What happened?

When trying to get the changed_count (which requires generating a diff) we end up executing an O(n^2) loop on millions of records which hangs without a timeout or error, preventing the cron from finishing and also not raising any error other than the grafana (filter not uploaded) error

grafana: https://earthangel-b40313e5.influxcloud.net/d/IWZpIQgMk?orgId=1
logs: https://console.cloud.google.com/kubernetes/pod/us-west1/webservices-high-prod/amo-prod/addons-server-v1-cronjob-upload-mlbf-to-remote-setti-28860rv9s9/details?invt=Abhivw&project=moz-fx-webservices-high-prod&cloudshell=true

What did you expect to happen?

The loop should take a matter of seconds and have complexity O(n)

Additionally, if this type of error were to happen, we should be able to catch it on stage before prod and finally we should get more explicit warnings/errors that this kind of loop is running for a long time. It has been over 12 hours and no notifications.

Is there an existing issue for this?

• I have searched the existing issues

┆Issue is synchronized with this Jira Task