On September 30th, in a 20-minute period between 15:24 and 15:44 UTC, a rolling restart caused 6,299 5XX errors served to 25 customers. The rolling restart was self-healing and did not require developer intervention.
In an effort to problem-solve an unrelated concern, a service in the Spreedly Production Environment was restarted. The rolling restart cause intermittent 5XX errors to be returned as the Core API Service attempted to authenticate transactions against the restarting servers.
Spreedly will investigate and implement strategies for better application resiliency during temporary service outages.