From 07:23 UTC to 12:28 UTC our Heroku logplex endpoint was down. This resulted in Heroku Platform errors not being collected and alerted during this window. The problem was caused by a misconfigured AMI that, when booted, would not pass the ELB health check. The outage was as long as it was because our monitoring was not configured correctly to wake us up for failures with this endpoint. The AMI problem has been corrected and additional monitoring will be added.