mesh-delivery/modules/mesh-delivery-assigner/internal/assign/wants.go

package assign

import (
	"fmt"
	"slices"
	"sort"

	"git.novox.be/novox/mesh-sdk/go/catalogue"
	"git.novox.be/novox/mesh-sdk/go/delivery"
)

// Module is what the six rules read of one module (ADR 0429 §6, ADR 0528 clause 3): its newest fingerprint F at the
// trunk (the indexer's modules), whether the builder's build of F is done, whether F is held, whether the controller
// issued its credential for F, its manifest at F, and the nodes it is assigned to.
type Module struct {
	Name     string
	FP       string
	Built    bool
	Held     bool
	Granted  bool
	Manifest catalogue.Manifest
	Nodes    []string
}

// Node is what the rules read of the nodes: whether each is online (the watchman's presence) and each module's have.
type Nodes struct {
	Online map[string]bool
	Haves  map[string]map[string]delivery.Have
	// Providers is, by provision, the modules that provide it, with the nodes each is assigned to.
	Providers map[string]map[string][]string
}

// healthy says whether a have runs fp healthy.
func healthy(h delivery.Have, fp string) bool {
	return h.FP == fp && h.Phase == delivery.PhaseRunning && h.Health == delivery.HealthOK
}

// Canary is the module's canary: the first of its nodes, by name, that is online (ADR 0435 §3; one ring, ordered by
// name, ADR 0528 clause 3). Empty when none is online.
func Canary(nodes []string, online map[string]bool) string {
	sorted := slices.Sorted(slices.Values(nodes))
	for _, n := range sorted {
		if online[n] {
			return n
		}
	}
	return ""
}

// Holds is whether every rule holds for moving node's want for m to m.FP, and if not, the first rule that does not,
// with why (ADR 0429 §6). Rule 4, at most `step` others ahead, always holds: the mesh keeps no step, and step 0 is no
// limit (ADR 0528 clause 3).
func Holds(m Module, node string, n Nodes) (bool, int, string) {
	switch {
	case !m.Built:
		return false, delivery.RuleBuildDone, fmt.Sprintf("the build of %s at %s is not done", m.Name, short(m.FP))
	case m.Held:
		return false, delivery.RuleNotHeld, fmt.Sprintf("%s is held at %s", m.Name, short(m.FP))
	}
	canary := Canary(m.Nodes, n.Online)
	switch {
	case canary == "":
		return false, delivery.RuleCanary, fmt.Sprintf("none of the nodes of %s is online, so it has no canary", m.Name)
	case node != canary && !healthy(n.Haves[canary][m.Name], m.FP):
		return false, delivery.RuleCanary, fmt.Sprintf("the canary %s does not run %s healthy yet", canary, short(m.FP))
	}
	for _, provision := range m.Manifest.Requires {
		if ok, why := providedHealthy(provision, node, n); !ok {
			return false, delivery.RuleProviders, why
		}
	}
	if !m.Granted {
		return false, delivery.RuleGrant, fmt.Sprintf("the controller has not issued the credential of %s for %s", m.Name, short(m.FP))
	}
	return true, 0, ""
}

// providedHealthy is rule 5 for one provision: some module providing it runs healthy on the node, or, assigned
// elsewhere only, at its canary. A provision no module provides is not a module's to judge (the mesh or the node
// offers it) and holds.
func providedHealthy(provision, node string, n Nodes) (bool, string) {
	providers := n.Providers[provision]
	if len(providers) == 0 {
		return true, ""
	}
	names := make([]string, 0, len(providers))
	for p := range providers {
		names = append(names, p)
	}
	sort.Strings(names)
	for _, p := range names {
		nodes := providers[p]
		at := node
		if !slices.Contains(nodes, node) {
			at = Canary(nodes, n.Online)
		}
		if h, ok := n.Haves[at][p]; ok && at != "" && h.Phase == delivery.PhaseRunning && h.Health == delivery.HealthOK {
			return true, ""
		}
	}
	return false, fmt.Sprintf("no provider of %s runs healthy on %s or at its canary", provision, node)
}

// keepsData says whether a module holds data a set-back would lose: an own item of a class that keeps (irreplaceable,
// valuable or rebuildable; cache and none do not).
func keepsData(m catalogue.Manifest) bool {
	for _, it := range m.DataItems() {
		if it.Class == catalogue.ClassIrreplaceable || it.Class == catalogue.ClassValuable || it.Class == catalogue.ClassRebuildable {
			return true
		}
	}
	return false
}

// Outcome is what one module's pass decides: each node's want to write (only those that change), and a hold to write.
type Outcome struct {
	Wants map[string]delivery.Want
	// Held is the hold of m.FP to write, when a node's health check failed on it and it is not held yet.
	Held *delivery.Held
	// Rules is, by node, the rule holding the node back from m.FP and why, for the comparison.
	Rules map[string]Holding
}

// Holding is a rule holding a node back, and why.
type Holding struct {
	Rule  int
	Words string
}

// Wants is the pass of one module m: for each of its nodes, the want by the six rules, starting from the want written
// before (wants, by node); lastHealthy is, by node, the newest fingerprint other than m.FP the node ran healthy. A failed
// health check of m.FP on a node holds it: every node wanting m.FP is set back to its last healthy fingerprint, but a
// module holding data is not set back (in release 5 it is asked about from release 6 on: wants are only compared). A
// node that wants m but is no longer among its nodes gets its want written removed, never deleted (ADR 0408 §4).
func Wants(m Module, n Nodes, wants map[string]delivery.Want, lastHealthy map[string]string) Outcome {
	out := Outcome{Wants: map[string]delivery.Want{}, Rules: map[string]Holding{}}
	held := m.Held
	if !held && m.FP != "" {
		for _, node := range slices.Sorted(slices.Values(m.Nodes)) {
			if h, ok := n.Haves[node][m.Name]; ok && h.FP == m.FP && h.Health == delivery.HealthFailed {
				out.Held = &delivery.Held{Node: node, Words: h.Detail}
				held = true
				break
			}
		}
	}
	mm := m
	mm.Held = held
	for _, node := range m.Nodes {
		if m.FP == "" {
			break
		}
		before, had := wants[node]
		var next delivery.Want
		switch ok, rule, words := Holds(mm, node, n); {
		case ok:
			next = delivery.Want{FP: m.FP}
		case held && had && before.FP == m.FP:
			out.Rules[node] = Holding{Rule: rule, Words: words}
			last := lastHealthy[node]
			switch {
			case keepsData(m.Manifest):
				next = before
				next.Words = fmt.Sprintf("%s holds data, so it is not set back from %s: the operator is asked from release 6", m.Name, short(m.FP))
			case last == "":
				next = before
				next.Words = fmt.Sprintf("%s is held at %s and %s ran no other fingerprint healthy to set it back to", m.Name, short(m.FP), node)
			default:
				next = delivery.Want{FP: last, Newest: m.FP, Rule: rule, Words: words}
			}
		default:
			out.Rules[node] = Holding{Rule: rule, Words: words}
			if !had || before.Removed {
				continue
			}
			next = delivery.Want{FP: before.FP, Newest: m.FP, Rule: rule, Words: words}
			if before.FP == m.FP {
				next = delivery.Want{FP: m.FP}
			}
		}
		if !had || !delivery.SameWant(normal(before), normal(next)) {
			out.Wants[node] = next
		}
	}
	for node, before := range wants {
		if !slices.Contains(m.Nodes, node) && !before.Removed {
			out.Wants[node] = delivery.Want{Removed: true, Words: fmt.Sprintf("%s is no longer assigned to %s", m.Name, node)}
		}
	}
	return out
}

// normal is a want without its version, as a decoded and a composed one are compared.
func normal(w delivery.Want) delivery.Want {
	w.V = 0
	return w
}

func short(fp string) string {
	part, err := delivery.FingerprintKeyPart(fp)
	if err != nil || len(part) < 12 {
		return fp
	}
	return part[:12]
}