LTFS Tape Won't Mount After Vendor Swap
A post-production studio stores 380 TB of camera RAW on 65 LTO-7 LTFS tapes. The tapes were written by an IBM TS4300 library using IBM LTFS on Linux. After a merger, the studio installs a Quantum Scalar i3 with LTO-8 drives and uses a Windows LTFS implementation. The first eight tapes mount, twelve fail with 'cannot read LTFS index', and five mount but show missing clip folders. The tapes are drive-encrypted, and the key server is available and holds the correct keys. The studio has one remaining LTO-7 drive in a storage closet, but no IBM library. A client needs 60 TB delivered in 48 hours. Reformatting is forbidden because these are the only copies. The IT team wants to know whether this is a key problem, a drive problem, or an LTFS interchange problem, and how to recover without making it worse.
1. What is the most likely explanation for the mixed mount results, and what immediate triage steps would you take?
Read the worked answer
Mixed results make a global key failure unlikely: keys are present and eight tapes mount. More likely the Windows LTFS implementation is not parsing IBM-written indexes or extensions correctly. LTFS is a standard, but implementations differ in index partition size, block size, XML index version, Unicode handling, extended attributes, and vendor-specific fields. A tape can mount but omit folders if the index is only partially read. Triage: stop retrying failing tapes; write-protect them; use the remaining LTO-7 drive with the original IBM LTFS on Linux; mount read-only; run a read-only ltfsck; export the index; compare file counts and checksums. Do not reformat or write to originals.
2. Propose a recovery workflow for the 60 TB deadline that minimizes risk to the originals.
Read the worked answer
Recover to staging disk first. Use the LTO-7 drive and original software; mount read-only; copy the 60 TB client list; verify checksums. For tapes whose index will not mount, use ltfsck to rebuild the index to a new tape or disk image, never in place. If the original drive cannot read a tape, try a known-good LTO-7 drive before escalating. Only after a verified copy should you create an LTO-8 canonical copy. Keep a manifest of recovered files. If time runs out, deliver the verified subset and document what remains. The originals stay write-protected until the client accepts the copy.
3. What procurement and documentation controls should be added to prevent a repeat?
Read the worked answer
Require LTFS conformance to a stated version, fixed block size, index partition size, no proprietary extensions, and acceptance testing with the other site's reader. Keep one drive of each generation in the production environment, not in a closet. Document format parameters and checksum manifests. Use a third-party reader for validation. Train staff that LTFS is not automatically interchangeable just because both products say LTFS. Add a clause that the vendor must demonstrate interchange with the existing tapes before the old library is removed.