Interface for the ansible  `kubernetes.core.k8s` module[1].
 This accepts only module args, but there are a few ways to pass them.  
 See the docs in 'ansible.adhoc/<module>' for discussion of examples.

 * `[1]`: https://docs.ansible.com/ansible/latest/collections/kubernetes/core/k8s_module.html

 Checks whether the cluster is available.  
 This just returns the exit status of cluster-info, and not 
 whether pods are all in a ready state. For that, see 'k8s.wait'

 EXAMPLE: 
   ./k8s.mk k8s.cluster.ready

 EXAMPLE: ( in a loop )
   ./k8s.mk flux.loop.until/k8s.cluster.ready

 REFS:
   * `[1]`: https://github.com/alecjacobs5401/kubectl-sick-pods

 Waits until all pods in all namespaces are ready.  (Alias for 'k8s.namespace.wait/all')

 Shows deployments for the given namespace
 Not for parsing; this info is human friendly, and output to logging channel.

 USAGE:
   k8s.get.svc/<ns>/<name>/<filter>
   k8s.get.svc/ab-testing/coin-service/.spec.clusterIP

 Alias for k8s.graph/all/pods.  This returns dot-format data.

 Alias for 'k8s.graph.tui/all/pods'.  This prints a visual graph on the terminal.

 Previews topology for a given kubernetes <namespace>/<kind> in a way that is terminal-friendly.

 This is a human-friendly way to visualize progress or changes, because it supports 
 very large input data from complex deployments with lots of services/pods, either in 
 one namespace or across the whole cluster. To do that, it has to throw away some 
 information compared with raw kubectl output, and node labels on the graph aren't 
 visible.

 This is basically a pipeline from graphs in dot format, generated by kubectl-graph, 
 then passed through some image-magick transformations, and then pushed into 
 the 'chafa' tool for generating ASCII-art from images.

 USAGE: (same as k8s.graph)
   ./k8s.mk k8s.graph.tui/<namespace>/<kind>
   ./k8s.mk k8s.graph.tui/<namespace>/<kind>,<outfile>

 Previews topology for a given kubernetes <namespace>/<kind> in a way that is terminal-friendly.

 This is a human-friendly way to visualize progress or changes, because it supports 
 very large input data from complex deployments with lots of services/pods, either in 
 one namespace or across the whole cluster. To do that, it has to throw away some 
 information compared with raw kubectl output, and node labels on the graph aren't 
 visible.

 This is basically a pipeline from graphs in dot format, generated by kubectl-graph, 
 then passed through some image-magick transformations, and then pushed into 
 the 'chafa' tool for generating ASCII-art from images.

 USAGE: (same as k8s.graph)
   ./k8s.mk k8s.graph.tui/<namespace>/<kind>
   ./k8s.mk k8s.graph.tui/<namespace>/<kind>,<outfile>

 Graphs resources under the given namespace, for the given kind, in dot-format.
 Pipe Friendly: results are always dot files.  Caller should handle any errors.

 This requires the krew plugin "graph" (installed by default with k8s-tools.yml).

 USAGE: 
     ./k8s.mk k8s.graph/<namespace>/<kind>
     ./k8s.mk k8s.graph/<namespace>/<kind>,<outfile>

 Argument for 'kind' must be provided, but may be "all".  

 Graphs resources under the given namespace, for the given kind, in dot-format.
 Pipe Friendly: results are always dot files.  Caller should handle any errors.

 This requires the krew plugin "graph" (installed by default with k8s-tools.yml).

 USAGE: 
     ./k8s.mk k8s.graph/<namespace>/<kind>
     ./k8s.mk k8s.graph/<namespace>/<kind>,<outfile>

 Argument for 'kind' must be provided, but may be "all".  

 Shows targets for just the 'k8s' namespace.

 Wait for all jobs in all namespaces

 Wait for all jobs in the given namespace

 Wait for all jobs in the given namespace

 Returns `kubectx` result

 Context-manager.  Activates the given namespace, creating it first if necessary.

 NB: This modifies state in the kubeconfig, so that it can effect contexts 
 outside of the current process, therefore this is not thread-safe.

 USAGE: 
    ./k8s.mk k8s.kubens.create/<namespace>

 Context-manager.  Activates the given namespace.
 NB: This modifies state in the kubeconfig, so that it can effect contexts 
 outside of the current process, and therefore this is not thread-safe.

 USAGE:  
   ./k8s.mk k8s.kubens/<namespace>

 Appends the given label to the given namespace.

 USAGE: 
   key=<key> val=<val> ./k8s.mk k8s.namespace.label/<namespace>
   ./k8s.mk k8s.namespace.label/<namespace>/<key>/<val>

 Runs a separate purge for every matching namespace.
 NB: This isn't likely to clean everything, see the docs for your dependencies.

 USAGE: 
    ./k8s.mk k8s.namespace.purge.by.prefix/<prefix>

 Wipes everything inside the given namespace.

 USAGE: 
    k8s.namespace.purge/<namespace>

 Wait for all pods in all namespaces

 Waits for every pod/job in the given namespace to be ready.

 This uses only kubectl/jq to loop on pod-status, but assumes that 
 the krew-plugin 'sick-pods'[1] is available for formatting the 
 user-message.  See `k8s.wait` for an alias that waits on all pods.

 NB: If the parameter is "all" then this uses --all-namespaces

 USAGE: 
   ./k8s.mk k8s.namespace.wait/<namespace>

 REFS:
   * `[1]`: https://github.com/alecjacobs5401/kubectl-sick-pods

 Waits for every pod/job in the given namespace to be ready.

 This uses only kubectl/jq to loop on pod-status, but assumes that 
 the krew-plugin 'sick-pods'[1] is available for formatting the 
 user-message.  See `k8s.wait` for an alias that waits on all pods.

 NB: If the parameter is "all" then this uses --all-namespaces

 USAGE: 
   ./k8s.mk k8s.namespace.wait/<namespace>

 REFS:
   * `[1]`: https://github.com/alecjacobs5401/kubectl-sick-pods

 Context-manager.  Activates the given namespace.

 NB: This modifies state in the kubeconfig, so that it can effect contexts 
 outside of the current process, therefore this is not thread-safe.

 USAGE:  
     ./k8s.mk k8s.namespace/<namespace>

 This drops into a debugging shell for the named pod using `kubectl exec`,
 plus a streaming version of the same which allows for working with pipes.

 NB: This target assumes that the named pod actually exists.  If you want
 an idempotent create-operation for such a pod.. see `k8s.test_harness`.

 NB: This target attempts to be "universal", so that it may run from the 
 docker host or inside the `k8s:base` container.  This works by detecting 
 aspects of the caller context.  In general, it tries to use k8s-tools.yml 
 when that makes sense and if it's present, falling back to kubectl.

 If `user` is set as an environment variable, this tries to use `su` to overcome 
 the fact that `kubectl exec --user` does no work like `docker exec --user`.
 See also: https://github.com/kubernetes/kubernetes/issues/30656

 USAGE: Interactive shell in pod:
   ./k8s.mk k8s.shell/<namespace>/<pod_name>

 Returns information about pods in all namespaces.  
 Not for parsing; this info is human friendly, and output to logging channel.

 Returns information about pods in all namespaces.  
 Not for parsing; this info is human friendly, and output to logging channel.

 Wait for all pods in all namespaces

 Wait for all pods in the given namespace

 Wait for all pods in the given namespace

 Shows pods for the given namespace, excluding ones with "completed" status.
 Not for parsing; this info is human friendly, and output to logging channel.

 Shows pods for the given namespace, excluding ones with "completed" status.
 Not for parsing; this info is human friendly, and output to logging channel.

 Checks whether the cluster is available.  
 This just returns the exit status of cluster-info, and not 
 whether pods are all in a ready state. For that, see 'k8s.wait'

 EXAMPLE: 
   ./k8s.mk k8s.cluster.ready

 EXAMPLE: ( in a loop )
   ./k8s.mk flux.loop.until/k8s.cluster.ready

 REFS:
   * `[1]`: https://github.com/alecjacobs5401/kubectl-sick-pods

 Shows cluster, kube, and docker environment variables

 Shows all namespaces for the current cluster.
 (This is just the output of `kubens` with no arguments)

 Shows services for the given namespace
 Not for parsing; this info is human friendly, and output to logging channel.

 Starts a test-pod with a random name in the given namespace, optionally blocking until it's ready.

 USAGE: 
    `k8s.test_harness.random`

 Starts a test-pod in the given namespace, optionally blocking 
 until it's ready. If `img` is provided in environment it will 
 be used, otherwise defaults to 'IMG_ALPINE_K8S'.

 USAGE: 
    `k8s.test_harness/<namespace>/<pod_name>`

 Waits until all pods in all namespaces are ready.  (Alias for 'k8s.namespace.wait/all')

 An interface into the named ansible module.  Just pass the module-arguments.  
 Like adhoc ansible, this allows you to call a task without a playbook.
 This actually generates a playbook JIT though, which makes things 
 more flexible.

 USAGE:
   echo '<arg_json>' | ./compose.mk ansible.adhoc/<ansible_module_name>

 An interface into the named ansible module.  Just pass the module-arguments.  
 Like adhoc ansible, this allows you to call a task without a playbook.
 This actually generates a playbook JIT though, which makes things 
 more flexible.

 USAGE:
   echo '<arg_json>' | ./compose.mk ansible.adhoc/<ansible_module_name>

 An interface into the named ansible module.  Just pass the module-arguments.  
 Like adhoc ansible, this allows you to call a task without a playbook.
 This actually generates a playbook JIT though, which makes things 
 more flexible.

 USAGE:
   echo '<arg_json>' | ./compose.mk ansible.adhoc/<ansible_module_name>

 An interface into the named ansible module.  Just pass the module-arguments.  
 Like adhoc ansible, this allows you to call a task without a playbook.
 This actually generates a playbook JIT though, which makes things 
 more flexible.

 USAGE:
   echo '<arg_json>' | ./compose.mk ansible.adhoc/<ansible_module_name>

 Interface for the ansible  block-in-file module[1].
 This accepts only module args, but there are several ways to pass them.  
 See the docs in ansible.adhoc/<module> for discussion of examples.

 USAGE:
   echo <json_data> | ./k8s.mk ansible.blockinfile
   ./k8s.mk jb <key1>=<val1> <keyn>=<valn> | ./k8s.mk ansible.blockinfile
   data="<key1>=<val1> <keyn>=<valn>" ./k8s.mk ansible.blockinfile

 EXAMPLE:
   path=.gitignore block=".flux.stage.*" | ./k8s.mk ansible.blockinfile

 * `[1]`: https://docs.ansible.com/ansible/latest/collections/ansible/builtin/blockinfile_module.html

 Interface for the ansible  helm module[1].
 This accepts only module args, but there are a few ways to pass them.  
 See the docs in 'ansible.adhoc/<module>' for discussion of examples.

 * `[1]`: https://docs.ansible.com/ansible/latest/collections/kubernetes/core/helm_module.html

 Interface for the ansible  `kubernetes.core.k8s` module[1].
 This accepts only module args, but there are a few ways to pass them.  
 See the docs in 'ansible.adhoc/<module>' for discussion of examples.

 * `[1]`: https://docs.ansible.com/ansible/latest/collections/kubernetes/core/k8s_module.html

 Runs the input-stream as an ansible playbook.
 This calls ansible in a way that ensures all output is JSON.

 USAGE:
   cat <playbook> | ./compose.mk ansible.run

 Like `ansible.run.tasks`, but reads content from given define-block

 Runs ansible tasks defined by stdin.

 Runs the given playbook file.
 This calls ansible in a way that ensures all output is JSON.

 USAGE: 
   ./k8s.mk ansible.run/<path>

 List for the default namespace (i.e. "argo")

 Returns the results of 'argo list' for the current argo context.

 Returns the results of 'argo list' for the current argo context.

 USAGE:
  cat manifest.yml | k8s.mk argo.submit.stdin

 USAGE:
  url="http://../manifest.yml" ./k8s.mk argo.submit.url

 Pass the given path to `argo submit`.
 This fills in CLI args for `--wait` and `--log` based 
 on whether corresponding environment variables are set.
 USAGE:
  ./k8s.mk argo.submit/manifest.yml 

 Succeeds only when the given environment exists 
 for the currently active namespace.  No output.

 Creates the named fission environment if it doesn't already exist.
 Desired namespace should already be activated and `img` must be 
 set in the environment.

 Newline-separated list of all the fission environments available.

 USAGE:
   fission_env=<env_name> 
     fission_code=<path> ./k8s.mk 
      fission.function.create/<fxn_name>

 Describe fission status for the current namespace.
 This only writes to stderr, combining the following
 commands and failing if any one of them fails:
   fission check
   fission env list
   fission function list

 Lists only the agents for the given cluster.

 Creates a k3d cluster with the given name, using the given configuration.
 This supports most of the usual command-line options,
 but they must be passed as variables.

 USAGE:
   k3d_servers=.. k3d_agents=.. 
   k3d_port=.. k3d_api_port=.. 
     ./k8s.mk k3d.cluster.create/<cluster_name>

 Idempotent version of k3d cluster delete 

 USAGE:
   ./k8s.mk k3d.cluster.delete/<cluster_name>

 Succeeds iff cluster exists.

 Create a k3d cluster if it does not already exist.

 Returns cluster-names, newline delimited.  Available as a macro.
 For more details in a human friendly summary, see `k3d.stat`.

 USAGE:  
   ./k8s.mk k3d.cluster.list

 Starts a 4-pane TUI dashboard, using the commander layout.  
 This opens 'lazydocker', 'ktop', and other widgets that are convenient for working with k3d.

 USAGE:  
   KUBECONFIG=.. ./k8s.mk k3d.commander/<namespace>

 Sets the given k3d cluster as the active kubectx.
 NB: This appends the usual "k3d-" prefix

 Shows targets for just the 'k3d' namespace.

 Returns cluster-names, newline delimited.  Available as a macro.
 For more details in a human friendly summary, see `k3d.stat`.

 USAGE:  
   ./k8s.mk k3d.cluster.list

 Non-graceful stop for everything that is k3d related. 

 USAGE:  
   ./k8s.mk k3d.panic

 Deletes all known k3d clusters.

 Show status for k3d.

 Lists only the agents for the given cluster.

 Creates a k3d cluster with the given name, using the given configuration.
 This supports most of the usual command-line options,
 but they must be passed as variables.

 USAGE:
   k3d_servers=.. k3d_agents=.. 
   k3d_port=.. k3d_api_port=.. 
     ./k8s.mk k3d.cluster.create/<cluster_name>

 Idempotent version of k3d cluster delete 

 USAGE:
   ./k8s.mk k3d.cluster.delete/<cluster_name>

 Succeeds iff cluster exists.

 Create a k3d cluster if it does not already exist.

 Returns cluster-names, newline delimited.  Available as a macro.
 For more details in a human friendly summary, see `k3d.stat`.

 USAGE:  
   ./k8s.mk k3d.cluster.list

 Starts a 4-pane TUI dashboard, using the commander layout.  
 This opens 'lazydocker', 'ktop', and other widgets that are convenient for working with k3d.

 USAGE:  
   KUBECONFIG=.. ./k8s.mk k3d.commander/<namespace>

 Sets the given k3d cluster as the active kubectx.
 NB: This appends the usual "k3d-" prefix

 Shows targets for just the 'k3d' namespace.

 Returns cluster-names, newline delimited.  Available as a macro.
 For more details in a human friendly summary, see `k3d.stat`.

 USAGE:  
   ./k8s.mk k3d.cluster.list

 Non-graceful stop for everything that is k3d related. 

 USAGE:  
   ./k8s.mk k3d.panic

 Deletes all known k3d clusters.

 Show status for k3d.

 This wraps the given target, using the given given kubefwd arguments.
 See `flux.context_manager` docs for more details.

 Roughly equivalent:
   kubefwd.ctx/1st,2nd => kubefwd.start/2nd 1st kubefwd.stop/2nd

 Runs kubefwd for the given namespace, finding and forwarding ports/DNS for the given 
 service, or for all services. This is idempotent, and implicitly stops port-forwarding 
 if it is running, then restarts it. 

 NB: This target should only run from the docker host (not from the kubefwd container),  
 and it assumes k8s-tools.yml is present with that filename. Simple port-mapping and 
 filtering by name is supported; other usage with selectors/labels/reservations/etc 
 should just invoke kubefwd directly.

 USAGE: 
   ./k8s.mk kubefwd/<namespace>
   ./k8s.mk kubefwd/<namespace>/<svc_name>
     kubefwd_mapping="8080:80" ./k8s.mk kubefwd/<namespace> 
   kubefwd_mapping="8080:80" ./k8s.mk kubefwd/<namespace>/<svc_name>

 Stops the named kubefwd instance.  Mostly for internal usage, 
 usually you want 'kubefwd.start' or 'kubefwd.panic' instead.

 USAGE:
    ./k8s.mk kubefwd.stop/<namespace>/<svc_name>

 Shows targets for just the 'kubefwd' namespace.

 Non-graceful stop for everything that is kubefwd related.

 Emergency use only; this can clutter up your /etc/hosts
 file as kubefwd may not get a chance to clean things up.

 USAGE:  
   ./k8s.mk kubefwd.panic

 Display status info for all kubefwd instances that are running

 Returns JSON for the currently available helm repositories.

 Shows the status of things that were installed with helm
 This is human friendly output on stderr, use `helm.list` for a JSON dump.