banzaicloud / banzai-cli Goto Github PK

List available organizations, let the user select one, and store the choice.

Don't list the hidden secrets by default, add --all option to keep displaying them.

Set up env vars to select the given cluster/org/token etc as the default for banzai-cli invocations inside the subshell.

Implement full oauth (local webserver + browser) login flow

Currently from CLI with (default format) we can't show the secret values

Retrieve and save a pipeline token and the base url from the instance created with banzai cp up.

by name or id

• Install AWS infrastructure for PKE
• Fetch PKE K8S credential from instance over SSH
• Automatic credential import, from env or aws credential( ~/.aws/credential)

Check things like below in the git repo of the current working directory:

• Every Spotguide file must be within a .banzaicloud directory
• Must contain a README.md file providing a simple description/documentation for the Spotguide
• Must contain an icon.svg file
• Must contain a CI/CD configuration template in pipeline.yaml
• Must contain Spotguide configuration (eg. name) and questionnaire in spotguide.yaml

Use a config value (default: ~/.banzai/pipeline/default) for the working dir of the installer instead of the current working directory.

• Add image verions flag, default (latest)
• Add image pull policy flag

Currently the cluster create command accepts a cluster descriptor on the standard input, but in that case the command runs in non-interactive mode.

Passing the file with a flag to the command would be nice with keeping the interactive mode.

Generate a set of tags based on the local username and hostname to provider_config of the ec2 installation. Handle it in the installer (and attach the uuid too).

https://gist.github.com/mickep76/ca29cca70f0b458aee4d

Add a 'deployment' subcommand for:

• list deployments that are running on a cluster
• get deployment - get the details of a deployment identified by release name
• add deployment - add deployment to a cluster
• delete deployment - delete a deployment identified by release name
• update deployment - update deployment identified by release name

Interactive flow:

• select type from available types
• ask for each required and optional field
• ask for tags
• offer editing

Non-interactively: read request from stdin

Let the user create a cluster based on a yaml file containing the request.

• Download specific kind version into .banzai/bin folder
• Create a cluster with kind

When the CLI interactively asks something SIGINT does not terminate the process, but skips the question.

for example: #9

Currently goreleaser config is not up to date with the latest spec.

Also, currently it's not integrated into the circle CI build.

In certain situations, the certificate of the Pipeline (and Dex) is signed by an unknown authority, in which case we need to enable passing the CA of the authority or skip TLS peer verification at all.

Give good defaults for creating an aws-pke cluster without touching the request

Let them select one "default" secret for each context.

Split from #5

It should be better if in this case the response shows that the Pipeline endpoint is nat a real endpoint.

bug:
./build/banzai login (docker-desktop/vault)
? Pipeline endpoint: https://not.a.real.endpoint
? Pipeline token: xxxxxxxxxxxx
2019/07/11 10:48:23 could not list organizations: 404 Not Found

expected:
./build/banzai login (docker-desktop/vault)
? Pipeline endpoint: https://not.a.real.endpoint
? Pipeline token: xxxxxxxxxxxx
2019/07/11 11:28:08 given URL is not a Pipeline API endpoint: 404 Not Found

Interactive flow:

• select secret
• ask for field to update
• ask for tags

Non-interactively: read request from stdin

This is a relatively small and new project, might worth testing go modules here.

Allow multiple configurations by setting the BANZAICONFIG environment variable.

Let the user list all clusters, and get the details of one of them by name/id.

bootstrap the cli project

Take care of providing a suitable helm command in the PATH of the shell executed by the banzai cluster shell command.

• Create a directory, add it to the beginning of $PATH, and place a helm wrapper there (this might be a symlink to the banzai-cli, or a single line shell-script that calls an internal banzai-cli endpoint).
• If the wrapped helm is called (or maybe even in advance in a background branch), ask the cluster for the version of tiller installed, and find in a cache or download the same version of helm.
• Create or use another directory for the cluster, set it up as $HELM-HOME, and set up the repositories that are provided by the Pipeline API.
• After that, exec helm with the original parameters.

Provide an option (--disable-helm-wrapper) to switch off this whole behavior.

Find user's aws-cli credentials in env vars and in ~/.aws/credentials' default section and use for installation/secret creation. Ask for approval. (Preparation step, will be used later by the AWS "demo" installer.)

Step required to SSH onto nodes (PKE cluster):

• Download the SSH secret (id_rsa)
• Grab the Node IP (public ip or via jump host)
• Parameter SSH command generate ssh config file with the private key and host (+ optionally ProxyJump)

fix problems, for example

• nodeport :80/443 ingress controller on aws/pke cp
• local storage

Let the user list the available secrets.

Add a banzai cluster import command that creates a BYOC cluster on Pipeline from the current Kubeconfig. In interactive mode it should ask for confirmation after printing some basic information like k get nodes, and offer running or skipping of the default posthooks (fixme?).

Save pipeline credentials and endpoint data to workspaces (~/.banzai/workspaces/*) instead of the general config file. Add a command to select default/current workspace, and a global command line option + env var to override the workspace selection. Merge in the config file in the current workspace.
banzai login should offer to create and select a new workspace if you are already logged in.

Give a command like banzai cluster p[ort-]f[orward] where the user can choose interactively a clustert, a namespace, a service/deploymentet/pod and an exposed pod, then optionally open it in a browser.

Ask pipeline user token, and validate/store it.

see #21

Contribute a generic solution for the selector of the survey project, to allow users to select a row from a table. Use it for selecting clusters, secrets etc.

Originally posted by @orymate in #79

offer aws and kind in addition to local kubecontext in banzai cp up

• if values file is provided, run non-interactively
• if no values file,
  • ask for provider, credentials (use local aws creds or exit with error)
  • write values file (omit credentials)

Summary after creating cluster is broken:

723 modern-go-application #(template: Status:1:2: executing "Status" at <.Status>: can't evaluate field Status in type client.CreateClusterResponse202)

thanks @sagikazarmark for reporting

Add cluster delete command.

Also, add dex based support with callback server. Related to #21

(Remaining task from #83)

Print something like banzai login --endpoint=.../pipeline