceSeconds: 600
 resources:
 - group: “”
   version: v1
   kind: Secret
 - group: “”
   version: v1
   kind: Deployment

where field schedule is defined in Cron format.

list

techsupport list can be used to display all collected techsupports:

kubectl exec -it -n projectsveltos sveltosctl-0 -- ./sveltosctl techsupport list --techsupport=hourly 
+--------------------+---------------------+
| TECHSUPPORT POLICY |        DATE         |
+--------------------+---------------------+
| hourly             | 2022-10-10:22:00:00 |
| hourly             | 2022-10-10:23:00:00 |
+--------------------+---------------------+

Snapshot

When running sveltosctl as pod in the management cluster, it can take configuration snapshot.

A snapshot allows an administrator to perform the following tasks:

  1. Live snapshots of the running configuration deployed by ClusterProfiles in each CAPI cluster;
  2. Recurring snapshots;
  3. Versioned storage of the configuration
  4. Full viewing of any snapshot configuration including the differences between snapshots
  5. Rollback to any previous configuration snapshot.

Define a Snapshot instance, following for instance will take a snaphost every hour.

apiVersion: utils.projectsveltos.io/v1beta1
kind: Snapshot
metadata:
  name: hourly
spec:
  schedule: "00 * * * *"
  storage: /snapshot

where field schedule is defined in Cron format.

The configuration snapshots consist of text files containing:

  1. ClusterProfiles;
  2. All ConfigMaps/Secrets referenced by at least one ClusterProfile;
  3. CAPI Cluster labels;
  4. few other internal config.projectsveltos.io CRD instances.

The snapshot contains the configuration at the time of the snapshot stored. Each snapshot is stored with a version identifier. The version identifier is automatically generated by concatenating the date with the time of the snapshot.

list

snapshot list can be used to display all available snapshots:

kubectl exec -it -n projectsveltos sveltosctl-0 -- ./sveltosctl snapshot list --snapshot=hourly 
+-----------------+---------------------+
| SNAPSHOT POLICY |        DATE         |
+-----------------+---------------------+
| hourly          | 2022-10-10:22:00:00 |
| hourly          | 2022-10-10:23:00:00 |
+-----------------+---------------------+

diff

snapshot diff can be used to display all the configuration changes between two snapshots:

kubectl exec -it -n projectsveltos sveltosctl-0 -- ./sveltosctl snapshot diff --snapshot=hourly  --from-sample=2022-10-10:22:00:00 --to-sample=2022-10-10:23:00:00 
+-------------------------------------+--------------------------+-----------+----------------+----------+------------------------------------+
|               CLUSTER               |      RESOURCE TYPE       | NAMESPACE |      NAME      |  ACTION  |              MESSAGE               |
+-------------------------------------+--------------------------+-----------+----------------+----------+------------------------------------+
| default/sveltos-management-workload | helm release             | mysql     | mysql          | added    |                                    |
| default/sveltos-management-workload | helm release             | nginx     | nginx-latest   | added    |                                    |
| default/sveltos-management-workload | helm release             | kyverno   | kyverno-latest | modified | To version: v2.5.0 From            |
|                                     |                          |           |                |          | version v2.5.3                     |
| default/sveltos-management-workload | /Pod                     | default   | nginx          | added    |                                    |
| default/sveltos-management-workload | kyverno.io/ClusterPolicy |           | no-gateway     | modified | To see diff compare ConfigMap      |
|                                     |                          |           |                |          | default/kyverno-disallow-gateway-2 |
|                                     |                          |           |                |          | in the from folderwith ConfigMap   |
|                                     |                          |           |                |          | default/kyverno-disallow-gateway-2 |
|                                     |                          |           |                |          | in the to folder                   |
+-------------------------------------+--------------------------+-----------+----------------+----------+------------------------------------+

To see Sveltos CLI for snapshot in action, have a look at this video

rollback

Rollback is when a previous configuration snapshot is used to replace the current configuration deployed by ClusterProfiles. This can be done on the granularity of:.

  1. namespace: Rollbacks only ConfigMaps/Secrets and Cluster labels in this namespace. If not specified all namespaces are considered;
  2. cluster: Rollback only labels for cluster with this name. If not specified all cluster's labels are updated;
  3. clusterprofile: Rollback only clusterprofile with this name. If not specified all clusterprofiles are updated

When all of the configuration files for a particular version are used to replace the current configuration, this is referred to as a full rollback.

Following for instance will bring system back to the state it had at 22:00

kubectl exec -it -n projectsveltos sveltosctl-0 -- ./sveltosctl snapshot rollback --snapshot=hourly  --sample=2022-10-10:22:00:00

To see Sveltos CLI for snapshot in action, have a look at this video

Admin RBACs

snapshot show admin-rbac can be used to display admin's RBACs per cluster:

./bin/sveltosctl show admin-rbac                                                                           
+---------------------------------------------+-------------+-----------+------------+-----------+----------------+----------------+
|                   CLUSTER                   |  ADMIN      | NAMESPACE | API GROUPS | RESOURCES | RESOURCE NAMES |     VERBS      |
+---------------------------------------------+-------------+-----------+------------+-----------+----------------+----------------+
| Cluster:default/sveltos-management-workload | eng/devops  | default   |            | pods      | pods           | get,watch,list |
+---------------------------------------------+-------------+-----------+------------+-----------+----------------+----------------+

Contributing

❤️ Your contributions are always welcome! If you want to contribute, have questions, noticed any bug or want to get the latest project news, you can connect with us in the following ways:

  1. Open a bug/feature enhancement on github contributions welcome
  2. Chat with us on the Slack in the #projectsveltos channel Slack
  3. Contact Us

License

Copyright 2022.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.