2021 Facebook outage
The Facebook Outage of 2021 was a service outage on the US social network, Facebook and its subsidiaries, including Messenger, Instagram, WhatsApp, Mapillary and Oculus, starting at 2:40 pm UTC on October 4, 2021. The networks were unavailable globally for more than six hours. The outage was caused by a Border Gateway Protocol (BGP) removal of all IP routes to its Domain Name Servers (DNS), all of which are then self-hosted. After these platforms stopped, users began to migrate to Twitter, Discord and Telegram, resulting in server outages for these applications. Facebook's DNS services began to be accessed again at 9:05 pm. At 10pm, Facebook reported that app layer services were slowly being restored to Facebook, Instagram and WhatsApp after more than six hours of outage. Gradually, WhatsApp, Facebook and Instagram returned to work on some devices.
The outage came a day after former Facebook product manager Frances Haugen relayed information to CBS News' 60 Minutes, saying Facebook, Inc. was more concerned with "growth than security." The report showed that Facebook gave preferential treatment to politicians and celebrities, having been partially used to organize the attack on the US Capitol in 2021. Haugen is expected to testify before a Senate subcommittee on 4 October.
Security experts identified the problem as a Border Gateway Protocol's removal of the IP address prefixes on which Facebook's domain name servers were hosted, making it impossible for users to resolve Facebook and related domain names, and reach the services. The effects were visible globally: Swiss internet provider Init7, for example, saw a massive drop in internet traffic to Facebook's servers after the change to BGP. or receive external emails, access the corporate directory, and authenticate to some Google Docs and Zoom services. The New York Times reported that employees were unable to access buildings and conference rooms with their security badges. The website Downdetector, which monitors network outages, has recorded tens of thousands of incidents around the world. Steve Gibson, security researcher, said a "routine BGP update went wrong" blocking "people with remote access" to servers to fix the bug, with people with physical access not authorized to fix the bug. around 3:40 pm Facebook made a significant number of BGP updates, announcing the removal of routes to their servers. At the same time, Facebook's DNS servers went offline. Around 3:50 pm, after the propagation of route updates, the Facebook domain name became unavailable in Cloudflare's DNS resolver. Around 9pm, Facebook resumed announcing BGP updates, with the Facebook domain name becoming resolvable again at 9:20pm. Facebook began to have its services gradually restored after a team gained access to computers in the data center located in California and restarted the servers. By 10:28 pm, Facebook's main services appeared to be reconnected to the global Internet. A CNBC report said the outage was the worst Facebook has experienced since 2008. On the day of the stoppage, the company's shares fell nearly 5%.
Facebook's chief technology officer, Mike Schroepfer, drafted an apology after the time.