At 05:00 (UTC) on August 26, due to routing changes in tier-1 backbone providers’ networks, a huge amount of traffic to AdGuard DNS from across Europe began to be routed to our Istanbul location (instead of the usual Frankfurt, London, and Amsterdam). Unfortunately, the Istanbul PoP is not capable of handling such a massive volume of traffic, as it normally only serves users from the Middle East.
However, instead of a complete outage (which we would have noticed immediately), overload protection mechanisms only started limiting user connections. This led to an “accumulation” of unresolved DNS queries on user devices. Because of this, we only discovered the incident at 06:30 and manually disabled the Istanbul location.
Immediately afterward, all accumulated user queries from the previous hour and a half were redirected to the more powerful Frankfurt PoP. Processing this backlog took about an hour, during which users could have experienced significant delays in DNS resolution.
All issues were finally resolved at 08:30
As a result of this incident, we decided to: