How to Audit What Data Your Old Host Still Holds After You Leave

How to Audit What Data Your Old Host Still Holds After You Leave

If your old host still has my data after migration, the honest answer is that it almost certainly does, at least for a while, and probably in more places than you expect. Closing or abandoning an account rarely deletes anything on the server side; it usually just removes your ability to log in and look. The realistic goal is to find out what is still there, export anything you might need, and then get written confirmation of deletion rather than assuming the account closure did it.

The reason data lingers is structural. A control panel account is a thin layer over a filesystem, a database server, a mail store and a set of rotating backups, and those four things are usually cleaned up by different processes on different schedules. Your home directory might be removed promptly, the MySQL or MariaDB databases it referenced might sit untouched for much longer, and the backup generations that contain all of it are typically retained on a fixed rotation that has nothing to do with your cancellation date.

Start by inventorying what the old host still has my data in

Before you can delete anything you need a list, and the list has to come from your own records because you no longer have access to their panel. Work from your last known state: the file tree under your home directory, every database name you created, every mailbox and forwarder, every cron job, every DNS zone they hosted, and any TLS private keys or API credentials that lived on the box. If you kept a deployment script, a wp-config.php, a .env file or an old ~/.my.cnf, those are your best inventory, because they name the databases and paths that existed.

Pull the database list out of whatever config you still have locally. A quick pass over a backup of your old site will usually reveal the schema names:

grep -rhoE "DB_NAME|DB_DATABASE|database[\"']?\s*[:=]\s*[\"'][^\"']+" ./old-site-config/ \
  | sort -u
# DB_NAME='wp_prod'
# DB_DATABASE=app_live
# "database": "analytics_staging"

Mail is the part people forget. Every address on the domain is a separate store, and catch-all or default addresses can accumulate years of inbound mail you never read. Forwarders leave no mail behind, but autoresponders, vacation messages and mailing list archives often do. If the old host ran the DNS zone, the zone file itself is data: it lists subdomains, mail exchangers and verification records that describe your infrastructure.

Logs and metadata count as data

Access logs, error logs, mail logs and any application logs written under your account contain visitor IP addresses, user agents, requested URLs and email envelope data. Under privacy regimes those are personal data even though they are not your content. A host that keeps raw logs for a retention window is still holding personal data about your visitors after you leave, so ask about log retention explicitly and separately from file deletion.

Request an export before you request deletion

Order matters. Ask for an export first, verify it, and only then ask for deletion. Deletion that happens before you have a verified copy is the one mistake you cannot undo. A proper export is a filesystem archive, a logical database dump and a mail archive, delivered over a channel you control. If they offer a download link, check that it is over HTTPS and that it expires; a link that stays live indefinitely is itself a data exposure.

When the export arrives, verify it rather than trusting the filename. A database dump is only useful if it restores, and a tarball is only complete if it extracts without errors and contains what you expect:

tar -tzf export-home.tar.gz | wc -l
mysqldump --no-data --skip-comments mydb > /dev/null  # sanity check client works
gzip -t export-db.sql.gz && echo "gzip OK"
grep -c "CREATE TABLE" export-db.sql
# 4123
# gzip OK
# 27

Compare the table count against your own notes. A dump with fewer tables than you created is a partial export, and you should go back and ask for the rest before anyone deletes anything.

Get deletion in writing, with scope and a date

A support ticket saying "your account has been closed" is not a deletion confirmation. What you want is a statement that names the categories: home directory contents, all databases associated with the account, mailboxes and mail archives, DNS zones, and backups. Backups are the item hosts most often omit, because backup retention is managed separately and sometimes by a different team or a third party storage provider. Ask directly whether your data will age out of backups on a rotation or be purged, and ask for the retention period in days.

If you are dealing with a formal request under a privacy regulation, use the terminology that triggers their process: a request for erasure of personal data, and a request for confirmation of completion. Keep it in writing, keep the ticket number, and keep the timestamps. A useful confirmation names the systems touched and the date the last backup containing your data will expire. If they will not put that in writing, that is itself information about how they handle deletion.

Close the loose ends on your side

Deletion on their side does not revoke credentials that are still valid elsewhere. Rotate every secret that ever lived on that server: database passwords, API keys, SMTP credentials, and any deploy keys in version control. Remove the old host's IP addresses from your DNS SPF record and from any allowlists, and delete the old hostnames from your TLS certificate requests. If the old host held your domain registration, transfer it and confirm the registrar of record changed, because an expired registration is a far bigger risk than a stale file.

Finally, check what the public can still see. Query DNS for the old hostnames and look at the certificate transparency logs for certificates issued for your domains; a certificate issued after you left is a sign that something is still configured on their side.

dig +short old.example.com A
dig +short example.com MX
curl -sI https://old.example.com/ | head -n 5
# HTTP/2 200
# server: nginx
# content-type: text/html

A live response from a hostname you thought was retired means there is still a vhost, a document root and probably a database behind it. That is the clearest possible evidence that data remains, and it is worth putting in front of the host in your deletion request.

What to do next: build the inventory from your own configs and notes today, request the full export in writing, verify the archive and the dump before you agree to anything, then ask for scoped deletion including backups with a stated expiry date. Rotate every credential that touched the old server in parallel, and keep the whole exchange somewhere you will still be able to find it if you need to prove the request was made.

Related articles

Subscribe to our newsletter

Get the latest hosting tips, performance insights, and industry news.