From 17:45 until 02:45AM UTC users may have experienced intermittent HTTPS request failures. This was due to a misbehaving NGINX server that needed to be restarted. Unfortunately, the failure rate was sufficiently low to avoid escalation by our alerts but high enough that is may have led to problems using the service. We are adding more aggressive alerting to avoid future impacts in the event of a recurrence of this problem and will be evaluating the stability of our current NGNIX configuration in light of this incident.
Posted about 3 years ago. Apr 13, 2015 - 03:29 UTC
This incident affected: Website, SSH, Git via HTTPS, Mercurial via HTTPS, and Source downloads.