Keep only what you need and delete it on time

How to do it

List each field you collect and why, and drop the rest. Keep logs without IP addresses or query strings. Default formats, such as nginx's combined, log both. Delete records automatically after a fixed time, and delete on request. Keep passwords and tokens out of logs too, as in keep private data out of logs.

Example

# Access log without IP addresses
log_format noip '$time_iso8601 "$request_method $uri" $status $body_bytes_sent';
access_log /var/log/nginx/access.log noip;

# crontab: every night at 03:00, delete enquiries older than 12 months
0 3 * * * sqlite3 /var/lib/app/data.db "DELETE FROM enquiries WHERE created_at < date('now', '-12 months')"

How to check

A scheduled job deletes expired data, and a test checks it.