What causes "getattr inode %d: %s" in juicedata/juicefs?

Re-mount with a valid existing subdirectory path (--subdir) and avoid deleting mounted ancestors

getattr inode %d: %s

Error message

getattr inode %d: %s

What it means

When resolving a directory path (e.g. for session info or subdir display), getDirPath walks parents via doGetAttr; if getattr on an inode fails, this error wraps the status code. It means an inode in the ancestor chain could not be stat'ed — typically a dangling or concurrently-deleted ancestor, or a metadata lookup error.

Source

Thrown at pkg/meta/base.go:2364

	}
}

func (m *baseMeta) GetPaths(ctx Context, inode Ino) []string {
	if inode == RootInode {
		return []string{"/"}
	}

	if inode == TrashInode {
		return []string{"/.trash"}
	}

	outside := "path not shown because it's outside of the mounted root"
	getDirPath := func(ino Ino) (string, error) {
		var names []string
		var attr Attr
		for ino != RootInode && ino != m.root {
			if st := m.en.doGetAttr(ctx, ino, &attr); st != 0 {
				return "", fmt.Errorf("getattr inode %d: %s", ino, st)
			}
			if attr.Typ != TypeDirectory {
				return "", fmt.Errorf("inode %d is not a directory", ino)
			}
			var entries []*Entry
			if st := m.en.doReaddir(ctx, attr.Parent, 0, &entries, -1); st != 0 {
				return "", fmt.Errorf("readdir inode %d: %s", ino, st)
			}
			var name string
			for _, e := range entries {
				if e.Inode == ino {
					name = string(e.Name)
					break
				}
			}
			if attr.Parent == RootInode && ino == TrashInode {
				name = TrashName
			}

View on GitHub (pinned to c9a67b23e8)

Solutions

Re-mount with a valid existing subdirectory path (--subdir) and avoid deleting mounted ancestors
Check metadata consistency (juicefs fsck) for dangling inodes/parent links
Investigate the underlying doGetAttr status from metadata engine logs

Example fix

// before: subdir deleted while mounted
juicefs mount redis://meta /jfs --subdir /gone
// after: use an existing subdir
juicefs mount redis://meta /jfs --subdir /data

Defensive patterns

Strategy: try-catch

Validate before calling

// before mounting with --subdir, verify the path exists:
// juicefs info <meta-url> /data

Try / catch

if _, err := buildDirPath(ctx, rootIno); err != nil { if strings.Contains(err.Error(), "getattr inode") { log.Warnf("ancestor missing; remount required: %v", err) } }

Prevention

Never delete directories that are mount roots/subdirs of running clients
Run `juicefs fsck` after suspected metadata corruption
Use stable, dedicated paths as subdir mount points

When it happens

Trigger: An ancestor inode of m.root was deleted while building the path; metadata engine lookup failure returning a nonzero status for an internal inode; corrupted parent links.

Common situations: Deleting the mounted subdirectory while a client is running; metadata inconsistencies after failed loads; races during concurrent rm -rf of ancestor directories.

Understand the failure class

Background: "Not found" and "does not exist" errors: why "Task not found", "No such folder", and "Can't find" fire when a lookup comes back empty — this error's family across 14 libraries.

Related errors

AI-assisted analysis of juicedata/juicefs@c9a67b23e8 (2026-09-06). Data as JSON: /api/errors/4cb34b36cadc3de2. Report an issue: GitHub.