What are inodes and directories really? And what problem does journaling solve?

🎤 Say this first

An inode is the file: metadata (size, permissions, timestamps) plus pointers to its data blocks — everything except the name. A directory is just a file mapping names → inode numbers; a hard link is a second name for the same inode (files die when link count AND open-FD count hit zero — the Unix unlink-while-open trick). Journaling fixes crash consistency: multi-block updates (allocate block + update inode + update directory) can be half-done at power loss; the FS first writes an intent record to a journal, then applies it — recovery replays or discards whole entries, never leaving half-states. Same idea as a database WAL, one layer down.