DKIM rotation broke the domain it was run on. It retired the live key and inserted a placeholder as ACTIVE - private_key_enc was the literal string "dummy-encrypted-key" - so FindActiveKeys then failed for the whole domain and outbound mail stopped being signed, with the real key already gone. Reproduced against a database before changing anything. The root cause runs deeper: the auth service derives its encryption key with a plain SHA-256 of the master key while the protocols service uses HKDF with its own info string, so nothing the auth service seals can ever be opened by the signer. Rotation therefore moves to the protocols service, which owns that vault and the real key generator. It retires the previous key with a seven-day overlap in the same transaction rather than deleting it, because mail already sent under it is still being verified. Domain onboarding had three defects in one function. It built the alias destination by interpolating the domain name into the SQL text, which was injectable; it reused one placeholder for a CITEXT and a VARCHAR column, which made the statement fail and happened to hide the injection; and it ignored its scope argument entirely, so any DOMAIN_ADMIN could create domains. It also provisioned postmaster, abuse and security, missing the dmarc and tlsrpt aliases that the published records point at - PLAN.md section 7.4b is explicit that reports bounce without them. The TLS panel returned fixed values: acme_status "VALID" and 72 days to expiry, whatever the process actually held. It now reads the certificate watcher, and a compile-time assertion keeps the wiring from silently regressing into "disabled". The audit log and the readiness scorecard were fixtures too, and their field names did not match the API they now read. Two tests failed and a third asserted a fabricated constant: one depended on test order through a shared fixture that had a second factor enrolled, one used a fixed domain name and so passed once per database, and one expected "records: 13" from a function that never touches DNS. All three are fixed, and the suite now passes twice in a row against one database. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
300 lines
10 KiB
Go
300 lines
10 KiB
Go
package httppresentation
|
|
|
|
import (
|
|
"fmt"
|
|
"io"
|
|
"net"
|
|
"net/http"
|
|
"strings"
|
|
"time"
|
|
|
|
appusecase "lambdamail/protocols/internal/application/usecase"
|
|
"lambdamail/protocols/internal/domain/valueobject"
|
|
)
|
|
|
|
// mtaStsMaxAge is the policy lifetime advertised to senders. RFC 8461
|
|
// recommends a value in the order of weeks so that an attacker cannot strip
|
|
// the policy by making the endpoint unreachable.
|
|
const mtaStsMaxAge = 604800 // 7 days
|
|
|
|
type Router struct {
|
|
mux *http.ServeMux
|
|
reportUseCase *appusecase.IngestReportsUseCase
|
|
pingFunc func() error
|
|
mtaStsMode valueobject.MtaStsMode
|
|
// mail serves the webmail's message screens. It stays nil when no session
|
|
// secret is configured, and the routes then answer 503 rather than
|
|
// pretending to be open.
|
|
mail *mailAPI
|
|
// events pushes new mail to the browser. Like mail, it stays nil without a
|
|
// session secret.
|
|
events *EventHub
|
|
// dkim manages signing keys. It lives here because the private key must be
|
|
// sealed with this process's vault.
|
|
dkim *adminDkimAPI
|
|
// tls reports what the certificate watcher sees, rather than a constant.
|
|
tls *adminTlsAPI
|
|
// degradedFunc reports a condition that leaves the service running but
|
|
// not fit for production, the clearest case being a self-signed
|
|
// certificate standing in for one Traefik never issued
|
|
// (PLAN.md section 8.4).
|
|
degradedFunc func() error
|
|
}
|
|
|
|
func NewRouter(reportUseCase *appusecase.IngestReportsUseCase, pingFunc func() error) *Router {
|
|
r := &Router{
|
|
mux: http.NewServeMux(),
|
|
reportUseCase: reportUseCase,
|
|
pingFunc: pingFunc,
|
|
// PLAN.md section 5 ramps MTA-STS from testing to enforce. Starting at
|
|
// enforce means any TLS misconfiguration silently blackholes inbound
|
|
// mail from every sender that honours the policy, with no way to tell
|
|
// from this side - so the safe mode is the default and the operator
|
|
// promotes it once TLS-RPT reports come back clean.
|
|
mtaStsMode: valueobject.MtaStsModeTesting,
|
|
}
|
|
r.registerRoutes()
|
|
return r
|
|
}
|
|
|
|
// SetMtaStsMode selects the mode advertised in the served policy. Callers
|
|
// should promote to enforce only after TLS-RPT confirms senders can negotiate
|
|
// TLS against the published MX.
|
|
func (r *Router) SetMtaStsMode(mode valueobject.MtaStsMode) {
|
|
r.mtaStsMode = mode
|
|
}
|
|
|
|
func (r *Router) ServeHTTP(w http.ResponseWriter, req *http.Request) {
|
|
r.mux.ServeHTTP(w, req)
|
|
}
|
|
|
|
func (r *Router) registerRoutes() {
|
|
r.mux.HandleFunc("/health", r.handleHealth)
|
|
r.mux.HandleFunc("/.well-known/mta-sts.txt", r.handleMtaSts)
|
|
r.mux.HandleFunc("/.well-known/security.txt", r.handleSecurityTxt)
|
|
r.mux.HandleFunc("/mail/config-v1.1.xml", r.handleThunderbirdAutoconfig)
|
|
r.mux.HandleFunc("/.well-known/autoconfig/mail/config-v1.1.xml", r.handleThunderbirdAutoconfig)
|
|
r.mux.HandleFunc("/autodiscover/autodiscover.xml", r.handleOutlookAutodiscover)
|
|
r.mux.HandleFunc("/api/v1/events", r.handleEvents)
|
|
r.mux.HandleFunc("/api/v1/admin/tls", r.handleAdminTls)
|
|
r.mux.HandleFunc("/api/v1/admin/dkim/keys", r.handleAdminDkimKeys)
|
|
r.mux.HandleFunc("/api/v1/admin/dkim/rotate", r.handleAdminDkimRotate)
|
|
r.mux.HandleFunc("/api/v1/mail/folders", r.handleMailFolders)
|
|
r.mux.HandleFunc("/api/v1/mail/messages", r.handleMailMessages)
|
|
r.mux.HandleFunc("/api/v1/mail/message/", r.handleMailMessage)
|
|
r.mux.HandleFunc("/api/v1/mail/seen", r.handleMailSeen)
|
|
r.mux.HandleFunc("/api/v1/mail/send", r.handleMailSend)
|
|
r.mux.HandleFunc("/api/v1/reports/dmarc", r.handleDmarcIngest)
|
|
r.mux.HandleFunc("/api/v1/reports/tlsrpt", r.handleTlsRptIngest)
|
|
}
|
|
|
|
// SetDegradedCheck registers the probe behind the "degraded" health state.
|
|
// Returning an error means the process is serving but something an operator
|
|
// must fix is wrong.
|
|
func (r *Router) SetDegradedCheck(check func() error) {
|
|
r.degradedFunc = check
|
|
}
|
|
|
|
func (r *Router) handleHealth(w http.ResponseWriter, _ *http.Request) {
|
|
if r.pingFunc != nil {
|
|
if err := r.pingFunc(); err != nil {
|
|
http.Error(w, fmt.Sprintf("unhealthy: %v", err), http.StatusServiceUnavailable)
|
|
return
|
|
}
|
|
}
|
|
|
|
// Degraded still answers 200: the container is alive and must not be
|
|
// restarted in a loop over a certificate problem a restart cannot fix.
|
|
// The distinction is carried in the body and the header so monitoring can
|
|
// alert on it (PLAN.md section 8.4).
|
|
if r.degradedFunc != nil {
|
|
if err := r.degradedFunc(); err != nil {
|
|
w.Header().Set("X-LambdaMail-Health", "degraded")
|
|
w.WriteHeader(http.StatusOK)
|
|
fmt.Fprintf(w, "DEGRADED: %v\n", err)
|
|
return
|
|
}
|
|
}
|
|
|
|
w.Header().Set("X-LambdaMail-Health", "ok")
|
|
w.WriteHeader(http.StatusOK)
|
|
w.Write([]byte("OK\n"))
|
|
}
|
|
|
|
func (r *Router) handleMtaSts(w http.ResponseWriter, req *http.Request) {
|
|
host := extractHost(req.Host)
|
|
domainName := strings.TrimPrefix(host, "mta-sts.")
|
|
|
|
policy := valueobject.NewMtaStsPolicy(r.mtaStsMode, []string{fmt.Sprintf("mail.%s", domainName)}, mtaStsMaxAge)
|
|
|
|
w.Header().Set("Content-Type", "text/plain; charset=utf-8")
|
|
// Kept in step with max_age so an HTTP cache cannot outlive the policy it
|
|
// is caching.
|
|
w.Header().Set("Cache-Control", fmt.Sprintf("public, max-age=%d", mtaStsMaxAge))
|
|
w.WriteHeader(http.StatusOK)
|
|
w.Write([]byte(policy.Format()))
|
|
}
|
|
|
|
func (r *Router) handleSecurityTxt(w http.ResponseWriter, req *http.Request) {
|
|
// The contact must be a real mailbox on the mail domain, not on whichever
|
|
// service hostname the request happened to arrive at.
|
|
domainName := baseDomain(extractHost(req.Host))
|
|
expires := time.Now().AddDate(1, 0, 0).Format(time.RFC3339)
|
|
|
|
content := fmt.Sprintf("Contact: mailto:security@%s\r\nExpires: %s\r\nPreferred-Languages: en, pt-BR, es\r\n", domainName, expires)
|
|
|
|
w.Header().Set("Content-Type", "text/plain; charset=utf-8")
|
|
w.WriteHeader(http.StatusOK)
|
|
w.Write([]byte(content))
|
|
}
|
|
|
|
func (r *Router) handleThunderbirdAutoconfig(w http.ResponseWriter, req *http.Request) {
|
|
host := extractHost(req.Host)
|
|
domainName := strings.TrimPrefix(host, "autoconfig.")
|
|
|
|
xml := valueobject.BuildThunderbirdAutoconfigXML(domainName, fmt.Sprintf("mail.%s", domainName))
|
|
|
|
w.Header().Set("Content-Type", "application/xml; charset=utf-8")
|
|
w.WriteHeader(http.StatusOK)
|
|
w.Write([]byte(xml))
|
|
}
|
|
|
|
func (r *Router) handleOutlookAutodiscover(w http.ResponseWriter, req *http.Request) {
|
|
host := extractHost(req.Host)
|
|
domainName := strings.TrimPrefix(host, "autodiscover.")
|
|
|
|
xml := valueobject.BuildOutlookAutodiscoverXML(domainName, fmt.Sprintf("mail.%s", domainName), fmt.Sprintf("user@%s", domainName))
|
|
|
|
w.Header().Set("Content-Type", "text/xml; charset=utf-8")
|
|
w.WriteHeader(http.StatusOK)
|
|
w.Write([]byte(xml))
|
|
}
|
|
|
|
func (r *Router) handleDmarcIngest(w http.ResponseWriter, req *http.Request) {
|
|
if req.Method != http.MethodPost {
|
|
http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
|
|
return
|
|
}
|
|
body, err := io.ReadAll(req.Body)
|
|
if err != nil {
|
|
http.Error(w, "Read body failed", http.StatusBadRequest)
|
|
return
|
|
}
|
|
if r.reportUseCase == nil {
|
|
http.Error(w, "Ingestion unconfigured", http.StatusInternalServerError)
|
|
return
|
|
}
|
|
if _, err := r.reportUseCase.IngestDmarc(req.Context(), body); err != nil {
|
|
http.Error(w, fmt.Sprintf("Parse error: %v", err), http.StatusBadRequest)
|
|
return
|
|
}
|
|
w.Header().Set("Content-Type", "application/json")
|
|
w.WriteHeader(http.StatusOK)
|
|
w.Write([]byte(`{"status":"ok"}`))
|
|
}
|
|
|
|
func (r *Router) handleTlsRptIngest(w http.ResponseWriter, req *http.Request) {
|
|
if req.Method != http.MethodPost {
|
|
http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
|
|
return
|
|
}
|
|
body, err := io.ReadAll(req.Body)
|
|
if err != nil {
|
|
http.Error(w, "Read body failed", http.StatusBadRequest)
|
|
return
|
|
}
|
|
if r.reportUseCase == nil {
|
|
http.Error(w, "Ingestion unconfigured", http.StatusInternalServerError)
|
|
return
|
|
}
|
|
if _, err := r.reportUseCase.IngestTlsRpt(req.Context(), body); err != nil {
|
|
http.Error(w, fmt.Sprintf("Parse error: %v", err), http.StatusBadRequest)
|
|
return
|
|
}
|
|
w.Header().Set("Content-Type", "application/json")
|
|
w.WriteHeader(http.StatusOK)
|
|
w.Write([]byte(`{"status":"ok"}`))
|
|
}
|
|
|
|
// extractHost strips the optional port from a Host header, handling the
|
|
// bracketed IPv6 literal form as well.
|
|
func extractHost(hostHeader string) string {
|
|
if hostHeader == "" {
|
|
return "example.test"
|
|
}
|
|
if host, _, err := net.SplitHostPort(hostHeader); err == nil && host != "" {
|
|
return host
|
|
}
|
|
return strings.Trim(hostHeader, "[]")
|
|
}
|
|
|
|
// serviceHostPrefixes are the per-service names published under a mail domain
|
|
// (PLAN.md section 7.3); stripping one yields the mail domain itself.
|
|
var serviceHostPrefixes = []string{"mta-sts.", "autoconfig.", "autodiscover.", "mail."}
|
|
|
|
func baseDomain(host string) string {
|
|
for _, prefix := range serviceHostPrefixes {
|
|
if strings.HasPrefix(host, prefix) {
|
|
return strings.TrimPrefix(host, prefix)
|
|
}
|
|
}
|
|
return host
|
|
}
|
|
|
|
// SetMailAPI wires the webmail message API. Without it the mail routes exist
|
|
// but report that the feature is not configured, which is a clearer failure
|
|
// than a 404 that looks like a typo in the path.
|
|
func (r *Router) SetMailAPI(useCase *appusecase.WebmailUseCase, sessionSecret string) {
|
|
if useCase == nil || sessionSecret == "" {
|
|
return
|
|
}
|
|
r.mail = &mailAPI{useCase: useCase, sessions: NewWebSessionVerifier(sessionSecret)}
|
|
}
|
|
|
|
func (r *Router) mailReady(w http.ResponseWriter) bool {
|
|
if r.mail == nil {
|
|
writeError(w, http.StatusServiceUnavailable, "MAIL_API_DISABLED",
|
|
"The mail API needs JWT_SECRET to verify webmail sessions")
|
|
return false
|
|
}
|
|
return true
|
|
}
|
|
|
|
func (r *Router) handleMailFolders(w http.ResponseWriter, req *http.Request) {
|
|
if r.mailReady(w) {
|
|
r.mail.handleFolders(w, req)
|
|
}
|
|
}
|
|
|
|
func (r *Router) handleMailMessages(w http.ResponseWriter, req *http.Request) {
|
|
if r.mailReady(w) {
|
|
r.mail.handleMessages(w, req)
|
|
}
|
|
}
|
|
|
|
func (r *Router) handleMailMessage(w http.ResponseWriter, req *http.Request) {
|
|
if r.mailReady(w) {
|
|
r.mail.handleMessage(w, req)
|
|
}
|
|
}
|
|
|
|
func (r *Router) handleMailSeen(w http.ResponseWriter, req *http.Request) {
|
|
if req.Method != http.MethodPost {
|
|
writeError(w, http.StatusMethodNotAllowed, "METHOD_NOT_ALLOWED", "POST required")
|
|
return
|
|
}
|
|
if r.mailReady(w) {
|
|
r.mail.handleSeen(w, req)
|
|
}
|
|
}
|
|
|
|
func (r *Router) handleMailSend(w http.ResponseWriter, req *http.Request) {
|
|
if req.Method != http.MethodPost {
|
|
writeError(w, http.StatusMethodNotAllowed, "METHOD_NOT_ALLOWED", "POST required")
|
|
return
|
|
}
|
|
if r.mailReady(w) {
|
|
r.mail.handleSend(w, req)
|
|
}
|
|
}
|