prvng_core/nulib/main_provisioning/commands/infrastructure.nu
Jesús Pérez 894046ef5a
feat(core): three-layer DAG, unified component arch, commands-registry cache, Nushell 0.112.2 migration
- DAG architecture: `dag show/validate/export` (nulib/main_provisioning/dag.nu),
    config loader (lib_provisioning/config/loader/dag.nu), taskserv dag-executor.
    Backed by schemas/lib/dag/*.ncl; orchestrator emits NATS events via
    WorkspaceComposition::into_workflow. See ADR-020, ADR-021.
  - Unified Component Architecture: components/mod.nu, main_provisioning/
    {components,workflow,extensions,ontoref-queries}.nu. Full workflow engine with
    topological sort and NATS subject emission. Blocks A-H complete (libre-daoshi).
  - Commands-registry: nulib/commands-registry.ncl (Nickel source, 314 lines) +
    JSON cache at ~/.cache/provisioning/commands-registry.json rebuilt on source
    change. cli/provisioning fast-path alias expansion avoids cold Nu startup.
    ADDING_COMMANDS.md documents new-command workflow.
  - Platform service manager: service-manager.nu (+573), startup.nu (+611),
    service-check.nu (+255); autostart/bootstrap/health/target refactored.
  - Nushell 0.112.2 migration: removed all try/catch and bash redirections;
    external commands prefixed with ^; type signatures enforced. Driven by
    scripts/refactor-try-catch{,-simplified}.nu.
  - TTY stack: removed shlib/*-tty.sh; replaced by cli/tty-dispatch.sh,
    tty-filter.sh, tty-commands.conf.
  - New domain modules: images/ (golden image lifecycle), workspace/{state,sync}.nu,
    main_provisioning/{bootstrap,cluster-deploy,fip,state}.nu, commands/{state,
    build,integrations/auth,utilities/alias}.nu, platform.nu expanded (+874).
  - Config loader overhaul: loader/core.nu slimmed (-759), cache/core.nu
    refactored (-454), removed legacy loaders/file_loader.nu (-330).
  - Thirteen new provisioning-<domain>.nu top-level modules for bash dispatcher.
  - Tests: test_workspace_state.nu (+351); updates to test_oci_registry,
    test_services.
  - README + CHANGELOG updated.
2026-04-17 04:27:33 +01:00

672 lines
25 KiB
Text

# Infrastructure Command Handlers
# Handles: server, taskserv, cluster, infra commands
use ../flags.nu *
# REMOVED: use ../../lib_provisioning * - causes circular import
use ../../lib_provisioning/plugins/auth.nu *
# Pre-load server module to preserve plugin context (tera, auth, kms, etc.)
# This is needed so template rendering and other plugin operations work
# in the same Nushell process
use ../../servers/create.nu *
# Helper to run module commands
# Modules are pre-loaded above to preserve plugin context
def run_module [
args: string
module: string
subcommand?: string # Optional explicit subcommand (for create operations)
--exec
] {
# Convert args string to list by splitting on spaces
let args_list = if ($args | is-not-empty) {
$args | split row " " | where {|x| ($x | str trim | is-not-empty) }
} else {
[]
}
# Call the appropriate module's main function
# Server module is pre-loaded above, so plugins (tera, auth, kms, etc.) are in scope
match $module {
"server" => {
# For server: call the "main create" function directly from the already-loaded servers/create.nu
# This preserves the tera plugin context in the same process
# If subcommand is explicitly provided (from handle_server), use it
# Otherwise, extract from args
let actual_subcommand = if ($subcommand | is-not-empty) {
$subcommand
} else {
let op_list = ($args | split row " " | where { |x| ($x | is-not-empty) })
if ($op_list | length) > 0 { $op_list | first } else { "help" }
}
# For now, only handle "create" directly. For others, use -mod
match $actual_subcommand {
"create" | "c" | "list" | "l" => {
# The servers/create.nu and servers/list.nu are loaded modules
# Call "main create" or "main list" function directly with the arguments
# This preserves context (env vars, plugins, etc.) in the same process
let use_debug = if ($env.PROVISIONING_DEBUG? | default false) { "-x" } else { "" }
let cmd_args = [-mod, "server", $actual_subcommand, ...$args_list]
exec $"($env.PROVISIONING_NAME)" $use_debug ...$cmd_args
}
_ => {
# For other operations (delete, ssh, price, status, etc.), use -mod with explicit subcommand
let use_debug = if ($env.PROVISIONING_DEBUG? | default false) { "-x" } else { "" }
let cmd_args = [-mod, "server", $actual_subcommand, ...$args_list]
exec $"($env.PROVISIONING_NAME)" $use_debug ...$cmd_args
}
}
}
"taskserv" | "task" => {
# Taskserv uses exec mode
let use_debug = if ($env.PROVISIONING_DEBUG? | default false) { "-x" } else { "" }
let cmd_args = [-mod, $module, ...$args_list, --notitles]
exec $"($env.PROVISIONING_NAME)" $use_debug ...$cmd_args
}
"cluster" => {
# Cluster uses exec mode
let use_debug = if ($env.PROVISIONING_DEBUG? | default false) { "-x" } else { "" }
let cmd_args = [-mod, $module, ...$args_list, --notitles]
exec $"($env.PROVISIONING_NAME)" $use_debug ...$cmd_args
}
"infra" => {
# Infra uses exec mode since it's a legacy module
let use_debug = if ($env.PROVISIONING_DEBUG? | default false) { "-x" } else { "" }
let cmd_args = [-mod, $module, ...$args_list, --notitles]
exec $"($env.PROVISIONING_NAME)" $use_debug ...$cmd_args
}
_ => {
print $"❌ Unknown module: ($module)"
exit 1
}
}
}
# Show infrastructure commands help
def show_infrastructure_help [] {
print ""
print "INFRASTRUCTURE"
print ""
print " s server Server lifecycle — create, delete, list, ssh, price"
print " t taskserv L2 provisioning — create, update, reset, delete, status"
print " list → components filtered to mode=taskserv"
print " show → component show"
print " c component Unified component catalog and workspace instances"
print " e component (ext) list [--mode taskserv|cluster|container] [--workspace <ws>]"
print " show <name> [--workspace <ws>] [--ext]"
print " status <name> --workspace <ws>"
print " vm Virtual machine management"
print ""
print "ORCHESTRATION"
print ""
print " w workflow WorkflowDef lifecycle — list, show, run, validate, status"
print " j job Orchestrator job management — list, status, monitor, submit"
print " b batch Batch operations"
print " o orchestrator Orchestrator daemon lifecycle"
print ""
print "Examples:"
print " prvng c list # all components"
print " prvng c list --mode cluster # cluster-mode only"
print " prvng c show postgresql --workspace libre-daoshi # full component view"
print " prvng c status k0s --workspace libre-daoshi # FSM state only"
print " prvng w list --workspace libre-daoshi # workspace workflows"
print " prvng w run deploy-services-libre-daoshi --workspace libre-daoshi"
print " prvng t create --infra libre-daoshi # L2 provisioning"
print " prvng s list # server list"
print " prvng j list # orchestrator jobs"
print ""
}
# Main infrastructure command dispatcher
export def handle_infrastructure_command [
command: string
ops: string
flags: record
] {
set_debug_env $flags
match $command {
"create" | "c" => {
# Handle: provisioning create server/taskserv/cluster <name> ...
let create_ops_list = if ($ops | is-not-empty) {
$ops | split row " " | where {|x| ($x | is-not-empty) }
} else { [] }
let resource_type = if (($create_ops_list | length) > 0) {
$create_ops_list | first
} else { "" }
let resource_name_and_args = if (($create_ops_list | length) > 1) {
$create_ops_list | skip 1 | str join " "
} else { "" }
match $resource_type {
"server" | "s" => {
let server_args = $"create ($resource_name_and_args)"
handle_server $server_args $flags
}
"taskserv" | "task" | "t" => {
let taskserv_args = $"create ($resource_name_and_args)"
handle_taskserv $taskserv_args $flags
}
"cluster" | "cl" => {
let cluster_args = $"create ($resource_name_and_args)"
handle_cluster $cluster_args $flags
}
_ => {
if ($resource_type | is-empty) {
print "❌ Resource type required for create command"
} else {
print $"❌ Unknown resource type for create: ($resource_type)"
}
print ""
print "Usage: provisioning create <resource> <name>"
print ""
print "Resources:"
print " server (s) - Create a server"
print " taskserv (t) - Create a task service"
print " cluster (cl) - Create a cluster"
exit 1
}
}
}
"delete" | "d" => {
# Handle: provisioning delete server/taskserv/cluster <name> ...
let delete_ops_list = if ($ops | is-not-empty) {
$ops | split row " " | where {|x| ($x | is-not-empty) }
} else { [] }
let resource_type = if (($delete_ops_list | length) > 0) {
$delete_ops_list | first
} else { "" }
let resource_name_and_args = if (($delete_ops_list | length) > 1) {
$delete_ops_list | skip 1 | str join " "
} else { "" }
match $resource_type {
"server" | "s" => {
let server_args = $"delete ($resource_name_and_args)"
handle_server $server_args $flags
}
"taskserv" | "task" | "t" => {
let taskserv_args = $"delete ($resource_name_and_args)"
handle_taskserv $taskserv_args $flags
}
"cluster" | "cl" => {
let cluster_args = $"delete ($resource_name_and_args)"
handle_cluster $cluster_args $flags
}
_ => {
print $"❌ Unknown resource type for delete: ($resource_type)"
exit 1
}
}
}
"bootstrap" | "bstrap" => { handle_bootstrap $ops $flags }
"fip" | "floating-ip" => { handle_fip $ops $flags }
"server" => { handle_server $ops $flags }
"taskserv" | "task" => { handle_taskserv $ops $flags }
"component" | "comp" => { handle_component $ops $flags }
"extension" | "ext" => { handle_extension $ops $flags }
"cluster" => { handle_component $ops $flags } # cluster → component (deprecated alias)
"vm" => {
# Import VM domain handler
use vm_domain.nu handle_vm_command
# Parse VM subcommand
let vm_ops_list = if ($ops | is-not-empty) {
$ops | split row " " | where {|x| ($x | is-not-empty) }
} else { [] }
let vm_command = if (($vm_ops_list | length) > 0) {
$vm_ops_list | first
} else { "vm" }
let vm_remaining_ops = if (($vm_ops_list | length) > 1) {
$vm_ops_list | skip 1 | str join " "
} else { "" }
handle_vm_command $vm_command $vm_remaining_ops $flags
}
"infra" | "infras" => {
# Show help if no ops provided
if ($ops | is-empty) {
show_infrastructure_help
} else {
handle_infra $ops $flags
}
}
"infrastructure" | "help" | "" => { show_infrastructure_help }
_ => {
print $"❌ Unknown command: ($command)"
show_infrastructure_help
exit 1
}
}
}
# Floating IP command handler
def handle_fip [ops: string, flags: record] {
use ../../main_provisioning/fip.nu *
let ops_list = if ($ops | is-not-empty) {
$ops | split row " " | where {|x| ($x | is-not-empty) }
} else { [] }
let subcommand = if ($ops_list | length) > 0 { $ops_list | first } else { "" }
let remaining = if ($ops_list | length) > 1 { $ops_list | skip 1 } else { [] }
let out_flag = ($flags | get --optional output_format | default "")
match $subcommand {
"list" | "l" => {
if ($out_flag | is-not-empty) { main list --out $out_flag } else { main list }
}
"show" | "s" => {
let name = if ($remaining | length) > 0 { $remaining | first } else {
error make { msg: "Usage: provisioning fip show <name>" }
}
if ($out_flag | is-not-empty) { main show $name --out $out_flag } else { main show $name }
}
"assign" => {
let name = if ($remaining | length) > 0 { $remaining | get 0 } else {
error make { msg: "Usage: provisioning fip assign <name> <server>" }
}
let server = if ($remaining | length) > 1 { $remaining | get 1 } else {
error make { msg: "Usage: provisioning fip assign <name> <server>" }
}
let yes = $flags.auto_confirm
main assign $name $server --yes=$yes
}
"unassign" => {
let name = if ($remaining | length) > 0 { $remaining | first } else {
error make { msg: "Usage: provisioning fip unassign <name>" }
}
let yes = $flags.auto_confirm
main unassign $name --yes=$yes
}
"protection" => {
let name = if ($remaining | length) > 0 { $remaining | get 0 } else {
error make { msg: "Usage: provisioning fip protection <name> <enable|disable>" }
}
let action = if ($remaining | length) > 1 { $remaining | get 1 } else {
error make { msg: "Usage: provisioning fip protection <name> <enable|disable>" }
}
main protection $name $action
}
_ => {
print "Floating IP Management"
print "====================="
print ""
print "Usage: provisioning fip <command> [args]"
print ""
print "Commands:"
print " list List all Floating IPs with role and protection"
print " show <name> Show detail for a specific FIP"
print " assign <name> <server> Assign FIP to a server"
print " unassign <name> Release FIP from its current server"
print " protection <name> enable|disable Toggle delete protection"
print ""
print "Examples:"
print " provisioning fip list"
print " provisioning fip show librecloud-fip-smtp"
print " provisioning fip assign librecloud-fip-smtp sgoyol-1"
print " provisioning fip unassign librecloud-fip-smtp"
print " provisioning fip protection librecloud-fip-smtp enable"
}
}
}
# Bootstrap command handler — L1 Hetzner resource provisioning
def handle_bootstrap [ops: string, flags: record] {
use ../../main_provisioning/bootstrap.nu *
let ws = ($flags | get --optional workspace | default "")
let dry = $flags.dry_run
if ($ws | is-not-empty) {
main bootstrap --workspace $ws --dry-run=$dry
} else {
main bootstrap --dry-run=$dry
}
}
# Server command handler
def handle_server [ops: string, flags: record] {
# Show help if no subcommand provided
if ($ops | is-empty) {
print "Server Management"
print "================="
print ""
print "Usage: provisioning server <command> [options]"
print ""
print "Commands:"
print " create <name> Create a new server"
print " delete <name> Delete a server"
print " list List all servers"
print " ssh <name> SSH into server"
print " price Show server pricing"
print ""
print "Examples:"
print " provisioning server create web-01"
print " provisioning server list"
print " provisioning server ssh web-01"
print ""
return
}
# Authentication check for server operations (metadata-driven)
let operation_parts = ($ops | split row " " | where {|x| ($x | is-not-empty)})
let action = if ($operation_parts | is-empty) { "" } else { $operation_parts | first }
# Determine operation type
let operation_type = match $action {
"create" | "c" => "create"
"delete" | "d" | "remove" => "delete"
"modify" | "update" => "modify"
_ => "read"
}
# Check authentication using metadata-driven approach
if not (is-check-mode $flags) and $operation_type != "read" {
let operation_name = $"server ($action)"
check-operation-auth $operation_name $operation_type $flags
}
# Extract the remaining arguments after the action verb (create/delete/list/etc)
let action_and_args = if ($operation_parts | length) > 1 {
$operation_parts | skip 1 | str join " "
} else {
""
}
let args = build_module_args $flags $action_and_args
# Pass the action as explicit subcommand so run_module knows which operation is being performed
# For create operations, this preserves plugin context by calling "main create" directly
run_module $args "server" $action --exec
}
# Task service command handler
def handle_taskserv [ops: string, flags: record] {
# Show help if no subcommand provided
if ($ops | is-empty) {
print "Task Service Management"
print "======================"
print ""
print "Usage: provisioning taskserv <command> [options]"
print ""
print "Commands:"
print " create <service> Create a task service"
print " delete <service> Delete a task service"
print " list List all task services"
print " generate <service> Generate task service config"
print ""
print "Service Mesh Options:"
print " istio - Full-featured service mesh with built-in ingress gateway"
print " linkerd - Lightweight service mesh (requires external ingress)"
print " cilium - CNI with service mesh capabilities"
print ""
print "Ingress Controller Options:"
print " nginx-ingress - Most popular, battle-tested ingress controller"
print " traefik - Modern cloud-native ingress with middleware"
print " contour - Envoy-based ingress with simple configuration"
print " haproxy-ingress - High-performance HAProxy-based ingress"
print ""
print "Examples:"
print " provisioning taskserv create kubernetes"
print " provisioning taskserv create istio"
print " provisioning taskserv create linkerd"
print " provisioning taskserv create nginx-ingress"
print " provisioning taskserv create traefik"
print " provisioning taskserv list"
print ""
print "Recommended Combinations:"
print " 1. Linkerd + Nginx Ingress - Lightweight mesh + proven ingress"
print " 2. Istio (standalone) - Full-featured with built-in gateway"
print " 3. Linkerd + Traefik - Lightweight mesh + modern ingress"
print " 4. No mesh + Nginx Ingress - Simple deployments"
print ""
return
}
# Authentication check for taskserv operations (metadata-driven)
let operation_parts = ($ops | split row " ")
let action = if ($operation_parts | is-empty) { "" } else { $operation_parts | first }
# Determine operation type
let operation_type = match $action {
"create" | "c" => "create"
"delete" | "d" | "remove" => "delete"
"modify" | "update" => "modify"
_ => "read"
}
# Check authentication using metadata-driven approach
if not (is-check-mode $flags) and $operation_type != "read" {
let operation_name = $"taskserv ($action)"
check-operation-auth $operation_name $operation_type $flags
}
# Show ontoref FSM state from both ontology instances:
# 1. provisioning project domain ($PROVISIONING/.ontology/)
# 2. active workspace domain ($PROVISIONING_KLOUD_PATH/.ontology/)
let ontoref_bin = (do { ^which ontoref } | complete | get stdout | str trim)
if ($ontoref_bin | is-not-empty) {
let prov_path = ($env.PROVISIONING? | default "")
let kloud_path = ($env.PROVISIONING_KLOUD_PATH? | default "")
let onto_roots = (
[$prov_path, $kloud_path]
| where { |p| ($p | is-not-empty) and ($p | path join ".ontology" "state.ncl" | path exists) }
| uniq
)
if ($onto_roots | is-not-empty) {
print ""
for root in $onto_roots {
do { cd $root; ^ontoref describe state } | complete | get stdout | print
}
}
}
let args = build_module_args $flags $ops
run_module $args "taskserv" --exec
}
# Cluster command handler
def handle_cluster [ops: string, flags: record] {
# Show help if no subcommand provided
if ($ops | is-empty) {
print "Cluster Management"
print "=================="
print ""
print "Usage: provisioning cluster <command> [options]"
print ""
print "Commands:"
print " deploy <layer> <cluster> Deploy L3 platform or L4 app extensions"
print " create <name> Create a new cluster"
print " delete <name> Delete a cluster"
print " list List all clusters"
print ""
print "Examples:"
print " provisioning cluster deploy platform sgoyol --ws librecloud_renew"
print " provisioning cluster deploy apps sgoyol --ws librecloud_renew"
print " provisioning cluster create k8s-prod"
print " provisioning cluster list"
print ""
return
}
let operation_parts = ($ops | split row " " | where { $in | is-not-empty })
let action = if ($operation_parts | is-empty) { "" } else { $operation_parts | first }
# Intercept deploy — routes to cluster-deploy.nu, not the old -mod cluster module
if $action in ["deploy"] {
use ../../main_provisioning/cluster-deploy.nu *
let rest = ($operation_parts | skip 1)
let layer = ($rest | get -o 0 | default "")
let cluster = ($rest | get -o 1 | default "")
if ($layer | is-empty) or ($cluster | is-empty) {
print "❌ Usage: provisioning cluster deploy <layer> <cluster> [--ws <workspace>]"
print " layer: platform | apps"
exit 1
}
let ws = ($flags | get --optional workspace | default "")
let dry = $flags.dry_run
let kube_cfg = ""
let sec_file = ""
if ($ws | is-not-empty) {
main cluster deploy $layer $cluster --workspace $ws --dry-run=$dry --kubeconfig $kube_cfg --secrets-file $sec_file
} else {
main cluster deploy $layer $cluster --dry-run=$dry --kubeconfig $kube_cfg --secrets-file $sec_file
}
return
}
# Determine operation type for auth check
let operation_type = match $action {
"create" | "c" => "create"
"delete" | "d" | "remove" | "destroy" => "delete"
"modify" | "update" => "modify"
_ => "read"
}
if not (is-check-mode $flags) and $operation_type != "read" {
let operation_name = $"cluster ($action)"
check-operation-auth $operation_name $operation_type $flags
}
let args = build_module_args $flags $ops
run_module $args "cluster" --exec
}
# Infrastructure command handler
def handle_infra [ops: string, flags: record] {
# Handle infra-specific argument building
let infra_arg = if ($flags.infra | is-not-empty) {
$"-i ($flags.infra)"
} else if ($flags.infras | is-not-empty) {
$"--infras ($flags.infras)"
} else {
$"-i (get_infra | path basename)"
}
let use_yes = if $flags.auto_confirm { "--yes" } else { "" }
let use_check = if $flags.check_mode { "--check" } else { "" }
let use_onsel = if ($flags.onsel | is-not-empty) {
$"--onsel ($flags.onsel)"
} else { "" }
let args = $"($ops) ($infra_arg) ($use_check) ($use_onsel) ($use_yes)" | str trim
run_module $args "infra"
}
# Price/cost command handler
export def handle_price_command [ops: string, flags: record] {
let use_check = if $flags.check_mode { "--check " } else { "" }
let str_infra = if ($flags.infra | is-not-empty) {
$"--infra ($flags.infra) "
} else { "" }
let str_out = if ($flags.outfile | is-not-empty) {
$"--outfile ($flags.outfile) "
} else { "" }
run_module $"($ops) ($str_infra) ($use_check) ($str_out)" "server" "price" --exec
}
# Create-server-task combined command handler
export def handle_create_server_task [ops: string, flags: record] {
# Create servers first
let server_args = build_module_args $flags $ops
run_module $server_args "server" "create"
# Check if server creation succeeded
if $env.LAST_EXIT_CODE != 0 {
_print $"🛑 Errors found in (_ansi yellow_bold)create-server(_ansi reset)"
exit 1
}
# Create taskservs
let taskserv_args = build_module_args $flags $"- ($ops)"
run_module $taskserv_args "taskserv" "create"
}
# Component command handler — unified view for extensions/components
def handle_component [ops: string, flags: record] {
let parts = ($ops | split row " ")
let action = if ($parts | is-empty) { "" } else { $parts | first }
let workspace = ($flags.workspace? | default ($flags.ws? | default ""))
let mode = ($flags.mode? | default "")
use ../../components/mod.nu *
match $action {
"list" | "ls" | "l" | "" => {
component-list $mode $workspace
}
"show" | "s" => {
if ($parts | length) < 2 {
print "❌ Error: component show requires a name"
return
}
let name = ($parts | get 1)
let ext_only = ($flags.ext? | default false)
component-show $name $workspace $ext_only
}
"status" | "st" => {
if ($parts | length) < 2 {
print "❌ Error: component status requires a name"
return
}
let name = ($parts | get 1)
component-status $name $workspace
}
"" => {
print "Component Management"
print "===================="
print ""
print "Usage: provisioning component <command> [options]"
print ""
print "Commands:"
print " list [--mode taskserv|cluster|container] [--workspace <ws>]"
print " show <name> [--workspace <ws>] [--ext]"
print " status <name> [--workspace <ws>]"
print ""
print "Examples:"
print " provisioning component list"
print " provisioning component list --mode cluster"
print " provisioning component show postgresql --workspace libre-daoshi"
print " provisioning component status k0s --workspace libre-daoshi"
}
_ => {
print $"❌ Unknown component subcommand: ($action)"
print "Use 'provisioning component' for help"
}
}
}
# Extension command handler — browses extension catalog (extensions/components/ definitions)
# e / ext → extension → shows metadata, modes, requires/provides without workspace context
def handle_extension [ops: string, flags: record] {
let parts = ($ops | split row " ")
let action = if ($parts | is-empty) { "" } else { $parts | first }
let mode = ($flags.mode? | default "")
use ../../components/mod.nu *
match $action {
"list" | "ls" | "l" | "" => {
# Extension catalog: no workspace filter (ext_only view)
component-list $mode ""
}
"show" | "s" => {
if ($parts | length) < 2 {
print "❌ Error: extension show requires a name"
return
}
component-show ($parts | get 1) "" true # ext_only = true
}
_ => {
print $"❌ Unknown extension subcommand: ($action)"
print "Use: prvng e list | prvng e show <name>"
}
}
}