feat(rgw): make the member-side RGW apply safe to repeat

microceph/api/certificates.go

@@ -24,16 +24,14 @@ func cmdCertificatesRGWPut(s mcTypes.State, r *http.Request) mcTypes.Response {
 		return mcTypes.InternalError(err)
 	}
 
-	err = ceph.UpdateRGWCertificates(interfaces.CephState{State: s}, req.SSLCertificate, req.SSLPrivateKey)
+	// UpdateRGWCertificates handles the --restart flag itself: the restart is
+	// part of the same serialized lifecycle, so a failure cannot leave a
+	// published-but-unservable certificate behind. Input failures (bad base64,
+	// mismatched pair) carry ceph.ErrRgwFrontendInvalid and map to 400;
+	// operational failures map to 500 (see serviceErrorResponse).
+	err = ceph.UpdateRGWCertificates(r.Context(), interfaces.CephState{State: s}, req.SSLCertificate, req.SSLPrivateKey, req.Restart)
 	if err != nil {
-		return mcTypes.SmartError(err)
-	}
-
-	if req.Restart {
-		err = ceph.RestartRGW()
-		if err != nil {
-			return mcTypes.SmartError(err)
-		}
+		return serviceErrorResponse(err)
 	}
 
 	return mcTypes.EmptySyncResponse

microceph/api/rgw_test.go

@@ -1,14 +1,18 @@
 package api
 
 import (
+	"bytes"
 	"context"
+	"encoding/json"
 	"errors"
 	"fmt"
 	"net/http"
 	"net/http/httptest"
 	"strings"
 	"testing"
 
+	lxdAPI "github.com/canonical/lxd/shared/api"
+	mcTypes "github.com/canonical/microcluster/v3/microcluster/types"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 
@@ -17,6 +21,45 @@ import (
 	"github.com/canonical/microceph/microceph/interfaces"
 )
 
+func TestRGWInvalidFrontendSurvivesHTTPTransport(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		response := cmdEnableServicePut(nil, r)
+		err := response.Render(w, r)
+		if err != nil {
+			t.Errorf("rendering RGW response: %v", err)
+		}
+	}))
+	t.Cleanup(server.Close)
+
+	cases := []struct {
+		name    string
+		payload string
+	}{
+		{"malformed payload", `{`},
+		{"null payload", `null`},
+		{"negative port", `{"Port":-1,"SSL":false}`},
+		{"TLS port overflow", `{"SSL":true,"SSLPort":65536}`},
+		{"same listener port", `{"Port":443,"SSL":true,"SSLPort":443}`},
+		{"missing private key", `{"SSL":true,"SSLCertificate":"Y2VydA=="}`},
+		{"invalid certificate", `{"SSL":true,"SSLCertificate":"!","SSLPrivateKey":"c2VjcmV0"}`},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			body, err := json.Marshal(types.EnableService{Name: "rgw", Wait: true, Payload: tc.payload})
+			require.NoError(t, err)
+			response, err := http.Post(server.URL+"/1.0/services/rgw", "application/json", bytes.NewReader(body))
+			require.NoError(t, err)
+			assert.Equal(t, http.StatusBadRequest, response.StatusCode)
+
+			// Microcluster reconstructs the error from HTTP status, not error_code.
+			_, err = mcTypes.ParseResponse(response)
+			require.Error(t, err)
+			assert.True(t, lxdAPI.StatusErrorCheck(err, http.StatusBadRequest))
+			assert.NotContains(t, err.Error(), "c2VjcmV0")
+		})
+	}
+}
+
 // TestEnableServicePutResponses verifies the handler's HTTP status for placement
 // outcomes, including failures that must not be returned as successful responses.
 func TestEnableServicePutResponses(t *testing.T) {

microceph/ceph/config.go

@@ -439,11 +439,12 @@ func UpdateConfig(ctx context.Context, s interfaces.StateInterface) error {
 	monitorAddresses = formatIPv6(monitorAddresses)
 
 	// Keep radosgw.conf's `mon host` line in sync with the current monitors.
-	// radosgw.conf is written once at RGW enable time (EnableRGW) and is never
-	// re-rendered from scratch, because the RGW frontend port/SSL settings are
-	// not persisted. We therefore rewrite only the mon host line in place,
-	// preserving the other settings. A missing file (RGW disabled) is a no-op;
-	// a failure here must not block the ceph.conf refresh below.
+	// radosgw.conf is owned by the RGW enable/update paths (applyRGWFrontend),
+	// which alone know the TLS material the frontend references, so it is not
+	// re-rendered from scratch here. We therefore rewrite only the mon host
+	// line in place, preserving the other settings. A missing file (RGW
+	// disabled) is a no-op; a failure here must not block the ceph.conf
+	// refresh below.
 	err = updateRadosGWMonHost(confPath, monitorAddresses)
 	if err != nil {
 		logger.Warnf("failed to refresh radosgw.conf mon host: %v", err)

microceph/ceph/configwriter.go

@@ -1,6 +1,7 @@
 package ceph
 
 import (
+	"bytes"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -32,6 +33,24 @@ func (c *Config) validateConfigFile() error {
 	return nil
 }
 
+// RenderConfig renders the configuration template with data into a byte slice
+// without touching disk. A caller compares this against the current on-disk
+// file to decide whether a rewrite (and service restart) is actually needed,
+// the "render from dqlite, compare in place" idiom (AGENTS.md conf-file
+// model). The output is byte-identical to WriteConfig for the same data.
+func (c *Config) RenderConfig(data map[string]any) ([]byte, error) {
+	err := c.validateConfigFile()
+	if err != nil {
+		return nil, err
+	}
+	var buf bytes.Buffer
+	err = c.configTemplate.Execute(&buf, data)
+	if err != nil {
+		return nil, fmt.Errorf("Couldn't render %s: %w", c.configFile, err)
+	}
+	return buf.Bytes(), nil
+}
+
 // WriteConfig writes the configuration file given a data bag and a filemode.
 // It writes atomically: the template is rendered to a uniquely-named temporary
 // file (created via os.CreateTemp in the same directory as the destination) and