mesh-controller/internal/inventory/secrets.go

package inventory

import (
	"context"
	"errors"
	"fmt"
	"slices"
	"sort"
	"strings"

	"github.com/jackc/pgx/v5"

	"github.com/novox/mesh-controller/internal/catalogue"
	"github.com/novox/mesh-controller/internal/secrets"
)

// Where sealed secrets live.
//
// The table holds nothing usable — see the migration and internal/secrets for why that is the
// design rather than an inconvenience.

// Secret is one provision's credential, sealed to each end.
type Secret struct {
	Name     string
	Consumer string
	// ConsumerModule is which module on that machine it is for.
	//
	// **Part of the key, not a label** (novox/hq 04-ISSUES/022). Two modules on one node wanting
	// the same provision are two consumers, and were one credential until this.
	ConsumerModule string
	// Local is the name the credential goes by inside the consumer where it keeps several for one
	// provision (novox/hq ADR 0094); empty for the ordinary one. Part of the key.
	Local       string
	Provider    string
	ForConsumer string
	ForProvider string
	ConsumerKey string
	ProviderKey string
	// Origin is `made` — the mesh generated it — or `accepted` — a person supplied it, for
	// something outside the mesh, and the mesh cannot make another (novox/hq 04-ISSUES/070).
	Origin string
}

// Where a pair credential came from.
const (
	OriginMade     = "made"
	OriginAccepted = "accepted"
)

// SecretFor is the credential one module uses for one provision, making it the first time.
//
// **Made once and kept**, rather than regenerated whenever it is asked for. A secret that changed
// on every declaration would restart both ends on every push and would mean the password a
// provider was told to create never matches the one a consumer was given — which is a mesh that
// reports success and cannot connect.
//
// **Remade when either end's sealing key changes.** A node that rejoined generated a new key and
// can no longer open what was sealed to the old one, so keeping the blob would deliver something
// unreadable for ever. The new secret reaches both ends in the same push, which is the only
// moment they can be changed together.
func (i *Inventory) SecretFor(ctx context.Context, name, consumer, consumerModule, provider, local string) (
	Secret, error) {
	consumerKey, err := i.SealingKeyOf(ctx, consumer)
	if err != nil {
		return Secret{}, err
	}
	providerKey, err := i.SealingKeyOf(ctx, provider)
	if err != nil {
		return Secret{}, err
	}

	consumerNode, err := i.NodeByName(ctx, consumer)
	if err != nil {
		return Secret{}, err
	}
	providerNode, err := i.NodeByName(ctx, provider)
	if err != nil {
		return Secret{}, err
	}

	var held Secret
	err = i.store.Pool().QueryRow(ctx,
		`select for_consumer, for_provider, consumer_key, provider_key, origin from secret
		 where name = $1 and consumer = $2 and consumer_module = $3 and provider = $4 and local = $5`,
		name, consumerNode.ID, consumerModule, providerNode.ID, local).
		Scan(&held.ForConsumer, &held.ForProvider, &held.ConsumerKey, &held.ProviderKey, &held.Origin)
	if err == nil && held.ConsumerKey == consumerKey && held.ProviderKey == providerKey {
		held.Name, held.Consumer, held.Provider = name, consumer, provider
		held.ConsumerModule, held.Local = consumerModule, local
		return held, nil
	}
	if err == nil && held.Origin == OriginAccepted {
		// A person supplied this, and the mesh does not hold the value: it cannot seal it to the
		// new key. Refused aloud rather than replaced by something the mesh made up, which would
		// be delivered, reported as applied, and fail to authenticate somewhere else entirely
		// (novox/hq 04-ISSUES/070).
		return Secret{}, fmt.Errorf(
			"%s's %q credential from %s was accepted from a person, and a sealing key at one end "+
				"has changed since. The mesh cannot re-seal a value it does not hold: accept it "+
				"again with `secret accept %s %s %s --provider %s%s`",
			consumerModule, name, provider, consumer, consumerModule, name, provider, localFlag(local))
	}

	// And to the operator, when the mesh has one (novox/hq ADR 0085, amended): the third copy that
	// makes a vault-provided secret recoverable, and nothing the mesh can open.
	operator, err := i.OperatorKey(ctx)
	if err != nil {
		return Secret{}, err
	}
	made, blob, err := secrets.MakeWithOperator(consumerKey, providerKey, operator)
	if err != nil {
		return Secret{}, err
	}
	forOperator, operatorKey := operatorColumns(operator, blob)
	_, err = i.store.Pool().Exec(ctx,
		`insert into secret (name, consumer, consumer_module, provider, for_consumer, for_provider,
		                     consumer_key, provider_key, operator_sealed, operator_key, local)
		 values ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10, $11)
		 on conflict (name, local, consumer, consumer_module, provider) do update set
		   for_consumer = excluded.for_consumer, for_provider = excluded.for_provider,
		   consumer_key = excluded.consumer_key, provider_key = excluded.provider_key,
		   created_at = now(),
		   operator_sealed = excluded.operator_sealed, operator_key = excluded.operator_key`,
		name, consumerNode.ID, consumerModule, providerNode.ID,
		made.ForConsumer, made.ForProvider, made.ConsumerKey, made.ProviderKey, forOperator, operatorKey, local)
	if err != nil {
		return Secret{}, err
	}
	return Secret{Name: name, Consumer: consumer, ConsumerModule: consumerModule, Local: local,
		Provider:    provider,
		ForConsumer: made.ForConsumer, ForProvider: made.ForProvider,
		ConsumerKey: made.ConsumerKey, ProviderKey: made.ProviderKey, Origin: OriginMade}, nil
}

// AcceptSecretForPair takes a value a person supplied into a pair credential — sealed to the
// consumer's node and to the provider's, and to the operator when the mesh has one — where the
// mesh would otherwise have made one (novox/hq 04-ISSUES/070, ADR 0092).
//
// This is the vault's third species: a credential for something outside the mesh, which only a
// person can supply. It is the counterpart to AcceptSecretForModule for a module's own secret;
// what differs is that both ends of the pair are sealed to, and that the record says `accepted`
// so a later read never replaces it with a minted one. The plaintext is discarded here.
func (i *Inventory) AcceptSecretForPair(ctx context.Context, name, consumer, consumerModule, provider, local, value string) error {
	// Refused for a requirement the module does not have, or a local it does not keep under it
	// (novox/hq 04-ISSUES/078): the credential would sit in the pair unread.
	m, err := i.declared(ctx, consumerModule)
	if err != nil {
		return err
	}
	if !slices.Contains(m.Requires, name) {
		return fmt.Errorf("%s does not require %q; it requires: %s", consumerModule, name, orNone(m.Requires))
	}
	if locals := m.SecretsMany[name]; len(locals) > 0 {
		if local == "" {
			return fmt.Errorf("%s keeps several secrets for %q; name one with --local: %s",
				consumerModule, name, orNone(sortedNames(locals)))
		}
		if _, kept := locals[local]; !kept {
			return fmt.Errorf("%s does not keep %q for %q; it keeps: %s",
				consumerModule, local, name, orNone(sortedNames(locals)))
		}
	} else if m.Secrets[name] == "" {
		return fmt.Errorf("%s requires %q but keeps no secret for it, so a delivered value would sit unread; "+
			"it keeps secrets for: %s", consumerModule, name, orNone(sortedNames(m.Secrets)))
	} else if local != "" {
		return fmt.Errorf("%s keeps one secret for %q, not several; drop --local", consumerModule, name)
	}
	consumerKey, err := i.SealingKeyOf(ctx, consumer)
	if err != nil {
		return err
	}
	providerKey, err := i.SealingKeyOf(ctx, provider)
	if err != nil {
		return err
	}
	if consumerKey == "" || providerKey == "" {
		return fmt.Errorf(
			"both %s and %s need a sealing key before a credential can be sealed to them — a "+
				"node joins to get one", consumer, provider)
	}
	consumerNode, err := i.NodeByName(ctx, consumer)
	if err != nil {
		return err
	}
	providerNode, err := i.NodeByName(ctx, provider)
	if err != nil {
		return err
	}
	sealed, err := secrets.Accept(value, consumerKey, providerKey)
	if err != nil {
		return err
	}
	forOperator, operatorKey, err := i.operatorSeal(ctx, value)
	if err != nil {
		return err
	}
	_, err = i.store.Pool().Exec(ctx,
		`insert into secret (name, consumer, consumer_module, provider, for_consumer, for_provider,
		                     consumer_key, provider_key, operator_sealed, operator_key, origin, local)
		 values ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10, $11, $12)
		 on conflict (name, local, consumer, consumer_module, provider) do update set
		   for_consumer = excluded.for_consumer, for_provider = excluded.for_provider,
		   consumer_key = excluded.consumer_key, provider_key = excluded.provider_key,
		   created_at = now(), origin = excluded.origin,
		   operator_sealed = excluded.operator_sealed, operator_key = excluded.operator_key`,
		name, consumerNode.ID, consumerModule, providerNode.ID,
		sealed.ForConsumer, sealed.ForProvider, sealed.ConsumerKey, sealed.ProviderKey,
		forOperator, operatorKey, OriginAccepted, local)
	return err
}

// RotateSecret discards what was there, so the next declaration carries a new one.
//
// Only a delete. Nothing reads the old value first, because nothing can — and making the
// replacement here rather than on the next read would be a second path to the same act, which is
// how two ends come to hold different passwords.
//
// The new secret then reaches both ends on the same push, together, which is what makes rotation
// a single event rather than a fanout with a window where half the mesh holds a dead credential.
//
// **An accepted credential is not rotated.** The mesh did not make it and cannot make its
// replacement; deleting it would have the next read mint one, which is exactly the wrong value
// delivered with the mesh insisting it was (novox/hq 04-ISSUES/070). Refused, and the remedy named.
func (i *Inventory) RotateSecret(ctx context.Context, name, consumer, consumerModule, provider, local string) error {
	consumerNode, err := i.NodeByName(ctx, consumer)
	if err != nil {
		return err
	}
	providerNode, err := i.NodeByName(ctx, provider)
	if err != nil {
		return err
	}
	var origin string
	err = i.store.Pool().QueryRow(ctx,
		`select origin from secret where name = $1 and consumer = $2 and consumer_module = $3
		   and provider = $4 and local = $5`,
		name, consumerNode.ID, consumerModule, providerNode.ID, local).Scan(&origin)
	if err == nil && origin == OriginAccepted {
		return fmt.Errorf(
			"%s's %q credential from %s was accepted from a person, and the mesh cannot make "+
				"its replacement. Accept the new value instead: `secret accept %s %s %s "+
				"--provider %s%s --from <file>`",
			consumerModule, name, provider, consumer, consumerModule, name, provider, localFlag(local))
	}
	_, err = i.store.Pool().Exec(ctx,
		`delete from secret where name = $1 and consumer = $2 and consumer_module = $3
		   and provider = $4 and local = $5`,
		name, consumerNode.ID, consumerModule, providerNode.ID, local)
	return err
}

// SecretsFrom is every credential a provider node was issued, so it can be told what to create.
func (i *Inventory) SecretsFrom(ctx context.Context, provider string) ([]Secret, error) {
	providerNode, err := i.NodeByName(ctx, provider)
	if err != nil {
		return nil, err
	}
	rows, err := i.store.Pool().Query(ctx,
		`select s.name, c.name, s.consumer_module, s.local, s.for_provider from secret s
		 join node c on c.id = s.consumer
		 where s.provider = $1 order by s.name, c.name, s.consumer_module, s.local`, providerNode.ID)
	if err != nil {
		return nil, err
	}
	defer rows.Close()

	var out []Secret
	for rows.Next() {
		s := Secret{Provider: provider}
		if err := rows.Scan(&s.Name, &s.Consumer, &s.ConsumerModule, &s.Local, &s.ForProvider); err != nil {
			return nil, err
		}
		out = append(out, s)
	}
	return out, rows.Err()
}

// SecretForModule is a secret a module needs in order to be itself, on one machine.
//
// Not the credential a consumer is given: a superuser password is not *for* anybody. Made once
// and kept, because regenerating it on every declaration would change the password a running
// database has already been started with — and remade when the node's sealing key changes, for
// the same reason as everything else sealed here.
// ModuleSecretIfIssued is what a module already holds on a node, and nothing if it holds nothing.
//
// **The read half of SecretForModule**, which mints one when there is none — an insert, and a row
// lock, on a path that also serves questions. Composing a declaration to answer *is this machine
// running what I would send it* went through the minting version for every module on every node,
// so asking wrote to the database and blocked against the machine it was asking about.
//
// A module with no secret yet has never been sent one, which is the same answer the caller wanted
// anyway: this machine is not running what the mesh would send it.
func (i *Inventory) ModuleSecretIfIssued(
	ctx context.Context, node, module, name string,
) (string, bool, error) {
	key, err := i.SealingKeyOf(ctx, node)
	if err != nil || key == "" {
		return "", false, err
	}
	record, err := i.NodeByName(ctx, node)
	if err != nil {
		return "", false, err
	}
	var sealed, against, origin string
	err = i.store.Pool().QueryRow(ctx,
		`select sealed, node_key, origin from module_secret
		  where node = $1 and module = $2 and name = $3`,
		record.ID, module, name).Scan(&sealed, &against, &origin)
	if errors.Is(err, pgx.ErrNoRows) {
		return "", false, nil
	}
	if err != nil {
		return "", false, err
	}
	// Sealed to a key the node no longer has is not something it holds. Reported as absent rather
	// than as an error: this is the read, and refusing here would make a question fail for a
	// condition its writing counterpart is the right place to explain.
	if against != key {
		return "", false, nil
	}
	return sealed, true, nil
}

func (i *Inventory) SecretForModule(ctx context.Context, node, module, name string) (string, error) {
	key, err := i.SealingKeyOf(ctx, node)
	if err != nil {
		return "", err
	}
	if key == "" {
		return "", fmt.Errorf(
			"%s needs a secret and %s has no sealing key, so nothing can be sealed to it",
			module, node)
	}
	record, err := i.NodeByName(ctx, node)
	if err != nil {
		return "", err
	}

	var sealed, against, origin string
	err = i.store.Pool().QueryRow(ctx,
		`select sealed, node_key, origin from module_secret
		  where node = $1 and module = $2 and name = $3`,
		record.ID, module, name).Scan(&sealed, &against, &origin)
	if err == nil && against == key {
		return sealed, nil
	}
	if err == nil && origin == "accepted" {
		// Sealed to a key this node no longer has, and not the mesh's to invent again. Making one
		// would put 32 random bytes where a working credential was: the machine would apply it,
		// report success, and whatever reads it would fail to authenticate somewhere else
		// entirely — with the mesh insisting the secret was delivered, which it was.
		return "", fmt.Errorf(
			"%s on %s holds %q, which was given to the mesh rather than made by it, and %s has "+
				"since generated a new sealing key. The mesh cannot make another; issue it again",
			module, node, name, node)
	}

	operator, err := i.OperatorKey(ctx)
	if err != nil {
		return "", err
	}
	// Sealed once to the machine — Make seals to two ends because a provision has two; here both
	// are the same machine, and only one copy is kept — and once more to the operator when the
	// mesh has one (novox/hq ADR 0085, amended), which is the copy a person can recover from.
	made, blob, err := secrets.MakeWithOperator(key, key, operator)
	if err != nil {
		return "", err
	}
	forOperator, operatorKey := operatorColumns(operator, blob)
	if _, err := i.store.Pool().Exec(ctx,
		`insert into module_secret (node, module, name, sealed, node_key, origin, operator_sealed, operator_key)
		 values ($1, $2, $3, $4, $5, 'made', $6, $7)
		 on conflict (node, module, name) do update set
		   sealed = excluded.sealed, node_key = excluded.node_key,
		   origin = excluded.origin, made_at = now(),
		   operator_sealed = excluded.operator_sealed, operator_key = excluded.operator_key`,
		record.ID, module, name, made.ForConsumer, key, forOperator, operatorKey); err != nil {
		return "", err
	}
	return made.ForConsumer, nil
}

// AcceptSecretForModule keeps a value somebody supplied as a module's own secret.
//
// The counterpart to SecretForModule, which generates one. Some of what a module needs the mesh
// cannot invent: a broker account exists because the broker was told about it, and the password is
// whatever was agreed with the broker at that moment. The mesh's job is to carry it to the machine
// that will use it without being able to read it afterwards.
//
// Sealed on the way in and the plaintext discarded, exactly as a generated one is — so the only
// difference between the two is where the value came from.
func (i *Inventory) AcceptSecretForModule(ctx context.Context, node, module, name, value string) error {
	// Refused for a name the module does not declare. A value stored under a name nothing reads
	// is a delivery that changed nothing and reported success — the shape of failure the mesh
	// is built to refuse (novox/hq 04-ISSUES/078).
	m, err := i.declared(ctx, module)
	if err != nil {
		return err
	}
	if _, own := m.OwnSecrets[name]; !own {
		return fmt.Errorf("%s does not declare %q as an own secret; %s", module, name, declaresOwn(m))
	}
	key, err := i.SealingKeyOf(ctx, node)
	if err != nil {
		return err
	}
	if key == "" {
		return fmt.Errorf(
			"%s has no sealing key, so nothing can be sealed to it — it joins again to get one",
			node)
	}
	record, err := i.NodeByName(ctx, node)
	if err != nil {
		return err
	}

	sealed, err := secrets.Accept(value, key, key)
	if err != nil {
		return err
	}
	// And to the operator, when the mesh has one: a value a person supplied is the one a person
	// most needs to get back, since the mesh cannot make another (novox/hq ADR 0085, amended).
	forOperator, operatorKey, err := i.operatorSeal(ctx, value)
	if err != nil {
		return err
	}
	_, err = i.store.Pool().Exec(ctx,
		`insert into module_secret (node, module, name, sealed, node_key, origin, operator_sealed, operator_key)
		 values ($1, $2, $3, $4, $5, 'accepted', $6, $7)
		 on conflict (node, module, name) do update set
		   sealed = excluded.sealed, node_key = excluded.node_key,
		   origin = excluded.origin, made_at = now(),
		   operator_sealed = excluded.operator_sealed, operator_key = excluded.operator_key`,
		record.ID, module, name, sealed.ForConsumer, key, forOperator, operatorKey)
	return err
}

// Holder is one end-to-end credential: who gets it and who must create it.
type Holder struct {
	Provision string
	Consumer  string
	// ConsumerModule is which module on that machine holds it. Part of what identifies a
	// credential (novox/hq 04-ISSUES/022), so rotating one consumer's does not touch another's.
	ConsumerModule string
	// Local is the credential's name inside the consumer where it holds several (ADR 0094).
	Local    string
	Provider string
}

// HoldersOf is every pair sharing a credential for one provision.
//
// **The question rotation has to ask, and the one HAL could not.** There, a provision had a single
// shared credential and rotating it updated the provider's row; nothing enumerated who else held
// the old one, so three nodes carried dead credentials for two days and the mesh reported success
// (novox/hq ADR 0001). Here each pair has its own credential, and this is the list that makes
// "every consumer" a set the mesh can name rather than a hope.
//
// Empty consumer means all of them.
func (i *Inventory) HoldersOf(ctx context.Context, provision, consumer string) ([]Holder, error) {
	rows, err := i.store.Pool().Query(ctx,
		`select s.name, c.name, s.consumer_module, s.local, p.name from secret s
		   join node c on c.id = s.consumer
		   join node p on p.id = s.provider
		  where s.name = $1 and ($2 = '' or c.name = $2)
		  order by c.name, s.consumer_module, s.local, p.name`, provision, consumer)
	if err != nil {
		return nil, err
	}
	defer rows.Close()

	var out []Holder
	for rows.Next() {
		var h Holder
		if err := rows.Scan(&h.Provision, &h.Consumer, &h.ConsumerModule, &h.Local, &h.Provider); err != nil {
			return nil, err
		}
		out = append(out, h)
	}
	return out, rows.Err()
}

// localFlag is the `--local` a remedy has to name where a credential has a local name.
func localFlag(local string) string {
	if local == "" {
		return ""
	}
	return " --local " + local
}

// declared is the manifest the mesh holds for a module — what a delivered value is checked
// against, so a delivery for a name the module does not have is refused rather than stored.
func (i *Inventory) declared(ctx context.Context, module string) (catalogue.Manifest, error) {
	known, err := i.Catalogue(ctx)
	if err != nil {
		return catalogue.Manifest{}, err
	}
	m, ok := known[module]
	if !ok {
		return catalogue.Manifest{}, fmt.Errorf("%s is not a module the mesh knows; `module add` it first", module)
	}
	return m, nil
}

func declaresOwn(m catalogue.Manifest) string {
	if len(m.OwnSecrets) == 0 {
		return "it declares no own secrets"
	}
	return "it declares: " + strings.Join(sortedNames(m.OwnSecrets.Paths()), ", ")
}

func sortedNames(of map[string]string) []string {
	names := make([]string, 0, len(of))
	for name := range of {
		names = append(names, name)
	}
	sort.Strings(names)
	return names
}

func orNone(names []string) string {
	if len(names) == 0 {
		return "none"
	}
	return strings.Join(names, ", ")
}

// ErrNotRotatable says why the mesh will not rotate a module's own secret; the words are the caller's
// to print, and the remedy is in them.
type ErrNotRotatable struct{ Why string }

func (e ErrNotRotatable) Error() string { return e.Why }

// RotateModuleSecret makes a module's own secret anew, the way the first mint did (novox/hq
// ADR 0114, issue 180). The caller sends the node, so the module is started again on the new value.
//
// **Only a secret the module reads when it starts.** A secret the module's code applies to a
// backend that takes it once would, rotated this way, leave the backend on the old value and the
// module reading the new one — the fault issue 179 was. That form is staged, which the mesh does
// not build yet, and is refused by name. A secret whose manifest says neither is refused with the
// word to write; a secret given to the mesh rather than made by it is refused as 0113 says: the
// mesh will not replace what it cannot read.
func (i *Inventory) RotateModuleSecret(ctx context.Context, node, module, name string) error {
	m, err := i.declared(ctx, module)
	if err != nil {
		return err
	}
	own, declared := m.OwnSecrets[name]
	if !declared {
		return fmt.Errorf("%s does not declare %q as an own secret; %s", module, name, declaresOwn(m))
	}
	switch own.Taken {
	case catalogue.TakenAtStart:
	case catalogue.TakenApplied:
		return ErrNotRotatable{Why: fmt.Sprintf(
			"%s applies %q to a backend that takes it once, so a rotation must be staged beside the "+
				"current value until the module confirms it — the mesh does not do that yet (ADR 0114). "+
				"Changing it is a person's work: change it in %s, then `secret accept %s %s %s`",
			module, name, module, node, module, name)}
	default:
		return ErrNotRotatable{Why: fmt.Sprintf(
			"%s does not say how it takes %q, so the mesh will not rotate it: a secret rotated under "+
				"software that never reads it again is worse than one left alone. Its definition says "+
				"\"own-secrets\": {%q: {\"path\": …, \"taken\": \"at-start\"}} when the module reads it as it "+
				"starts, or \"applied\" when its own code applies it",
			module, name, name)}
	}
	record, err := i.NodeByName(ctx, node)
	if err != nil {
		return err
	}
	key, err := i.SealingKeyOf(ctx, node)
	if err != nil {
		return err
	}
	if key == "" {
		return fmt.Errorf("%s has no sealing key, so nothing can be sealed to it", node)
	}
	var origin string
	err = i.store.Pool().QueryRow(ctx,
		`select origin from module_secret where node = $1 and module = $2 and name = $3`,
		record.ID, module, name).Scan(&origin)
	if errors.Is(err, pgx.ErrNoRows) {
		return fmt.Errorf("%s on %s holds no %q yet; the first push makes it", module, node, name)
	}
	if err != nil {
		return err
	}
	if origin == OriginAccepted {
		return ErrNotRotatable{Why: fmt.Sprintf(
			"%s on %s holds %q as a value given to the mesh, not made by it, and the mesh will not "+
				"replace what it cannot read (ADR 0113). Change it where it lives, then `secret accept "+
				"%s %s %s` with the new value",
			module, node, name, node, module, name)}
	}
	operator, err := i.OperatorKey(ctx)
	if err != nil {
		return err
	}
	made, blob, err := secrets.MakeWithOperator(key, key, operator)
	if err != nil {
		return err
	}
	forOperator, operatorKey := operatorColumns(operator, blob)
	_, err = i.store.Pool().Exec(ctx,
		`update module_secret set sealed = $4, node_key = $5, origin = 'made', made_at = now(),
		        operator_sealed = $6, operator_key = $7
		  where node = $1 and module = $2 and name = $3`,
		record.ID, module, name, made.ForConsumer, key, forOperator, operatorKey)
	return err
}