🎯 Goal

Stop chaos from spreading.

Two practical goals:

1) Audit + restore order in a messy external SSD media archive (Drone/GoPro) without manual hunting.
2) Regain control of Time Machine behavior by approaching scheduling in a clean, reversible way.

Not “hacking.” This was operational discipline.

âś… What I Did

1) Audited missing / empty DJI subtitle files (SRT) safely

Built a find ... -print0 pipeline to scan DJI clips and check whether matching .SRT files existed and were non-empty.

Results:

  • found at least one clip with missing SRT
  • found .SRT files that existed but were 0 bytes
  • moved empty .SRT files into a holding folder (_HOLD_EMPTY_SRT/...) instead of deleting (reversible)

Created a CSV summary (date,clip,status,mp4_path,srt_path,notes) to track archive health without terminal spam.

2) Hit a shell trap and corrected the workflow

While pasting “script-like chunks” into zsh, I triggered errors like:

  • zsh: command not found: #
  • zsh: read-only variable: status

Meaning:

  • pasted comments got interpreted as commands in that context
  • used a variable name (status) that can clash depending on shell/environment

Fix mindset:

  • run multi-step logic as a controlled block (script/heredoc), not pasted fragments
  • use safer variable names (tm_status, srt_status, etc.)
  • validate with dry-run output first, then move files

3) Mapped the SSD structure to stop guessing

Since tree wasn’t installed, I generated:

  • a directory map (“tree replacement”)
  • a media index CSV (counts per extension + sample paths)

Outcome: I could see the real structure and avoid moving files based on assumptions.

4) Built a restore plan instead of blindly moving files

Created a GoPro restore plan CSV using anchors:

  • .LRV and .THM in the “UnProcessed” folder used as destination truth markers
  • actions labeled as:
    • MOVE, OK, CONFLICT, AMBIGUOUS, UNPLACED

The point was: review the decision file before touching anything.

5) Took control of Time Machine frequency (macOS)

macOS doesn’t provide a clean “backup every X hours” toggle.

Clean approach:

  • disable automatic backups in Settings
  • use launchd scheduling to run tmutil startbackup --auto at a defined interval

Key principle: reversible automation (create → enable → verify logs → disable/remove).

đź”— Key Cybersecurity Connections

  • Backups are security
    • ransomware + deletion + corruption become survivable if backups exist and are testable.
  • Treat file operations like incident response
    • observe → inventory → plan → execute
    • log outputs and preserve reversibility with holding folders
  • Verify, don’t assume
    • wrong scan scope creates wrong conclusions — same failure mode as security investigations.

⚠️ Challenges

  • no tree → needed a fast replacement to map structure
  • shell paste mistakes in zsh created misleading errors
  • archive structure drift: multiple “truths” (dated folders vs organized folders)
  • risk of losing hours without automation and a plan

🧠 What I Learned

  • A CSV plan is a weapon against chaos:
    • prevents destructive decisions
    • makes actions reviewable and measurable
  • “Safe moves” beat “cleanup”:
    • _HOLD_* folders are part of a professional workflow
    • deletion is last, not first
  • macOS control still exists even when the GUI hides it:
    • launchd + tmutil gives auditable, removable scheduling.

⏭️ Next Steps

  • execute only safe MOVE actions from the restore plan (no overwrites)
  • create a second report focused on UNPLACED items:
    • group by filename patterns and search wider scope
  • build a reusable “archive health check” command:
    • counts, orphan files, missing anchors
  • for Time Machine:
    • confirm schedule triggers
    • confirm backups complete successfully
    • confirm logs show clean runs

đź’­ Reflection

Not glamorous — but it’s the difference between “using computers” and “running systems”:

  • controlled changes
  • traceable actions
  • reversible operations
  • minimal drama

I’m done trusting “looks right.” If it isn’t provable, it isn’t real.

âś… Lessons Learned

What worked

  • inventory-first approach (map → index)
  • CSV plan before moving anything
  • holding folders instead of deletion

What broke

  • assumptions about file locations
  • pasting mixed fragments into zsh

Why it broke

  • wrong scan scope = wrong conclusions
  • the shell executes what you paste literally unless you contain it properly

Fix / takeaway

  • map first, then automate
  • plan → review → execute, never execute → regret

  • use clean script blocks and safe variable names