There were no end-user visible certificate errors during the outage, so I suspect it was an internal/backend cert.