Any solution for uncategorizing a group of products by mistake?

[William Davis]

Any solution for uncategorizing a group of products by mistake? With thousands of products, this can turn into a nightmare, especially if products that belonged to assigned an category do searchable distinguishing text.

Fortunately, in this case they do and I also had am recent export file of said category that was only several weeks old.

[Kent Multer]

Miva makes daily backups of stores that they host. -- or at least they used to; I assume they still do.
So if you notice the mistake right away, it can be repaired with the help of someone with database skills. He can get a copy of the previous day's backup, and copy some data from the backup to your live database.
If you've already added some products to the category, or deleted some products that used to be in the category, that could cause problems. But if you notice the problem right away, and don't make any changes to the category while you;re waiting for the repair, it should be a straightforward process.

(If I'm mistaken about any of the above, I hope someone from in-house will correct me. I've done similar repairs from time to time in the past, but not recently. There may have been changes to the DB that make the situation more complicated.)

[Rick Wilson]

We make backups and snapshots but there is a fee to recover, so keep that in mind.

[ILoveHostasaurus]

The overall solution to avoid an issue like this, or more easily recover from such a mistake, is process management, change control, and risk analysis. If someone occasionally performs bulk category manipulation, then they should use a procedure that allows for recovery from a mistake, given not doing this daily greatly increases the chance of a mistake. So, that could be always creating an importable export before the operation, so the assignments could be put back. Or, it could be a script that retrieves this data via API, so you just run it on your computer before such work.

If it's being done more often, perhaps a scheduled task that does the export, or local scheduled job that pulls this data via API daily, etc.

There's no easy way for us to code an 'undo' into such changes, because it's not only possibly a very large amount of data, but then how many iterations do we keep? Could be the same person is changing ten batches of category memberships and broke things nine changes ago. It can also trigger third party modules and integrations, where it may not even be safe to just revert the data.

Miva makes daily backups of stores that they host. -- or at least they used to; I assume they still do.
So if you notice the mistake right away, it can be repaired with the help of someone with database skills. He can get a copy of the previous day's backup, and copy some data from the backup to your live database.
If you've already added some products to the category, or deleted some products that used to be in the category, that could cause problems. But if you notice the problem right away, and don't make any changes to the category while you;re waiting for the repair, it should be a straightforward process.

(If I'm mistaken about any of the above, I hope someone from in-house will correct me. I've done similar repairs from time to time in the past, but not recently. There may have been changes to the DB that make the situation more complicated.)

This should never be handled 'by someone with database skills,' which presumably means hopefully piecing what changed back together with subsets of information taken from a restored copy of the data. There could be numerous other changes that occurred within a store as a result of the action, including within third party modules and third party integrations. Using bits and pieces of data and hoping to alter things back to how they had been via direct manipulation is not safe and could create unforeseen consequences, that perhaps don't show themselves until well later when there is no longer a recovery path.

The recovery path from such a change would be:
1) Restore the entire site to what may be as little as an hour prior
2) Restore the site to the prior evening, or to the state it was in on a given day if it took multiple days to realize what occurred
3) Restore a clone of the site to a non-production URL and use supported means (API, export/import, eyeball copy and paste) to return things to how they were in a safe manner