After a job fails halfway: a cleanup routine before you try again

A long job stops partway through. A file is half written, an email batch went to some people but not others, a folder holds forty new files and nobody knows which ones are complete. The tempting move is to run it again from the top. It is also the move most likely to make the mess worse.

Cleaning up first costs ten minutes and saves hours. The routine below works whether the job was run by a person, a script or an AI assistant, and it assumes nothing about why the job failed. That question comes later. The examples below are illustrative.

Step 1: Stop everything that might still be running

Check that the job has actually stopped, including scheduled repeats and anything it started somewhere else. A second run on top of a first that has not finished is the most common way to double a problem. Pause the schedule, not just the run. If the job writes to a shared place, tell the people who share it that a cleanup is under way.

Step 2: Write down what it touched

List every place the job could have changed something: files and folders, sheets, messages, accounts, anything shared. Look at modified dates and sent folders. Do not fix anything yet. Just write the list. It is the map you will clean from.

Step 3: Sort the results into three piles

Done and good, done but doubtful, not done. The first pile stays. The second pile is the one that needs reading, because a half-finished file looks a lot like a finished one. Do not trust names and sizes alone. Open a few of the files marked done, and make sure to include the first and the last of the run. The third pile is the work for the rerun.

Step 4: Undo or quarantine whatever cannot stay

Move the doubtful files into a folder named with the date and the word “quarantine” instead of deleting them. If messages went to the wrong people, decide now who needs to know, and send the short correction yourself. Delete only what you have a copy of.

Step 5: Find the first thing that went wrong

Look at the first sign of trouble, not the last: an error message near the start, a missing input, a changed setting. The end of a log tells you where the job gave up. The beginning tells you why.

Step 6: Restart small

Do not rerun the whole thing. Run the first small piece that failed, check it and widen in steps. Write down what a good result looks like for that piece before you run it, so that you can tell at a glance. A rerun that starts at the failure point, with the fix in place, gives you a clear answer in minutes, where a full rerun gives you another long wait.

A hypothetical cleanup, and the four-line note

A job that renames and moves three hundred files stops while handling file 140. The cleanup finds the schedule still active and pauses it. The touched list: the source folder, the destination folder and one index sheet. Sorting shows 139 files moved and renamed correctly, one half-moved file with the new name and no content, and 160 not started. The half-moved file goes into quarantine, the index sheet is restored from its previous copy, and the rerun starts with ten files. Then write four lines for whoever meets this job next, quite possibly you: what failed, what it touched, what you cleaned and what you changed.

Where Octuo fits

We’re the Octuo team. Octuo is a personal assistant and one starting point for work across AI, tools and specialist services. It operates external solutions on your behalf, and it reports back when a job is done. A cleanup routine is worth writing down before the first big job, not after it. Octuo is available for macOS.