docs: strip over-commenting from server.go and reports.go

reports.go

@@ -9,12 +9,7 @@ import (
 	"strings"
 )
 
-// mergeUnion concatenates every backend's records and drops duplicates by key.
-// Reports and events are immutable history, so a certname that migrated between
-// PuppetDBs legitimately has records in both and the union — not a per-node
-// winner — is the correct merged view. results must be ordered by precedence;
-// the first backend holding a key supplies the record. A key func returning
-// ok=false means the record has no dedupe identity and is always kept.
+// results must be ordered by precedence; a key func returning ok=false means the record has no identity and is always kept.
 func mergeUnion(results []backendResult, key func(record) (string, bool)) []json.RawMessage {
 	seen := make(map[string]bool)
 	out := []json.RawMessage{}
@@ -32,10 +27,7 @@ func mergeUnion(results []backendResult, key func(record) (string, bool)) []json
 	return out
 }
 
-// reportKey identifies a report by its content hash, which PuppetDB guarantees
-// is unique per report. An `extract`/`group_by` query returns synthetic rows
-// with no hash and no identity — two backends can emit byte-identical aggregate
-// rows that both count — so those are never deduped.
+// extract/group_by rows are synthetic and carry no hash, so two backends can legitimately emit identical ones.
 func reportKey(rec record) (string, bool) {
 	if rec.Hash == "" {
 		return "", false
@@ -43,19 +35,15 @@ func reportKey(rec record) (string, bool) {
 	return "hash\x00" + rec.Hash, true
 }
 
-// rawKey identifies a record by its verbatim JSON. Events carry no unique id,
-// but two byte-identical events from the same PuppetDB serialiser describe the
-// same resource change, so raw equality is a safe dedupe key.
+// Events carry no id, but byte-identical events from the same PuppetDB serialiser are the same change.
 func rawKey(rec record) (string, bool) { return "raw\x00" + string(rec.Raw), true }
 
-// orderField is one entry of PuppetDB's order_by param.
 type orderField struct {
 	Field string
 	Desc  bool
 }
 
-// parseOrderBy decodes PuppetDB's order_by param, a JSON array of
-// {"field":..., "order":"asc"|"desc"} objects. An empty param yields no fields.
+// order_by is a JSON array of {"field": ..., "order": "asc"|"desc"} objects.
 func parseOrderBy(s string) ([]orderField, error) {
 	if strings.TrimSpace(s) == "" {
 		return nil, nil
@@ -77,9 +65,7 @@ func parseOrderBy(s string) ([]orderField, error) {
 	return out, nil
 }
 
-// sortRecords re-sorts a merged record set by order. Each backend only ordered
-// its own slice, so the union has to be ordered again here. The sort is stable,
-// so ties keep backend precedence order.
+// Each backend ordered only its own slice, so the union is re-sorted here; stable, so ties keep backend precedence.
 func sortRecords(recs []json.RawMessage, order []orderField) {
 	if len(order) == 0 || len(recs) < 2 {
 		return
@@ -113,8 +99,7 @@ func sortRecords(recs []json.RawMessage, order []orderField) {
 	copy(recs, sorted)
 }
 
-// compareValues orders two decoded JSON values. Unlike types are ordered by
-// kind (null < bool < number < string) so a missing field always sorts first.
+// Unlike types order by kind (null < bool < number < string), so a missing field sorts first.
 func compareValues(a, b any) int {
 	ra, rb := valueRank(a), valueRank(b)
 	if ra != rb {
@@ -165,9 +150,6 @@ func valueRank(v any) int {
 	}
 }
 
-// paging holds the PuppetDB paging params a merged endpoint has to re-apply
-// itself: each backend applies limit/offset to its own result set only, so the
-// proxy must page the union instead.
 type paging struct {
 	limit     int // -1 when unset
 	offset    int
@@ -175,8 +157,6 @@ type paging struct {
 	wantTotal bool
 }
 
-// parsePaging reads limit, offset, order_by and include_total from a request's
-// query params.
 func parsePaging(v url.Values) (paging, error) {
 	p := paging{limit: -1}
 	if s := v.Get("limit"); s != "" {
@@ -202,9 +182,7 @@ func parsePaging(v url.Values) (paging, error) {
 	return p, nil
 }
 
-// upstreamParams rewrites the client's params for the fan-out. A backend must
-// return everything that could land in the merged page, so it is asked for the
-// first offset+limit records and the offset is applied locally instead.
+// Backends are asked for the first offset+limit records with no offset; the offset is applied to the union instead.
 func (p paging) upstreamParams(in url.Values) url.Values {
 	out := url.Values{}
 	for k, vs := range in {
@@ -217,7 +195,6 @@ func (p paging) upstreamParams(in url.Values) url.Values {
 	return out
 }
 
-// apply slices the merged, ordered record set down to the requested page.
 func (p paging) apply(recs []json.RawMessage) []json.RawMessage {
 	if p.offset >= len(recs) {
 		return []json.RawMessage{}
@@ -229,9 +206,7 @@ func (p paging) apply(recs []json.RawMessage) []json.RawMessage {
 	return recs
 }
 
-// sumTotals adds up the X-Records counts the backends reported, ignoring any
-// backend that did not send one. It returns -1 when no backend reported a count.
-// Deduped records are counted once per backend, so the total is an upper bound.
+// Returns -1 when no backend reported a count; duplicates count once per backend, so the sum is an upper bound.
 func sumTotals(results []backendResult) int {
 	total := -1
 	for _, res := range results {

server.go

@@ -22,14 +22,10 @@ const (
 	eventsPath  = "/pdb/query/v4/events"
 	queryV4     = "/pdb/query/v4/"
 
-	// recordsHeader is PuppetDB's total-result-count header, returned when a
-	// request carries include_total=true.
+	// PuppetDB only sends this when the request carries include_total=true.
 	recordsHeader = "X-Records"
 )
 
-// backendResult is one backend's decoded response for a query. err is non-nil
-// when the backend failed (network/timeout/non-2xx); such results carry no
-// records and are excluded from the merge but logged.
 type backendResult struct {
 	name    string
 	records []record
@@ -37,7 +33,6 @@ type backendResult struct {
 	err     error
 }
 
-// Server proxies and merges PuppetDB queries across the configured backends.
 type Server struct {
 	cfg    Config
 	client *http.Client
@@ -49,7 +44,6 @@ type Server struct {
 	freshAt   time.Time
 }
 
-// NewServer builds a Server with an HTTP client bounded by cfg.Timeout.
 func NewServer(cfg Config, logger *log.Logger) *Server {
 	return &Server{
 		cfg:    cfg,
@@ -58,7 +52,6 @@ func NewServer(cfg Config, logger *log.Logger) *Server {
 	}
 }
 
-// Handler returns the HTTP mux for the proxy.
 func (s *Server) Handler() http.Handler {
 	mux := http.NewServeMux()
 	mux.HandleFunc("/healthz", s.handleHealth)
@@ -66,10 +59,6 @@ func (s *Server) Handler() http.Handler {
 	return mux
 }
 
-// handleQuery dispatches /pdb/query/v4/* requests: /facts and /nodes are merged
-// per node, /reports and /events are unioned across backends, a report's
-// sub-resources resolve to whichever backend stores that report, and every other
-// v4 path is transparently proxied to the primary.
 func (s *Server) handleQuery(w http.ResponseWriter, r *http.Request) {
 	if r.Method != http.MethodGet {
 		http.Error(w, "only GET is supported", http.StatusMethodNotAllowed)
@@ -93,9 +82,7 @@ func (s *Server) handleQuery(w http.ResponseWriter, r *http.Request) {
 	}
 }
 
-// isReportSubResource reports whether path is a per-report child endpoint —
-// /pdb/query/v4/reports/<hash>/{events,logs,metrics} — whose data lives in
-// exactly one backend.
+// Matches /pdb/query/v4/reports/<hash>/{events,logs,metrics}, whose data lives in exactly one backend.
 func isReportSubResource(path string) bool {
 	rest, ok := strings.CutPrefix(path, reportsPath+"/")
 	if !ok {
@@ -112,9 +99,6 @@ func isReportSubResource(path string) bool {
 	return false
 }
 
-// serveMerged fans out the request to all backends, then hands the per-backend
-// results to merge to produce the response body. If every backend fails it
-// returns 502; if some fail it serves the survivors and logs a warning.
 func (s *Server) serveMerged(w http.ResponseWriter, r *http.Request, path string, merge func([]backendResult) []json.RawMessage) {
 	alive, ok := s.aliveResults(w, r, path, queryParams(r.URL.Query().Get("query")))
 	if !ok {
@@ -123,10 +107,7 @@ func (s *Server) serveMerged(w http.ResponseWriter, r *http.Request, path string
 	writeJSON(w, merge(alive))
 }
 
-// serveUnion fans out a request whose records are immutable history — reports
-// and events — and serves the deduped union of every backend. Because each
-// backend ordered and paged only its own slice, the union is re-ordered and
-// re-paged here from the client's order_by/limit/offset.
+// Reports and events are immutable history, so both backends' records belong in the merged view.
 func (s *Server) serveUnion(w http.ResponseWriter, r *http.Request, path string, key func(record) (string, bool)) {
 	in := r.URL.Query()
 	page, err := parsePaging(in)
@@ -150,11 +131,7 @@ func (s *Server) serveUnion(w http.ResponseWriter, r *http.Request, path string,
 	writeJSON(w, page.apply(merged))
 }
 
-// serveFirstHolder answers a per-report sub-resource request. The report lives
-// in exactly one backend, so all are asked concurrently and the first one (in
-// precedence order) that actually holds it wins. Backends that do not have the
-// report answer 404, which is indistinguishable here from any other failure, so
-// an empty result is only served once every backend has been consulted.
+// A backend without the report answers 404, indistinguishable from a failure, so every backend is consulted before serving empty.
 func (s *Server) serveFirstHolder(w http.ResponseWriter, r *http.Request) {
 	results := s.fanOut(r.Context(), r.URL.Path, r.URL.Query())
 
@@ -179,8 +156,7 @@ func (s *Server) serveFirstHolder(w http.ResponseWriter, r *http.Request) {
 	writeJSON(w, nil)
 }
 
-// aliveResults fans out to every backend and returns the successful results.
-// It writes a 502 and returns ok=false when every backend failed.
+// Writes a 502 and returns ok=false only when every backend failed.
 func (s *Server) aliveResults(w http.ResponseWriter, r *http.Request, path string, params url.Values) ([]backendResult, bool) {
 	results := s.fanOut(r.Context(), path, params)
 
@@ -199,8 +175,6 @@ func (s *Server) aliveResults(w http.ResponseWriter, r *http.Request, path strin
 	return alive, true
 }
 
-// queryParams builds the upstream param set for a merged endpoint that only
-// forwards the PuppetDB query.
 func queryParams(query string) url.Values {
 	if query == "" {
 		return nil
@@ -208,7 +182,6 @@ func queryParams(query string) url.Values {
 	return url.Values{"query": []string{query}}
 }
 
-// rawRecords strips decoded metadata back down to the verbatim JSON elements.
 func rawRecords(recs []record) []json.RawMessage {
 	out := make([]json.RawMessage, 0, len(recs))
 	for _, rec := range recs {
@@ -217,21 +190,16 @@ func rawRecords(recs []record) []json.RawMessage {
 	return out
 }
 
-// mergeNodesResponse merges /nodes results (dedupe by certname, newer wins).
 func (s *Server) mergeNodesResponse(results []backendResult) []json.RawMessage {
 	return mergeNodes(s.byPrecedence(results))
 }
 
-// mergeFactsResponse merges /facts results at node granularity, choosing each
-// certname's owner via the configured merge strategy.
 func (s *Server) mergeFactsResponse(results []backendResult) []json.RawMessage {
 	ordered := s.byPrecedence(results)
 	if s.cfg.Merge == mergeStatic {
 		prefer := s.cfg.Prefer
 		return mergeFacts(ordered, func(string) string { return prefer })
 	}
-	// freshness merge: attribute each certname to the backend with the newer
-	// report_timestamp, taken from a short-TTL /nodes freshness map.
 	fresh := s.freshnessMap(context.Background(), ordered)
 	prefer := s.cfg.Prefer
 	return mergeFacts(ordered, func(cn string) string {
@@ -242,8 +210,7 @@ func (s *Server) mergeFactsResponse(results []backendResult) []json.RawMessage {
 	})
 }
 
-// byPrecedence orders results so the Prefer backend comes first, giving it the
-// tie-break on equal timestamps. Remaining backends keep config order.
+// Puts Prefer first so it wins ties; the rest keep config order.
 func (s *Server) byPrecedence(results []backendResult) []backendResult {
 	ordered := make([]backendResult, len(results))
 	copy(ordered, results)
@@ -253,15 +220,7 @@ func (s *Server) byPrecedence(results []backendResult) []backendResult {
 	return ordered
 }
 
-// freshnessMap returns a per-certname owner map derived from each backend's
-// /nodes report_timestamp, cached for cfg.FreshnessTTL. On cache miss it queries
-// /nodes from all backends; a backend that fails is simply absent from the map,
-// so its certnames fall back to precedence/Prefer.
-//
-// When the incoming request already carries /nodes data (results has records),
-// we still query /nodes broadly here because a /facts query's certname set can
-// differ from what the request's query filter returned. The cache keeps this
-// cheap under load.
+// Queries /nodes unfiltered rather than reusing the request's results, because a /facts query's certname set can differ.
 func (s *Server) freshnessMap(ctx context.Context, _ []backendResult) freshness {
 	s.mu.Lock()
 	if s.freshData != nil && time.Since(s.freshAt) < s.cfg.FreshnessTTL {
@@ -271,7 +230,6 @@ func (s *Server) freshnessMap(ctx context.Context, _ []backendResult) freshness
 	}
 	s.mu.Unlock()
 
-	// Empty query = all nodes; cheap enough for a short-TTL cache.
 	nodeResults := s.fanOut(ctx, nodesPath, nil)
 	var alive []backendResult
 	for _, res := range nodeResults {
@@ -290,8 +248,7 @@ func (s *Server) freshnessMap(ctx context.Context, _ []backendResult) freshness
 	return f
 }
 
-// fanOut queries every backend concurrently for path with the given params and
-// returns one backendResult per backend, in config order.
+// Returns one result per backend, in config order.
 func (s *Server) fanOut(ctx context.Context, path string, params url.Values) []backendResult {
 	results := make([]backendResult, len(s.cfg.Backends))
 	var wg sync.WaitGroup
@@ -307,8 +264,7 @@ func (s *Server) fanOut(ctx context.Context, path string, params url.Values) []b
 	return results
 }
 
-// queryBackend performs one GET b.URL+path?params and decodes the JSON array.
-// It also returns the upstream X-Records count, or -1 when the backend sent none.
+// The returned count is the upstream X-Records value, or -1 when the backend sent none.
 func (s *Server) queryBackend(ctx context.Context, b Backend, path string, params url.Values) ([]record, int, error) {
 	target := strings.TrimRight(b.URL, "/") + path
 	req, err := http.NewRequestWithContext(ctx, http.MethodGet, target, nil)
@@ -338,8 +294,6 @@ func (s *Server) queryBackend(ctx context.Context, b Backend, path string, param
 	return recs, total, err
 }
 
-// proxyPrimary transparently forwards a non-merged /pdb/query/v4/* request to
-// the primary backend and streams the response back verbatim.
 func (s *Server) proxyPrimary(w http.ResponseWriter, r *http.Request) {
 	b := s.cfg.PrimaryBackend()
 	target := strings.TrimRight(b.URL, "/") + r.URL.Path
@@ -365,15 +319,11 @@ func (s *Server) proxyPrimary(w http.ResponseWriter, r *http.Request) {
 	_, _ = io.Copy(w, resp.Body)
 }
 
-// healthReport is the /healthz JSON body.
 type healthReport struct {
 	Status   string            `json:"status"`
 	Backends map[string]string `json:"backends"` // name -> "ok" | error text
 }
 
-// handleHealth probes every backend's /nodes endpoint with a trivial query and
-// reports per-backend reachability. Overall status is "ok" if any backend is
-// reachable, "degraded" if some fail, "down" if all fail (503 in that case).
 func (s *Server) handleHealth(w http.ResponseWriter, r *http.Request) {
 	probe := `["=","certname","pdbmux-healthz-probe"]`
 	results := s.fanOut(r.Context(), nodesPath, queryParams(probe))
@@ -406,7 +356,6 @@ func (s *Server) handleHealth(w http.ResponseWriter, r *http.Request) {
 	_ = enc.Encode(report)
 }
 
-// writeJSON writes a JSON array of raw records as a PuppetDB-style response.
 func writeJSON(w http.ResponseWriter, recs []json.RawMessage) {
 	w.Header().Set("Content-Type", "application/json")
 	if recs == nil {