At around 11:04PM GMT+3, our router dropped all route table records and multiple BGP sessions.
The majority of connectivity was restored 5 minutes later, at 11:09PM, however a certain subnet (18.104.22.168/24 did not get properly announced and was visible out of only a small amount of locations, via a specific peer only. This issue was resolved 50 minutes later and connectivity for all of our customers is fully restored.
We are very sorry for the inconvenience caused and we're activately working on resolving permanently this issue. Updates on our progress in permanent solving the issue will be posted the following week.
Sunday, August 20, 2017