@celilo/cli 0.24.0 → 0.25.0

package/src/services/network-ensure.ts

@@ -0,0 +1,260 @@
+/**
+ * Ensure every network a module REQUIRES is defined before the deploy proceeds
+ * (openspec/changes/networks-are-declared-not-written/specs/network-declaration/spec.md).
+ *
+ * celilo owns the network namespace. A module names the networks it depends on
+ * under `requires.networks`; it never carries their values. When one of those
+ * networks has no `network.<name>.subnet` in system config, this asks for the
+ * range over the event bus and writes it — celilo asking, celilo writing.
+ *
+ * Two properties matter and both come from WHERE this runs rather than from
+ * anything clever it does:
+ *
+ *   - It runs in the deploy's interview phase, before generation and before any
+ *     hook. So by the time a hook executes, the network is defined — whether the
+ *     consumer reading declared networks captured them eagerly or reads at point
+ *     of use. That is what retires the live-reader mitigation the firewall
+ *     capability carries today (celilo#759).
+ *   - It asks through the generic bus interview, so it is answerable by whatever
+ *     responder is attached — a terminal, `celilo events reply`, an automated
+ *     policy — and behaves identically interactive or headless.
+ *
+ * ── What is asked, and what is merely observed ──
+ *
+ * Not every attribute of a network is a question. The rule is whether celilo can
+ * already SEE the answer:
+ *
+ *   - `subnet` is ASKED. It is an addressing-plan decision that predates every
+ *     module, and nothing in the fleet can be consulted for it. A well-known name
+ *     is offered a suggested range so an operator new to networking is not made to
+ *     invent one — an offer in a question, never a seeded row.
+ *   - `gateway` is OBSERVED, never asked. It is the address a router answers on
+ *     inside that subnet, which celilo already holds: `machine add` catalogues
+ *     every interface of every machine. Asking for it would be asking the operator
+ *     to retype something celilo can look up, which is the same failure this change
+ *     exists to remove, aimed at a different key.
+ *   - `vlan` is ASKED, optional, blank meaning untagged. It is NOT observable —
+ *     a catalogued interface carries `{name, ipAddress, zone}` and no tag — and it
+ *     IS load-bearing: every container-provisioning template reads
+ *     `$system:network.<zone>.vlan` as the Proxmox NIC tag. Leaving it uncollected
+ *     would provision containers untagged onto the wrong VLAN, silently.
+ *
+ * Which attributes a network HAS at all is celilo's answer too, taken from
+ * `schemas/system_config.json`: it declares `gateway`/`vlan` for the routed
+ * segments and omits both for the control-plane VPN, which has neither. So a
+ * module never states which attributes it reads, and asking a nonsense question
+ * is impossible by construction.
+ */
+
+import { subnetContains } from '@celilo/capabilities';
+import { eq } from 'drizzle-orm';
+import type { DbClient } from '../db/client';
+import { machines, moduleConfigs, systemConfig } from '../db/schema';
+import { type ModuleManifest, getRequiredNetworkNames } from '../manifest/schema';
+import { askText } from './bus-interview';
+import { loadSchema } from './system-init';
+
+export interface NetworkEnsureResult {
+  success: boolean;
+  error?: string;
+  /** What this call defined, as `<key> = <value>` lines for the deploy log. */
+  applied: string[];
+}
+
+/** The one piece of I/O here, injectable so the rules are testable without a bus. */
+export type NetworkAsker = (question: {
+  scope: string;
+  key: string;
+  message: string;
+  description: string;
+  defaultValue?: string;
+  placeholder?: string;
+  required: boolean;
+  pattern?: string;
+}) => Promise<string>;
+
+/**
+ * The module's own config, flat, for resolving a `from:` requirement.
+ *
+ * Reads `value_json` in preference to `value`: an array config value is stored
+ * as JSON, and reading the scalar column would yield the string form, which
+ * `getRequiredNetworkNames` would then have to guess at.
+ */
+function loadModuleConfigValues(db: DbClient, moduleId: string): Record<string, unknown> {
+  const values: Record<string, unknown> = {};
+  for (const row of db
+    .select()
+    .from(moduleConfigs)
+    .where(eq(moduleConfigs.moduleId, moduleId))
+    .all()) {
+    values[row.key] = row.valueJson ?? row.value;
+  }
+  return values;
+}
+
+function readSystemConfigValue(db: DbClient, key: string): string | undefined {
+  const row = db.select().from(systemConfig).where(eq(systemConfig.key, key)).get();
+  return row?.value && row.value.length > 0 ? row.value : undefined;
+}
+
+function writeSystemConfigValue(db: DbClient, key: string, value: string): void {
+  db.insert(systemConfig)
+    .values({ key, value })
+    .onConflictDoUpdate({ target: systemConfig.key, set: { value } })
+    .run();
+}
+
+/**
+ * The address a router answers on inside `subnet`, from the machine catalogue.
+ *
+ * Matched by CONTAINMENT rather than by an interface's recorded zone, because on
+ * a fleet that is still being described those are not the same thing: an
+ * interface is classified into a zone by comparing it against declared subnets,
+ * so before the subnet exists the interface has no zone. Containment answers the
+ * moment the operator supplies the range, which is exactly when this runs.
+ *
+ * `role: 'router'` is what distinguishes the gateway from any other host that
+ * happens to sit in the subnet.
+ */
+export function observeGateway(db: DbClient, subnet: string): string | undefined {
+  for (const machine of db.select().from(machines).all()) {
+    if (machine.role !== 'router') continue;
+    for (const iface of machine.interfaces) {
+      if (iface.ipAddress && subnetContains(subnet, iface.ipAddress)) {
+        return iface.ipAddress;
+      }
+    }
+  }
+  return undefined;
+}
+
+/**
+ * Ensure each of `manifest.requires.networks` is defined: a subnet, a vlan tag
+ * where one applies, and a gateway wherever celilo can see one.
+ */
+export async function ensureRequiredNetworks(
+  moduleId: string,
+  manifest: ModuleManifest,
+  db: DbClient,
+  ask: NetworkAsker = askText,
+): Promise<NetworkEnsureResult> {
+  // A `from:` requirement resolves against the module's OWN config, so the
+  // module's values have to be loaded before its required set is even knowable.
+  const names = getRequiredNetworkNames(manifest, loadModuleConfigValues(db, moduleId));
+  if (names.length === 0) return { success: true, applied: [] };
+
+  const schema = loadSchema();
+  const applied: string[] = [];
+
+  for (const name of names) {
+    const subnetKey = `network.${name}.subnet`;
+    const subnetProperty = schema.properties[subnetKey];
+    if (!subnetProperty) {
+      return {
+        success: false,
+        applied,
+        error: [
+          `Module "${moduleId}" requires a network called "${name}", which celilo does not know`,
+          `about: there is no "${subnetKey}" in celilo's system-config schema. Networks are`,
+          "celilo's to define, so a new one is added to schemas/system_config.json — a module",
+          'cannot introduce one.',
+        ].join(' '),
+      };
+    }
+
+    // 1. The subnet — asked, because nothing in the fleet can be consulted for it.
+    //
+    // Whether this network already EXISTED is the fact everything below turns
+    // on: celilo is here to define a network, not to audit one it already holds.
+    let subnet = readSystemConfigValue(db, subnetKey);
+    const defining = subnet === undefined;
+    if (defining) {
+      const answer = (
+        await ask({
+          scope: `network:${name}`,
+          key: 'subnet',
+          message: `Subnet CIDR for the "${name}" network:`,
+          description: [
+            `${moduleId} requires the "${name}" network, and celilo has no subnet for it.`,
+            "This becomes celilo's definition of the network — every module that needs it reads",
+            'this one value.',
+          ].join(' '),
+          defaultValue: subnetProperty.suggested,
+          // A defaultValue MUST have a matching placeholder, or the operator
+          // cannot see what pressing Enter would accept.
+          placeholder: subnetProperty.suggested,
+          required: true,
+          pattern: subnetProperty.pattern,
+        })
+      ).trim();
+
+      if (answer.length === 0) {
+        return {
+          success: false,
+          applied,
+          error: `No subnet supplied for the "${name}" network; ${moduleId} cannot deploy without it.`,
+        };
+      }
+      writeSystemConfigValue(db, subnetKey, answer);
+      applied.push(`${subnetKey} = ${answer}`);
+      subnet = answer;
+    }
+
+    // 2. The VLAN tag — asked, because it cannot be observed: a catalogued
+    //    interface carries no tag. Optional: an untagged segment genuinely has
+    //    none, and a blank answer writes nothing rather than writing "".
+    //
+    //    ONLY when celilo is defining the network. An existing network was
+    //    already defined without a tag — deliberately, or because it is
+    //    untagged — and re-opening that question every time a new module
+    //    requires it is not a question, it is a deploy that stops. Which is
+    //    exactly what happened: `iptables` requires `internal`, whose subnet the
+    //    management install had already recorded and whose vlan nothing ever
+    //    set, so a headless deploy died on `interview.required.network:internal.vlan`
+    //    with no responder to answer it. Absent means untagged; if that is wrong,
+    //    `celilo system config set network.<name>.vlan <tag>` says so once.
+    const vlanKey = `network.${name}.vlan`;
+    if (
+      defining &&
+      schema.properties[vlanKey] &&
+      readSystemConfigValue(db, vlanKey) === undefined
+    ) {
+      const answer = (
+        await ask({
+          scope: `network:${name}`,
+          key: 'vlan',
+          message: `VLAN tag for the "${name}" network (blank if untagged):`,
+          description: [
+            'Container provisioning reads this as the NIC tag, so a tagged fleet that leaves it',
+            'unset puts containers on the wrong VLAN without saying so. Leave it blank if this',
+            'segment is untagged.',
+          ].join(' '),
+          required: false,
+        })
+      ).trim();
+      if (answer.length > 0) {
+        writeSystemConfigValue(db, vlanKey, answer);
+        applied.push(`${vlanKey} = ${answer}`);
+      }
+    }
+
+    // 3. The gateway — OBSERVED. celilo catalogues every machine's interfaces at
+    //    `machine add`, so the router's address inside this subnet is a fact it
+    //    already holds. Absent means no router leg is catalogued there yet, which
+    //    a later `machine add` or firewall deploy resolves; it is not a question.
+    const gatewayKey = `network.${name}.gateway`;
+    if (
+      subnet &&
+      schema.properties[gatewayKey] &&
+      readSystemConfigValue(db, gatewayKey) === undefined
+    ) {
+      const observed = observeGateway(db, subnet);
+      if (observed) {
+        writeSystemConfigValue(db, gatewayKey, observed);
+        applied.push(`${gatewayKey} = ${observed} (observed from the machine catalogue)`);
+      }
+    }
+  }
+
+  return { success: true, applied };
+}

package/src/services/system-init.ts

@@ -15,7 +15,7 @@ import { systemConfig } from '../db/schema';
 /**
  * System configuration schema interface
  */
-interface SystemConfigSchema {
+export interface SystemConfigSchema {
   properties: Record<
     string,
     {
@@ -26,6 +26,17 @@ interface SystemConfigSchema {
       minimum?: number;
       maximum?: number;
       format?: string;
+      /**
+       * A starting offer for an interview, deliberately NOT a `default`.
+       *
+       * `getDefaultConfiguration()` seeds every `default:` it finds at `system
+       * init`, and network addressing is specifically not seeded
+       * (openspec/specs/progressive-zone-disclosure/spec.md). A separate field
+       * is structurally incapable of becoming a row nobody chose: it is only
+       * ever read to pre-fill a question the operator still has to answer
+       * ([[services/network-ensure.ts]]).
+       */
+      suggested?: string;
     }
   >;
 }
@@ -33,7 +44,7 @@ interface SystemConfigSchema {
 /**
  * Load system config schema from JSON file
  */
-function loadSchema(): SystemConfigSchema {
+export function loadSchema(): SystemConfigSchema {
   // Try common locations (relative to this file's directory and cwd)
   const thisDir = dirname(new URL(import.meta.url).pathname);
   const candidates = [
@@ -80,7 +91,7 @@ export function getDefaultConfiguration(): Record<string, string | number> {
  * @param subnet - CIDR notation (e.g., "10.0.10.0/24")
  * @returns Gateway IP (e.g., "10.0.10.1")
  */
-function computeGateway(subnet: string): string {
+export function computeGateway(subnet: string): string {
   const [network, _bits] = subnet.split('/');
   const octets = network.split('.').map(Number);
 

package/src/test-utils/bus-responder.ts

@@ -42,6 +42,7 @@ import { type DbClient, createDbClient } from '../db/client';
 import type {
   ConfigRequiredPayload,
   EnsureRequiredPayload,
+  InterviewRequiredPayload,
   SecretRequiredPayload,
 } from '../services/bus-interview';
 import {
@@ -79,6 +80,8 @@ export interface BusResponderFixture {
   seenSecretPayloads(): SecretRequiredPayload[];
   /** Snapshot of every ensure.required payload the responder saw. */
   seenEnsurePayloads(): EnsureRequiredPayload[];
+  /** Snapshot of every generic interview.required payload the responder saw. */
+  seenInterviewPayloads(): InterviewRequiredPayload[];
   /** Stop watching and close the bus + db connections. */
   close(): void;
 }
@@ -102,7 +105,16 @@ export function startBusResponderFixture(opts: BusResponderFixtureOptions): BusR
   const handle: ProgrammaticResponderHandle = startProgrammaticResponder({
     busDbPath: opts.busDbPath,
     db,
-    values: { config: opts.config, secrets: opts.secrets, ensures: opts.ensures },
+    // Every family the fixture's options type accepts must be forwarded, or a
+    // test supplies an answer that silently never arrives and the deploy hangs
+    // on a question nobody is listening for.
+    values: {
+      config: opts.config,
+      secrets: opts.secrets,
+      ensures: opts.ensures,
+      interview: opts.interview,
+      aspects: opts.aspects,
+    },
     onMissing: 'throw',
     emittedBy: 'test-bus-responder',
   });
@@ -114,6 +126,7 @@ export function startBusResponderFixture(opts: BusResponderFixtureOptions): BusR
     seenConfigPayloads: () => handle.seenConfigPayloads(),
     seenSecretPayloads: () => handle.seenSecretPayloads(),
     seenEnsurePayloads: () => handle.seenEnsurePayloads(),
+    seenInterviewPayloads: () => handle.seenInterviewPayloads(),
     close: () => {
       handle.close();
       db.$client.close();

package/src/variables/context.ts

@@ -185,9 +185,14 @@ function resolveSelfRefsInObject(
 }
 
 /**
- * Build resolution context for a module
+ * Build the resolution context for a module, provisioning as it goes.
  *
- * Execution function (Rule 10.1) - performs database queries
+ * Execution function (Rule 10.1) - performs database queries AND writes:
+ * it seeds `module_configs` with defaults and zone-derived networking, and
+ * records the module's deployed system (allocating IPAM addresses when the
+ * host is a celilo-provisioned container). This is the generate/deploy-time
+ * entrypoint. Anything that only wants to READ the resolved configuration
+ * wants {@link readResolutionContext} instead.
  *
  * @param moduleId - Module to build context for
  * @param db - Database client (optional, for testing)
@@ -197,6 +202,55 @@ export async function buildResolutionContext(
   moduleId: string,
   db = getDb(),
 ): Promise<ResolutionContext> {
+  return assembleResolutionContext(moduleId, db, { provision: true });
+}
+
+/**
+ * The same resolved configuration, computed without touching the database.
+ *
+ * Every derived value is recomputed from its current upstream, exactly as a
+ * build would compute it, but nothing is written: no config rows are seeded,
+ * no addresses are allocated, no deployed system is recorded. That makes it
+ * safe to call from read paths that run on a cadence — health checks, hook
+ * invocations, capability factories — where a provisioning side effect would
+ * be both surprising and, in the IPAM case, harmful.
+ *
+ * This is what {@link import('../hooks/load-hook-config').loadHookConfigMap}
+ * uses so a hook sees the value a derive currently produces rather than only
+ * the ones that happen to have been stored.
+ *
+ * @param moduleId - Module to build context for
+ * @param db - Database client (optional, for testing)
+ * @returns Resolution context with all data sources
+ */
+export async function readResolutionContext(
+  moduleId: string,
+  db = getDb(),
+): Promise<ResolutionContext> {
+  return assembleResolutionContext(moduleId, db, { provision: false });
+}
+
+/**
+ * The shared body of both entrypoints. `provision` is deliberately private
+ * (Rule 10.3): callers choose a named function, not a flag.
+ */
+async function assembleResolutionContext(
+  moduleId: string,
+  db: DbClient,
+  { provision }: { provision: boolean },
+): Promise<ResolutionContext> {
+  /**
+   * Seed a config row — a no-op when only reading. Every seeded value is also
+   * assigned into `selfConfig` by the caller, so the resolved context is the
+   * same either way; what differs is whether it is written down.
+   */
+  const persistConfig = (
+    key: string,
+    value: string | number | boolean | unknown[] | Record<string, unknown>,
+  ): void => {
+    if (provision) upsertModuleConfig(db, moduleId, key, value);
+  };
+
   // Fetch module manifest for VM resources
   const module = db.select().from(modules).where(eq(modules.id, moduleId)).get();
 
@@ -226,12 +280,12 @@ export async function buildResolutionContext(
 
     // Store assigned values in module config
     if (assigned.hostname) {
-      upsertModuleConfig(db, moduleId, 'hostname', assigned.hostname);
+      persistConfig('hostname', assigned.hostname);
       selfConfig.hostname = assigned.hostname;
     }
 
     if (assigned.zone) {
-      upsertModuleConfig(db, moduleId, 'zone', assigned.zone);
+      persistConfig('zone', assigned.zone);
       selfConfig.zone = assigned.zone;
     }
   }
@@ -248,9 +302,7 @@ export async function buildResolutionContext(
       // declared shape — e.g. `default: 2222` (YAML int) round-trips as
       // `number` not the string "2222". This is the root of Defect 1.
       if (variable.default !== undefined && !selfConfig[variable.name]) {
-        upsertModuleConfig(
-          db,
-          moduleId,
+        persistConfig(
           variable.name,
           variable.default as string | number | boolean | unknown[] | Record<string, unknown>,
         );
@@ -303,7 +355,7 @@ export async function buildResolutionContext(
       for (const { manifestKey, configKey, systemValue } of resourceMappings) {
         if (systemValue != null) {
           // Canonical system size — always wins so a resize propagates.
-          upsertModuleConfig(db, moduleId, configKey, systemValue);
+          persistConfig(configKey, systemValue);
           selfConfig[configKey] = String(systemValue);
           continue;
         }
@@ -312,9 +364,7 @@ export async function buildResolutionContext(
         // Pass them through unstringified so valueJson preserves the
         // shape — see comment in the variable-defaults block above.
         if (value !== undefined && !selfConfig[configKey]) {
-          upsertModuleConfig(
-            db,
-            moduleId,
+          persistConfig(
             configKey,
             value as string | number | boolean | unknown[] | Record<string, unknown>,
           );
@@ -333,7 +383,11 @@ export async function buildResolutionContext(
   //                         outputs (resolveInfrastructureVariables), not here.
   // This is the single place generate-time addresses are recorded — `target_ip`
   // no longer lives in module_configs.
-  if (module?.manifestData) {
+  //
+  // Provisioning only. A read must never reach this: allocating an address is
+  // not something looking at a config should do, and by the time any hook runs
+  // the row is already there for `buildInfraSystemsMap` below to read.
+  if (provision && module?.manifestData) {
     const manifest = module.manifestData as ModuleManifest;
     const declared = getDeclaredSystems(manifest);
     const hostname = selfConfig.hostname;
@@ -593,7 +647,7 @@ export async function buildResolutionContext(
 
     // If zone from manifest but not in selfConfig, store it as first-class config
     if (zone && !selfConfig.zone) {
-      upsertModuleConfig(db, moduleId, 'zone', zone);
+      persistConfig('zone', zone);
       selfConfig.zone = zone;
     }
 
@@ -625,7 +679,7 @@ export async function buildResolutionContext(
               }
               return value;
             })();
-            upsertModuleConfig(db, moduleId, field, coerced);
+            persistConfig(field, coerced);
             selfConfig[field] = String(coerced);
           }
         }
@@ -685,7 +739,7 @@ export async function buildResolutionContext(
       }
 
       if (isNew || isChanged) {
-        upsertModuleConfig(db, moduleId, key, value);
+        persistConfig(key, value);
         selfConfig[key] = value;
       }
     }

package/src/variables/declarative-derivation.test.ts

@@ -52,6 +52,59 @@ describe('resolveDeclarativeDerivation', () => {
       expect(result).toBe('10.0.10.0/24');
     });
 
+    test('resolves a system key containing hyphens', () => {
+      // celilo's zone names are kebab-case, so shipped manifests contain
+      // `$system:network.control-plane-vpn.subnet` and
+      // `$system:network.secure-mgmt.subnet`. While the match stopped at the
+      // first hyphen these looked up `network.control` / `network.secure`,
+      // threw, and — being optional — resolved to nothing in silence. That is
+      // half of the 2026-08-14 DNS outage.
+      const variable: VariableDeclare = {
+        name: 'vpn_subnet',
+        type: 'string',
+        required: false,
+        source: 'system',
+        derive_from: '$system:network.control-plane-vpn.subnet',
+      };
+
+      const context: ResolutionContext = {
+        moduleId: 'technitium',
+        selfConfig: {},
+        systemConfig: { 'network.control-plane-vpn.subnet': '10.255.255.0/24' },
+        systemSecrets: {},
+        secrets: {},
+        capabilities: {},
+      };
+
+      expect(resolveDeclarativeDerivation(variable, context)).toBe('10.255.255.0/24');
+    });
+
+    test('names the whole hyphenated key when it is missing', () => {
+      // The message has to name the key the manifest asked for. Reporting
+      // `network.secure` for a manifest that says `network.secure-mgmt` sends
+      // the reader looking for a key that was never requested.
+      const variable: VariableDeclare = {
+        name: 'secure_mgmt_subnet',
+        type: 'string',
+        required: true,
+        source: 'system',
+        derive_from: '$system:network.secure-mgmt.subnet',
+      };
+
+      const context: ResolutionContext = {
+        moduleId: 'technitium',
+        selfConfig: {},
+        systemConfig: {},
+        systemSecrets: {},
+        secrets: {},
+        capabilities: {},
+      };
+
+      expect(() => resolveDeclarativeDerivation(variable, context)).toThrow(
+        "Missing system config: network.secure-mgmt.subnet (required by variable 'secure_mgmt_subnet')",
+      );
+    });
+
     test('throws on missing system config', () => {
       const variable: VariableDeclare = {
         name: 'primary_domain',

package/src/variables/declarative-derivation.ts

@@ -56,8 +56,19 @@ function substituteVariables(
 ): string {
   let result = input;
 
-  // Replace $system:key patterns (both $system:key and ${system:key} forms)
-  result = result.replace(/\$\{?system:([a-zA-Z0-9_.]+)\}?/g, (_match, key) => {
+  // Replace $system:key patterns (both $system:key and ${system:key} forms).
+  //
+  // The key may contain hyphens. celilo's own zone names are kebab-case, so
+  // `$system:network.control-plane-vpn.subnet` and
+  // `$system:network.secure-mgmt.subnet` are both real keys in shipped
+  // manifests — and neither could ever resolve while this class excluded `-`:
+  // the match stopped at the first hyphen, looked up `network.control`, and
+  // threw. For an optional variable that throw is swallowed, so the derive
+  // simply produced nothing, forever, in silence. That is half of the
+  // 2026-08-14 DNS outage (`technitium.vpn_subnet`); the other half is that
+  // nothing re-derived the value at read time — see
+  // `hooks/load-hook-config.ts`.
+  result = result.replace(/\$\{?system:([a-zA-Z0-9_.-]+)\}?/g, (_match, key) => {
     const value = context.systemConfig[key];
     if (value === undefined) {
       throw new Error(`Missing system config: ${key} (required by variable '${variableName}')`);