Make an S3 listing the server cuts short an error (closes #297)

The SDK paginator takes a listing page marked truncated but carrying no
continuation token for the last page and stops without an error. A
server answering V2 listings V1-style sends such pages, so ListObjects
and ListObjectsStream returned the first page as the whole listing, and
prune treated the blobs of the snapshots on later pages as unreferenced.
Both now fetch each page through nextListPage, which returns
ErrNoContinuationToken for such a page; prune, snapshot list and remote
nuke already handle a listing error.

The test serves a fixed reply from a stub server, since gofakes3 always
sends the token.

Model: opus-5-5

README.md

@@ -349,9 +349,11 @@ the backup destination store. Blobs are NOT touched — deleting blobs
 requires reading every remaining remote manifest (the destination store
 may hold snapshots this host doesn't know about), which is what
 `vaultik prune` does. On success the command prints the exact `vaultik
-prune` invocation to run as a follow-up. Local row cleanup (files,
+prune` invocation to run as a follow-up. An ID that names no snapshot in
-chunks, blobs the snapshot was the last referrer for) runs
+the local index or on the destination store is an error; with
-automatically. If the destination store is unreachable, the local-DB
+`--local-only`, the snapshot must be in the local index. Local row
+cleanup (files, chunks, blobs the snapshot was the last referrer for)
+runs automatically. If the destination store is unreachable, the local-DB
 removal still completes and a warning is emitted; run `vaultik snapshot
 remove <snapshot-id>` again once the store is reachable to remove the
 snapshot's metadata from it (`vaultik prune` does not). To wipe everything
@@ -412,7 +414,8 @@ destination store.
 entirely. Remote storage is unaffected; the next backup will do a full
 scan and re-deduplicate against existing remote blobs, and the local
 index will re-bind to the currently configured storage destination.
-Use this after changing `storage_url` to a different destination.
+Use this after changing `storage_url`, or `s3.endpoint`, `s3.bucket` or
+`s3.prefix`, to a different destination.
 * `--force`: Skip confirmation prompt
 
 ---

TODO.md

@@ -22,6 +22,52 @@ the tag exists and is exercised; what is left is merging `next` to
 
 # Completed Steps
 
+- 2026-10-13: Made an S3 listing that the server cuts short an error
+  ([issue #297](https://git.eeqj.de/sneak/vaultik/issues/297)). A page
+  marked truncated with no continuation token, as a server answering V2
+  listings V1-style sends, ended the listing without an error, so
+  `prune` saw only the snapshots on the first page and could delete
+  blobs the rest still use. Such a page is now a listing error, and
+  `prune`, `snapshot list` and `remote nuke` handle it as a destination
+  that cannot be listed.
+
+- 2026-10-13: Made `snapshot remove` fail for a snapshot that does not
+  exist ([issue #293](https://git.eeqj.de/sneak/vaultik/issues/293)).
+  The command never checked the ID: it reported the snapshot removed
+  from the local database and from the destination store and exited 0,
+  under `--json`, `--dry-run` and `--local-only` too. An ID that names
+  no snapshot in the local index or on the destination store is now an
+  error; with `--local-only`, the snapshot must be in the local index.
+  The command now reports a removal from either side only when one
+  happened, and `remote_removed` is true only when metadata was deleted.
+
+- 2026-10-12: Bound the local index to the destination an `s3.*` config
+  names ([issue #298](https://git.eeqj.de/sneak/vaultik/issues/298)).
+  Only `storage_url` was checked, so after `s3.bucket` or `s3.prefix`
+  changed, the next backup treated every chunk as already stored,
+  uploaded nothing, and completed a snapshot that could not be restored.
+  Without `storage_url`, the local index is now bound to `s3.endpoint`,
+  `s3.bucket` and `s3.prefix`, and a change is refused as a changed
+  `storage_url` is.
+
+- 2026-10-12: Rejected a retention duration longer than about 292 years
+  ([issue #289](https://git.eeqj.de/sneak/vaultik/issues/289)). The
+  number given to `--older-than` or `--keep-newer-than` was multiplied
+  by the length of its unit with no limit, so from `293y`, `15251w`,
+  `3559mo` or `106752d`, or a sum of parts such as `200y200y`, the
+  result wrapped to a negative duration. The cutoff then lay in the
+  future, every snapshot of the name was deleted, the one just made
+  included, and the command exited 0. Such a duration is now rejected
+  as invalid and nothing is deleted.
+
+- 2026-10-08: Made a backup cancelled under `--skip-errors` stop at once
+  ([issue #286](https://git.eeqj.de/sneak/vaultik/issues/286)). After
+  Ctrl-C or SIGTERM, phase 2 treated the cancellation error from each
+  remaining file like an unreadable file: it opened the file, printed an
+  error line for it, counted it as failed and went on to the next one.
+  The run now stops at the first file after the cancellation, and no
+  file is counted as failed because of it.
+
 - 2026-10-08: Made `remote nuke` delete the `.partial` files that
   uploads cut off part-way leave on the destination store
   ([issue #281](https://git.eeqj.de/sneak/vaultik/issues/281)). The

internal/cli/snapshot.go

@@ -255,8 +255,12 @@ manifest (the destination store may hold snapshots this host doesn't
 know about), which is what 'vaultik prune' does. On success the command
 prints the exact 'vaultik prune' invocation to run as a follow-up.
 
+An ID that names no snapshot in the local index or on the destination
+store is an error.
+
 Use --local-only to skip the remote half (e.g. when you want to forget a
-snapshot locally without touching the destination store).
+snapshot locally without touching the destination store); the snapshot
+must then be in the local index.
 
 If the remote is unreachable, the local-database removal still completes
 and a warning is emitted; run 'vaultik snapshot remove <snapshot-id>' again

internal/s3/client.go

@@ -18,6 +18,13 @@ import (
 	"github.com/aws/smithy-go/logging"
 )
 
+// ErrNoContinuationToken is returned by ListObjects and ListObjectsStream
+// when the server marks a page of a listing truncated but sends no
+// continuation token to fetch the rest with, as a server answering V2
+// listings V1-style does.
+var ErrNoContinuationToken = errors.New(
+	"server marked the listing truncated but sent no continuation token")
+
 // Client wraps the AWS S3 client for vaultik operations.
 // It provides a simplified interface for S3 operations with automatic
 // prefix handling and connection management. All operations are performed
@@ -214,7 +221,7 @@ func (c *Client) ListObjects(ctx context.Context, prefix string) ([]string, erro
 	})
 
 	for paginator.HasMorePages() {
-		page, err := paginator.NextPage(ctx)
+		page, err := nextListPage(ctx, paginator)
 		if err != nil {
 			return nil, err
 		}
@@ -235,6 +242,26 @@ func (c *Client) ListObjects(ctx context.Context, prefix string) ([]string, erro
 	return keys, nil
 }
 
+// nextListPage fetches the next page of a listing. The SDK paginator takes
+// a page marked truncated without a continuation token for the last page
+// and stops without an error, so such a page is ErrNoContinuationToken
+// here instead.
+func nextListPage(
+	ctx context.Context, paginator *s3.ListObjectsV2Paginator,
+) (*s3.ListObjectsV2Output, error) {
+	page, err := paginator.NextPage(ctx)
+	if err != nil {
+		return nil, err
+	}
+
+	truncated := aws.ToBool(page.IsTruncated)
+	if truncated && aws.ToString(page.NextContinuationToken) == "" {
+		return nil, ErrNoContinuationToken
+	}
+
+	return page, nil
+}
+
 // HeadObject checks if an object exists in S3.
 // Returns true if the object exists, false otherwise.
 // The key is automatically prefixed with the configured prefix.
@@ -300,7 +327,7 @@ func (c *Client) ListObjectsStream(
 		})
 
 		for paginator.HasMorePages() {
-			page, err := paginator.NextPage(ctx)
+			page, err := nextListPage(ctx, paginator)
 			if err != nil {
 				ch <- ObjectInfo{Err: err}
 

internal/snapshot/scanner.go

@@ -1312,6 +1312,13 @@ func (s *Scanner) processPhase(
 
 	// Process each file
 	for _, fileToProcess := range filesToProcess {
+		// Check context cancellation
+		select {
+		case <-ctx.Done():
+			return ctx.Err()
+		default:
+		}
+
 		// Update progress
 		if s.progress != nil {
 			s.progress.GetStats().CurrentFile.Store(fileToProcess.Path)
@@ -1369,6 +1376,11 @@ func (s *Scanner) processFileWithErrorHandling(
 
 	err := s.processFileStreaming(ctx, fileToProcess, result)
 	if err != nil {
+		// A cancelled run stops here even under --skip-errors, rather than
+		// counting this file as failed and going on to the next one.
+		if ctx.Err() != nil {
+			return false, fmt.Errorf("processing file %s: %w", fileToProcess.Path, err)
+		}
 		// A packer/database/encryption/upload failure means the chunk's data
 		// may not have been stored. Skipping the file would let the snapshot
 		// record a file whose chunk is in no blob and cannot be restored, so

internal/snapshot/skip_errors_test.go

@@ -82,6 +82,32 @@ func (f *readFailFs) Open(name string) (afero.File, error) {
 	return file, nil
 }
 
+// cancelOnOpenFs cancels the run when the scanner opens the target file to
+// back it up, as Ctrl-C partway through processing would, and records each
+// file opened after that.
+type cancelOnOpenFs struct {
+	afero.Fs
+
+	target            string
+	cancel            context.CancelFunc
+	cancelled         bool
+	openedAfterCancel []string
+}
+
+//nolint:ireturn // afero.Fs.Open is defined to return the interface.
+func (f *cancelOnOpenFs) Open(name string) (afero.File, error) {
+	if f.cancelled {
+		f.openedAfterCancel = append(f.openedAfterCancel, name)
+	}
+
+	if name == f.target {
+		f.cancel()
+		f.cancelled = true
+	}
+
+	return f.Fs.Open(name)
+}
+
 // linkRemovedAfterLstatFs is the real filesystem, except that the symlink at
 // target is removed right after the walk lstats it, as happens when a link is
 // deleted during a backup. The scanner's readlink of it then fails.
@@ -128,15 +154,16 @@ func writeSkipErrorTestFile(t *testing.T, fs afero.Fs, path, content string) {
 	}
 }
 
-// runSkipErrorScan scans source on fs with the given skip-errors setting,
+// runSkipErrorScan scans source on fs under ctx with the given skip-errors
-// printing user-facing messages to uiw (nil discards them), and returns the
+// setting, printing user-facing messages to uiw (nil discards them), and
-// repositories (for inspection) and the scan error.
+// returns the repositories (for inspection) and the scan error.
 func runSkipErrorScan(
-	t *testing.T, fs afero.Fs, source string, skipErrors bool, uiw *ui.Writer,
+	ctx context.Context, t *testing.T, fs afero.Fs, source string,
+	skipErrors bool, uiw *ui.Writer,
 ) (*database.Repositories, error) {
 	t.Helper()
 
-	db, err := database.NewTestDB()
+	db, err := database.New(ctx, ":memory:")
 	if err != nil {
 		t.Fatalf("create test db: %v", err)
 	}
@@ -161,7 +188,6 @@ func runSkipErrorScan(
 		SkipErrors:       skipErrors,
 	})
 
-	ctx := context.Background()
 	snapshotID := "test-snapshot-skip-errors"
 	createTestSnapshotRecord(ctx, t, repos, snapshotID)
 
@@ -185,7 +211,7 @@ func TestScannerPackingFailureAbortsUnderSkipErrors(t *testing.T) {
 	writeSkipErrorTestFile(t, fs, "/source/file1.txt", "first file content")
 	writeSkipErrorTestFile(t, fs, "/source/file2.txt", "second file content")
 
-	repos, err := runSkipErrorScan(t, fs, "/source", true, nil)
+	repos, err := runSkipErrorScan(context.Background(), t, fs, "/source", true, nil)
 	if err == nil {
 		t.Fatal("expected scan to abort on the packer error, got nil")
 	}
@@ -212,7 +238,7 @@ func TestScannerReadErrorAbortsWithoutSkipErrors(t *testing.T) {
 	fs := &readFailFs{Fs: afero.NewMemMapFs(), target: target}
 	writeSkipErrorTestFile(t, fs, target, "content that cannot be read")
 
-	_, err := runSkipErrorScan(t, fs, "/source", false, nil)
+	_, err := runSkipErrorScan(context.Background(), t, fs, "/source", false, nil)
 	if err == nil {
 		t.Fatal("expected scan to fail on the read error, got nil")
 	}
@@ -228,7 +254,7 @@ func TestScannerReadErrorSkippedWithSkipErrors(t *testing.T) {
 	fs := &readFailFs{Fs: afero.NewMemMapFs(), target: target}
 	writeSkipErrorTestFile(t, fs, target, "content that cannot be read")
 
-	repos, err := runSkipErrorScan(t, fs, "/source", true, nil)
+	repos, err := runSkipErrorScan(context.Background(), t, fs, "/source", true, nil)
 	if err != nil {
 		t.Fatalf("expected scan to complete with --skip-errors, got %v", err)
 	}
@@ -267,7 +293,7 @@ func TestScannerUnreadableSymlinkAbortsWithoutSkipErrors(t *testing.T) {
 	sourceDir, linkPath := writeSymlinkSource(t)
 	fs := &linkRemovedAfterLstatFs{t: t, target: linkPath}
 
-	_, err := runSkipErrorScan(t, fs, sourceDir, false, nil)
+	_, err := runSkipErrorScan(context.Background(), t, fs, sourceDir, false, nil)
 	if !errors.Is(err, os.ErrNotExist) {
 		t.Fatalf("expected scan to fail on the removed symlink, got %v", err)
 	}
@@ -283,7 +309,7 @@ func TestScannerUnreadableSymlinkSkippedWithSkipErrors(t *testing.T) {
 	fs := &linkRemovedAfterLstatFs{t: t, target: linkPath}
 	uiw := ui.NewWithColor(io.Discard, false)
 
-	repos, err := runSkipErrorScan(t, fs, sourceDir, true, uiw)
+	repos, err := runSkipErrorScan(context.Background(), t, fs, sourceDir, true, uiw)
 	if err != nil {
 		t.Fatalf("expected scan to complete with --skip-errors, got %v", err)
 	}
@@ -302,3 +328,56 @@ func TestScannerUnreadableSymlinkSkippedWithSkipErrors(t *testing.T) {
 		t.Fatalf("expected %s not to be recorded", linkPath)
 	}
 }
+
+// TestScannerCancelStopsSkipErrorsRun checks that a --skip-errors backup
+// cancelled partway through processing returns the cancellation error,
+// opens no further file, and reports no file as failed.
+func TestScannerCancelStopsSkipErrorsRun(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		name          string
+		targetContent string
+	}{
+		// The cancellation lands while the target is being read.
+		{name: "while reading a file", targetContent: "first file content"},
+		// An empty target has no chunks, so the cancellation goes unnoticed
+		// until the run moves on to the next file.
+		{name: "between files", targetContent: ""},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			const target = "/source/a.txt"
+
+			ctx, cancel := context.WithCancel(context.Background())
+			defer cancel()
+
+			fs := &cancelOnOpenFs{
+				Fs: afero.NewMemMapFs(), target: target, cancel: cancel,
+			}
+			writeSkipErrorTestFile(t, fs, target, tt.targetContent)
+			writeSkipErrorTestFile(t, fs, "/source/b.txt", "second file content")
+			writeSkipErrorTestFile(t, fs, "/source/c.txt", "third file content")
+
+			uiw := ui.NewWithColor(io.Discard, false)
+
+			_, err := runSkipErrorScan(ctx, t, fs, "/source", true, uiw)
+			if !errors.Is(err, context.Canceled) {
+				t.Fatalf("expected the cancellation error, got %v", err)
+			}
+
+			if len(fs.openedAfterCancel) != 0 {
+				t.Fatalf("expected no file opened after the cancellation, got %v",
+					fs.openedAfterCancel)
+			}
+
+			if uiw.ErrorCount() != 0 {
+				t.Fatalf("expected no file reported as failed, got %d error lines",
+					uiw.ErrorCount())
+			}
+		})
+	}
+}

internal/storage/s3_test.go

@@ -4,6 +4,7 @@ import (
 	"bytes"
 	"context"
 	"errors"
+	"io"
 	"net/http"
 	"net/http/httptest"
 	"slices"
@@ -22,13 +23,16 @@ import (
 // s3TestBucket is the bucket created for each in-process S3 server.
 const s3TestBucket = "test-bucket"
 
-// Credentials for the tests that build a storer from a config.Config. The
+// Credentials for the tests' S3 clients. The test servers accept any.
-// in-process S3 server accepts any.
 const (
 	s3TestAccessKeyID     = "key"
 	s3TestSecretAccessKey = "secret"
 )
 
+// s3TestRegion is the region of the S3 clients built here with
+// s3.NewClient. The test servers ignore it.
+const s3TestRegion = "us-east-1"
+
 // newS3Storer builds an s3:// backend backed by a fresh in-process
 // S3 server. It reuses the same in-memory S3 harness (gofakes3 + s3mem
 // over httptest) that internal/s3 and the not-found regression test use,
@@ -55,7 +59,7 @@ func newS3Storer(t *testing.T) storage.Storer {
 		Bucket:          s3TestBucket,
 		AccessKeyID:     "test",
 		SecretAccessKey: "test",
-		Region:          "us-east-1",
+		Region:          s3TestRegion,
 	})
 	if err != nil {
 		t.Fatalf("new client: %v", err)
@@ -93,6 +97,68 @@ func TestS3StorerMissingKeyMapsToErrNotFound(t *testing.T) {
 	}
 }
 
+// TestS3ListingCutShortIsAnError checks that List and ListStream fail on a
+// listing page marked truncated that carries no continuation token, as a
+// server answering V2 listings V1-style sends it. Taken as the last page,
+// it would hide every later key, and prune would treat the blobs of the
+// snapshots listed there as unreferenced.
+func TestS3ListingCutShortIsAnError(t *testing.T) {
+	t.Parallel()
+
+	const (
+		prefix        = "metadata/"
+		truncatedPage = `<?xml version="1.0" encoding="UTF-8"?>
+<ListBucketResult xmlns="http://s3.amazonaws.com/doc/2006-03-01/">
+  <Name>test-bucket</Name>
+  <Prefix>metadata/</Prefix>
+  <KeyCount>1</KeyCount>
+  <MaxKeys>1</MaxKeys>
+  <IsTruncated>true</IsTruncated>
+  <Contents>
+    <Key>metadata/snap/manifest.json.zst</Key>
+    <Size>8</Size>
+  </Contents>
+</ListBucketResult>`
+	)
+
+	srv := httptest.NewServer(http.HandlerFunc(
+		func(w http.ResponseWriter, _ *http.Request) {
+			_, _ = io.WriteString(w, truncatedPage)
+		}))
+	t.Cleanup(srv.Close)
+
+	client, err := s3.NewClient(context.Background(), s3.Config{
+		Endpoint:        srv.URL,
+		Bucket:          s3TestBucket,
+		AccessKeyID:     s3TestAccessKeyID,
+		SecretAccessKey: s3TestSecretAccessKey,
+		Region:          s3TestRegion,
+	})
+	if err != nil {
+		t.Fatalf("new client: %v", err)
+	}
+
+	storer := storage.NewS3Storer(client)
+	ctx := context.Background()
+
+	keys, err := storer.List(ctx, prefix)
+	if !errors.Is(err, s3.ErrNoContinuationToken) {
+		t.Errorf("List = %q, %v; want ErrNoContinuationToken", keys, err)
+	}
+
+	var streamErr error
+
+	for obj := range storer.ListStream(ctx, prefix) {
+		if obj.Err != nil {
+			streamErr = obj.Err
+		}
+	}
+
+	if !errors.Is(streamErr, s3.ErrNoContinuationToken) {
+		t.Errorf("ListStream error = %v, want ErrNoContinuationToken", streamErr)
+	}
+}
+
 // TestS3URLPrefixKeyLayout pins the bucket keys an s3:// URL reads and
 // writes: the README's remote storage layout, with the prefix joined to
 // each key by one "/". s3://b/p and s3://b/p/ must be the same

internal/vaultik/helpers.go

@@ -3,6 +3,7 @@ package vaultik
 import (
 	"errors"
 	"fmt"
+	"math"
 	"regexp"
 	"strconv"
 	"strings"
@@ -47,6 +48,10 @@ const (
 	year  = 365 * day
 )
 
+// maxDuration is the largest value a time.Duration can hold, about 292
+// years.
+const maxDuration = time.Duration(math.MaxInt64)
+
 // A snapshot ID split on "_" has at least a hostname and a trailing
 // timestamp.
 const minSnapshotIDParts = 2
@@ -145,10 +150,10 @@ func parseSnapshotName(snapshotID, hostname string) string {
 // parseDuration parses a duration string with support for human-friendly units:
 // d/day/days, w/week/weeks, mo/month/months, y/year/years, plus standard Go
 // duration units. Following Go, m is minutes and mo is months. A bare number,
-// an unknown unit, and a negative value are all rejected. Outside the Go
+// an unknown unit, a negative value, and a total longer than maxDuration are
-// units the input must be whole numbers each followed directly by a unit
+// all rejected. Outside the Go units the input must be whole numbers each
-// (2w3d), so 1.0y and 30 days are errors; decimals work only in Go units
+// followed directly by a unit (2w3d), so 1.0y and 30 days are errors;
-// (1.5h).
+// decimals work only in Go units (1.5h).
 func parseDuration(s string) (time.Duration, error) {
 	if strings.HasPrefix(strings.TrimSpace(s), "-") {
 		return 0, errNegativeDuration
@@ -181,19 +186,30 @@ func parseDuration(s string) (time.Duration, error) {
 			return 0, fmt.Errorf("invalid number %q: %w", match[1], err)
 		}
 
+		var unitLength time.Duration
+
 		unit := strings.ToLower(match[2])
 		switch unit {
 		case "d", "day", "days":
-			total += time.Duration(n) * day
+			unitLength = day
 		case "w", "week", "weeks":
-			total += time.Duration(n) * week
+			unitLength = week
 		case "mo", "month", "months":
-			total += time.Duration(n) * month
+			unitLength = month
 		case "y", "year", "years":
-			total += time.Duration(n) * year
+			unitLength = year
 		default:
 			return 0, fmt.Errorf("%w %q", errUnknownTimeUnit, unit)
 		}
+
+		// Past maxDuration the sum wraps to a negative value, which would
+		// put the purge cutoff in the future.
+		if time.Duration(n) > (maxDuration-total)/unitLength {
+			return 0, fmt.Errorf("%w: %q is longer than about 292 years",
+				errInvalidDuration, s)
+		}
+
+		total += time.Duration(n) * unitLength
 	}
 
 	return total, nil

internal/vaultik/helpers_test.go

@@ -111,6 +111,17 @@ func TestParseDuration(t *testing.T) {
 		{"30 days", 0, true},
 		{"30 days ago", 0, true},
 		{"x7d", 0, true},
+		// The largest time.Duration is about 292 years. A larger value
+		// must be rejected, not wrapped to a negative one, which would
+		// put the cutoff in the future and select every snapshot.
+		{"292y", 292 * 365 * 24 * time.Hour, false},
+		{"293y", 0, true},
+		{"3559mo", 0, true},
+		{"15251w", 0, true},
+		{"106752d", 0, true},
+		{"2562048h", 0, true},
+		{"200y200y", 0, true},
+		{"292y6mo", 0, true},
 	}
 
 	for _, tt := range tests {

internal/vaultik/purge_per_name_test.go

@@ -267,6 +267,29 @@ func TestPurgeOlderThan_WithNameFilter(t *testing.T) {
 	assert.Contains(t, remaining, snapHomeT0)
 }
 
+func TestPurgeOlderThan_TooLargeDeletesNothing(t *testing.T) {
+	log.Initialize(log.Config{})
+	t.Parallel()
+
+	snapshotIDs := []string{snapHomeT0, snapHomeT1}
+
+	v := setupPurgeTest(t, snapHostname, snapshotIDs)
+
+	err := v.PurgeSnapshotsWithOptions(&vaultik.SnapshotPurgeOptions{
+		OlderThan: "1000y",
+		Force:     true,
+	})
+	require.Error(t, err)
+
+	err = v.PurgeSnapshotsWithOptions(&vaultik.SnapshotPurgeOptions{
+		OlderThan: "292y",
+		Force:     true,
+	})
+	require.NoError(t, err)
+
+	assert.ElementsMatch(t, snapshotIDs, listRemainingSnapshots(t, v))
+}
+
 func TestPurgeKeepLatest_ThreeNames(t *testing.T) {
 	log.Initialize(log.Config{})
 	t.Parallel()

internal/vaultik/remove_snapshot_test.go

@@ -3,17 +3,21 @@ package vaultik_test
 import (
 	"bytes"
 	"context"
+	"database/sql"
 	"io"
 	"strings"
 	"sync"
 	"testing"
+	"time"
 
 	"github.com/klauspost/compress/zstd"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
+	"sneak.berlin/go/vaultik/internal/database"
 	"sneak.berlin/go/vaultik/internal/log"
 	"sneak.berlin/go/vaultik/internal/snapshot"
 	"sneak.berlin/go/vaultik/internal/storage"
+	"sneak.berlin/go/vaultik/internal/types"
 	"sneak.berlin/go/vaultik/internal/vaultik"
 )
 
@@ -214,6 +218,37 @@ func addBlob(t *testing.T, store *testStorer, hash string) {
 	require.NoError(t, err)
 }
 
+// newLocalIndex returns an in-memory local index holding a completed
+// snapshot row for each of ids.
+func newLocalIndex(t *testing.T, ids ...string) *database.Repositories {
+	t.Helper()
+
+	ctx := context.Background()
+
+	db, err := database.New(ctx, ":memory:")
+	require.NoError(t, err)
+	t.Cleanup(func() { _ = db.Close() })
+
+	repos := database.NewRepositories(db)
+
+	for _, id := range ids {
+		completedAt := time.Now().UTC()
+		snap := &database.Snapshot{
+			ID:          types.SnapshotID(id),
+			Hostname:    snapHostname,
+			StartedAt:   completedAt,
+			CompletedAt: &completedAt,
+		}
+
+		err = repos.WithTx(ctx, func(ctx context.Context, tx *sql.Tx) error {
+			return repos.Snapshots.Create(ctx, tx, snap)
+		})
+		require.NoError(t, err)
+	}
+
+	return repos
+}
+
 // ============================================================================
 // Unit Tests for RemoveSnapshot
 // ============================================================================
@@ -233,6 +268,7 @@ func TestRemoveSnapshot_LocalOnly_PreservesRemote(t *testing.T) {
 	addBlob(t, store, blobA)
 
 	tv := vaultik.NewForTesting(store)
+	tv.Repositories = newLocalIndex(t, "snapshot-001")
 
 	opts := &vaultik.RemoveOptions{Force: true, LocalOnly: true}
 	result, err := tv.RemoveSnapshot("snapshot-001", opts)
@@ -268,6 +304,7 @@ func TestRemoveSnapshot_DefaultRemovesMetadataNotBlobs(t *testing.T) {
 	addBlob(t, store, blobShared)
 
 	tv := vaultik.NewForTesting(store)
+	tv.Repositories = newLocalIndex(t, "snapshot-001", "snapshot-002")
 
 	opts := &vaultik.RemoveOptions{Force: true}
 	result, err := tv.RemoveSnapshot("snapshot-001", opts)
@@ -316,6 +353,84 @@ func TestRemoveSnapshot_DryRun(t *testing.T) {
 	assert.Contains(t, tv.Stdout.String(), "[Dry run - no changes made]")
 }
 
+// TestRemoveSnapshot_UnknownIDIsError checks that an ID naming no
+// snapshot in the local index or on the destination store is an error
+// under every option, and that nothing is reported as removed.
+func TestRemoveSnapshot_UnknownIDIsError(t *testing.T) {
+	log.Initialize(log.Config{})
+	t.Parallel()
+
+	for name, opts := range map[string]vaultik.RemoveOptions{
+		"force":      {Force: true},
+		"json":       {JSON: true},
+		"dry-run":    {Force: true, DryRun: true},
+		"local-only": {Force: true, LocalOnly: true},
+	} {
+		t.Run(name, func(t *testing.T) {
+			t.Parallel()
+
+			store := newTestStorer()
+			addManifest(t, store, "snapshot-001", nil)
+
+			tv := vaultik.NewForTesting(store)
+			tv.Repositories = newLocalIndex(t, "snapshot-001")
+
+			result, err := tv.RemoveSnapshot("snapshot-999", &opts)
+
+			require.ErrorContains(t, err, "snapshot not found")
+			assert.False(t, result.RemoteRemoved)
+			assert.Empty(t, tv.Stdout.String())
+			assert.True(t, store.hasKey(remoteKeyPath("snapshot-001")))
+		})
+	}
+}
+
+// TestRemoveSnapshot_NoMetadataOnDestination checks that removing a
+// snapshot whose metadata is not on the destination store does not
+// report metadata removed from it.
+func TestRemoveSnapshot_NoMetadataOnDestination(t *testing.T) {
+	log.Initialize(log.Config{})
+	t.Parallel()
+
+	tv := vaultik.NewForTesting(newTestStorer())
+	tv.Repositories = newLocalIndex(t, "snapshot-001")
+
+	result, err := tv.RemoveSnapshot("snapshot-001",
+		&vaultik.RemoveOptions{Force: true})
+
+	require.NoError(t, err)
+	assert.False(t, result.RemoteRemoved)
+
+	out := tv.Stdout.String()
+	assert.Contains(t, out, "Removed snapshot 'snapshot-001' from local database")
+	assert.NotContains(t, out, "Removed snapshot metadata from remote storage")
+}
+
+// TestRemoveSnapshot_NotInLocalIndex checks that removing a snapshot
+// only the destination store holds removes its metadata there without
+// reporting a removal from the local database.
+func TestRemoveSnapshot_NotInLocalIndex(t *testing.T) {
+	log.Initialize(log.Config{})
+	t.Parallel()
+
+	store := newTestStorer()
+	addManifest(t, store, "snapshot-001", nil)
+
+	tv := vaultik.NewForTesting(store)
+	tv.Repositories = newLocalIndex(t)
+
+	result, err := tv.RemoveSnapshot("snapshot-001",
+		&vaultik.RemoveOptions{Force: true})
+
+	require.NoError(t, err)
+	assert.True(t, result.RemoteRemoved)
+	assert.False(t, store.hasKey(remoteKeyPath("snapshot-001")))
+
+	out := tv.Stdout.String()
+	assert.NotContains(t, out, "from local database")
+	assert.Contains(t, out, "Removed snapshot metadata from remote storage")
+}
+
 func TestRemoveAllSnapshots_RequiresForce(t *testing.T) {
 	log.Initialize(log.Config{})
 	t.Parallel()

internal/vaultik/snapshot.go

@@ -23,6 +23,7 @@ var (
 	errNoSnapshotsInConfig  = errors.New("no snapshots configured")
 	errSnapshotVerifyFailed = errors.New("verification failed")
 	errRemoveAllNeedsForce  = errors.New("--all requires --force")
+	errSnapshotNotFound     = errors.New("snapshot not found")
 	errInvalidTableName     = errors.New("invalid table name")
 	errNoAgeRecipients      = errors.New(
 		"creating a snapshot needs at least one public key in " +
@@ -643,7 +644,7 @@ func (v *Vaultik) confirmAndExecutePurge(
 				"snapshot_id", snapshotID, "error", err)
 		}
 
-		err = v.deleteRemoteSnapshotByKey(snapshot.RemoteSnapshotKey(snapshotID))
+		_, err = v.deleteRemoteSnapshotByKey(snapshot.RemoteSnapshotKey(snapshotID))
 		if err != nil {
 			return fmt.Errorf("deleting snapshot %s from remote: %w", snapshotID, err)
 		}
@@ -1153,6 +1154,11 @@ func (v *Vaultik) RemoveSnapshot(
 		return result, err
 	}
 
+	inLocalIndex, err := v.findSnapshotToRemove(snapshotID, opts)
+	if err != nil {
+		return result, err
+	}
+
 	if opts.DryRun {
 		return v.removeSnapshotDryRun(result, snapshotID, opts)
 	}
@@ -1161,11 +1167,13 @@ func (v *Vaultik) RemoveSnapshot(
 		return result, nil
 	}
 
-	log.Info("Removing snapshot from local database", "snapshot_id", snapshotID)
+	if inLocalIndex {
+		log.Info("Removing snapshot from local database", "snapshot_id", snapshotID)
 
-	err = v.deleteSnapshotFromLocalDB(snapshotID)
+		err = v.deleteSnapshotFromLocalDB(snapshotID)
-	if err != nil {
+		if err != nil {
-		return result, fmt.Errorf("removing from local database: %w", err)
+			return result, fmt.Errorf("removing from local database: %w", err)
+		}
 	}
 
 	if !opts.LocalOnly {
@@ -1184,9 +1192,11 @@ func (v *Vaultik) RemoveSnapshot(
 		return result, v.outputRemoveJSON(result)
 	}
 
-	v.stdoutf("Removed snapshot '%s' from local database\n", snapshotID)
+	if inLocalIndex {
+		v.stdoutf("Removed snapshot '%s' from local database\n", snapshotID)
+	}
 
-	if !opts.LocalOnly && result.RemoteRemoved {
+	if result.RemoteRemoved {
 		v.printlnStdout("Removed snapshot metadata from remote storage")
 		v.stdoutf("\nNote: The removed snapshot's blobs remain on the remote. "+
 			"Run '%s' to delete any blobs no longer referenced by any "+
@@ -1196,6 +1206,45 @@ func (v *Vaultik) RemoveSnapshot(
 	return result, nil
 }
 
+// findSnapshotToRemove reports whether the local index has a row for
+// snapshotID, and returns errSnapshotNotFound when there is nothing to
+// remove: no row and, unless LocalOnly is set, no metadata on the
+// destination store either. The destination store is consulted only
+// when the local index has no row, so a snapshot in the local index can
+// still be removed from it while the destination store is unreachable.
+func (v *Vaultik) findSnapshotToRemove(
+	snapshotID string, opts *RemoveOptions,
+) (bool, error) {
+	if v.Repositories != nil {
+		snap, err := v.Repositories.Snapshots.GetByID(v.ctx, snapshotID)
+		if err != nil {
+			return false, fmt.Errorf("looking up snapshot in local index: %w", err)
+		}
+
+		if snap != nil {
+			return true, nil
+		}
+	}
+
+	if opts.LocalOnly {
+		return false, fmt.Errorf("%w in the local index: %s",
+			errSnapshotNotFound, snapshotID)
+	}
+
+	objects, err := v.remoteSnapshotObjects(snapshot.RemoteSnapshotKey(snapshotID))
+	if err != nil {
+		return false, fmt.Errorf("looking up snapshot on destination store: %w", err)
+	}
+
+	if len(objects) == 0 {
+		return false, fmt.Errorf(
+			"%w in the local index or on the destination store: %s",
+			errSnapshotNotFound, snapshotID)
+	}
+
+	return false, nil
+}
+
 // removeSnapshotDryRun reports what RemoveSnapshot would do without
 // making changes.
 func (v *Vaultik) removeSnapshotDryRun(
@@ -1255,14 +1304,14 @@ func (v *Vaultik) confirmRemoveSnapshot(snapshotID string, opts *RemoveOptions)
 // didn't finish and to run `vaultik snapshot remove` for the snapshot
 // again once the destination store is reachable (`vaultik prune` never
 // removes snapshot metadata). Returns true when the remote removal
-// succeeded.
+// succeeded and deleted something.
 func (v *Vaultik) removeSnapshotRemote(snapshotID string, opts *RemoveOptions) bool {
 	log.Info("Removing snapshot metadata from remote storage",
 		"snapshot_id", snapshotID)
 
 	remoteKey := snapshot.RemoteSnapshotKey(snapshotID)
 
-	err := v.deleteRemoteSnapshotByKey(remoteKey)
+	deleted, err := v.deleteRemoteSnapshotByKey(remoteKey)
 	if err != nil {
 		log.Warn("Could not remove snapshot metadata from remote storage; "+
 			"run '"+snapshotRemoveCommandHint+"' with the snapshot's ID "+
@@ -1280,7 +1329,7 @@ func (v *Vaultik) removeSnapshotRemote(snapshotID string, opts *RemoveOptions) b
 		return false
 	}
 
-	return true
+	return deleted > 0
 }
 
 // RemoveAllSnapshots removes every snapshot known to the local
@@ -1572,7 +1621,7 @@ func (v *Vaultik) removeAllLocalSnapshots(
 		}
 
 		if !opts.LocalOnly {
-			err = v.deleteRemoteSnapshotByKey(
+			_, err = v.deleteRemoteSnapshotByKey(
 				snapshot.RemoteSnapshotKey(snapshotID))
 			if err != nil {
 				log.Warn("Failed to remove snapshot metadata from remote",
@@ -1598,7 +1647,7 @@ func (v *Vaultik) removeAllOrphanRemotes(
 	for _, key := range orphanRemoteKeys {
 		log.Info("Removing orphan remote snapshot", "remote_key", key)
 
-		err := v.deleteRemoteSnapshotByKey(key)
+		_, err := v.deleteRemoteSnapshotByKey(key)
 		if err != nil {
 			log.Warn("Failed to remove orphan from remote",
 				"remote_key", key, "error", err)
@@ -1645,34 +1694,44 @@ func (v *Vaultik) deleteSnapshotFromLocalDB(snapshotID string) error {
 }
 
 // deleteRemoteSnapshotByKey removes everything under
-// metadata/<remoteKey>/ on the destination store. The argument is a
+// metadata/<remoteKey>/ on the destination store and returns how many
-// remote key (double-SHA256 derivation), not a human snapshot ID;
+// objects it deleted. The argument is a remote key (double-SHA256
-// callers that have a human ID must hash via snapshot.RemoteSnapshotKey
+// derivation), not a human snapshot ID; callers that have a human ID
-// first.
+// must hash via snapshot.RemoteSnapshotKey first.
-func (v *Vaultik) deleteRemoteSnapshotByKey(remoteKey string) error {
+func (v *Vaultik) deleteRemoteSnapshotByKey(remoteKey string) (int, error) {
-	prefix := fmt.Sprintf("metadata/%s/", remoteKey)
+	objectsToDelete, err := v.remoteSnapshotObjects(remoteKey)
-	objectCh := v.Storage.ListStream(v.ctx, prefix)
+	if err != nil {
-
+		return 0, err
-	var objectsToDelete []string
-
-	for object := range objectCh {
-		if object.Err != nil {
-			return fmt.Errorf("listing objects: %w", object.Err)
-		}
-
-		objectsToDelete = append(objectsToDelete, object.Key)
 	}
 
-	for _, key := range objectsToDelete {
+	for i, key := range objectsToDelete {
-		err := v.Storage.Delete(v.ctx, key)
+		err = v.Storage.Delete(v.ctx, key)
 		if err != nil {
-			return fmt.Errorf("removing %s: %w", key, err)
+			return i, fmt.Errorf("removing %s: %w", key, err)
 		}
 
 		log.Debug("Deleted remote object", "key", key)
 	}
 
-	return nil
+	return len(objectsToDelete), nil
+}
+
+// remoteSnapshotObjects lists the keys of every object under
+// metadata/<remoteKey>/ on the destination store.
+func (v *Vaultik) remoteSnapshotObjects(remoteKey string) ([]string, error) {
+	prefix := fmt.Sprintf("metadata/%s/", remoteKey)
+
+	var keys []string
+
+	for object := range v.Storage.ListStream(v.ctx, prefix) {
+		if object.Err != nil {
+			return nil, fmt.Errorf("listing objects: %w", object.Err)
+		}
+
+		keys = append(keys, object.Key)
+	}
+
+	return keys, nil
 }
 
 // outputRemoveJSON outputs the removal result as JSON

internal/vaultik/storage_bind.go

@@ -9,7 +9,7 @@ import (
 )
 
 // errStorageBindingMismatch is returned when the local index database is
-// bound to a different destination than the configured storage_url.
+// bound to a different destination than the configured one.
 var errStorageBindingMismatch = errors.New(
 	"local index is bound to a different backup destination")
 
@@ -24,8 +24,9 @@ var errStorageBindingMismatch = errors.New(
 //
 // Behaviour:
 //   - On first use (empty stored value), record the configured
-//     storage_url and log the binding.
+//     destination (storage_url, or the s3.* endpoint, bucket and
-//   - When the stored value matches the configured storage_url, do
+//     prefix) and log the binding.
+//   - When the stored value matches the configured destination, do
 //     nothing and return nil.
 //   - When the two differ, refuse with an error that tells the user
 //     how to recover (revert the config, or run `vaultik database
@@ -46,12 +47,10 @@ func (v *Vaultik) EnsureStorageBinding() error {
 
 	configured := v.Config.StorageURL
 	if configured == "" {
-		// Some legacy configs still use the split s3.* keys instead of
+		// The s3.* form: bind on its endpoint, bucket and prefix,
-		// storage_url. Falling back to a synthetic URL for those would
+		// written the way storage_url would name the same destination.
-		// hide the fact that the binding is loose. Instead, treat
+		s3 := v.Config.S3
-		// unset as "nothing to bind against" — the check is
+		configured = "s3://" + s3.Bucket + "/" + s3.Prefix + "?endpoint=" + s3.Endpoint
-		// necessarily best-effort for those configs.
-		return nil
 	}
 
 	stored, err := v.Repositories.LocalMeta.Get(v.ctx, database.LocalMetaKeyStorageURL)
@@ -96,7 +95,8 @@ func buildBindingMismatchMessage(stored, configured string) string {
 		"future snapshots referencing blobs that don't exist at the new destination.\n" +
 		"\n" +
 		"To proceed, either:\n" +
-		"  - revert storage_url in your config to the bound destination, or\n" +
+		"  - revert storage_url (or the s3.* endpoint, bucket and prefix) in your\n" +
+		"    config to the bound destination, or\n" +
 		"  - run 'vaultik database delete' to discard the local index and rebuild it\n" +
 		"    from a fresh full backup against the new destination"
 }

internal/vaultik/storage_bind_test.go

@@ -76,3 +76,51 @@ func TestEnsureStorageBinding_MismatchRefuses(t *testing.T) {
 	assert.Contains(t, err.Error(), "file:///mnt/backups/new")
 	assert.Contains(t, err.Error(), "vaultik database delete")
 }
+
+// A config with no storage_url names its destination with s3.endpoint,
+// s3.bucket and s3.prefix; a change to any of them must be refused just
+// as a changed storage_url is.
+func TestEnsureStorageBinding_S3FieldsChangeRefuses(t *testing.T) {
+	log.Initialize(log.Config{})
+	t.Parallel()
+
+	bound := config.S3Config{
+		Endpoint: "https://s3.example.com",
+		Bucket:   "bucket-a",
+		Prefix:   "backups",
+	}
+
+	tests := []struct {
+		name    string
+		changed config.S3Config
+	}{
+		{"bucket", config.S3Config{
+			Endpoint: bound.Endpoint, Bucket: "bucket-b", Prefix: bound.Prefix,
+		}},
+		{"prefix", config.S3Config{
+			Endpoint: bound.Endpoint, Bucket: bound.Bucket, Prefix: "other",
+		}},
+		{"endpoint", config.S3Config{
+			Endpoint: "https://s3.other.com", Bucket: bound.Bucket, Prefix: bound.Prefix,
+		}},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			tv, _ := buildBindTestVaultik(t, "")
+
+			tv.Config.S3 = bound
+			require.NoError(t, tv.EnsureStorageBinding(), "first use binds")
+			require.NoError(t, tv.EnsureStorageBinding(), "same destination passes")
+
+			tv.Config.S3 = tt.changed
+			err := tv.EnsureStorageBinding()
+			require.Error(t, err)
+			assert.Contains(t, err.Error(),
+				"local index is bound to a different backup destination")
+			assert.Contains(t, err.Error(), "vaultik database delete")
+		})
+	}
+}