We apologise for the system downtime.
What happened: A configuration change was applied to the database cluster with rolling node restart. Later EDR Online connection pool experienced stale connections and eventually failed on all requests.
What solved the issue: Restart of EDR Online container cluster fixed the problem.
How will we make sure this does not happen again: We will add EDR Online cluster restart to the end of all database configuration jobs regardless of expected service impact to make sure we clear out any connection pool issues.