Facebook has revealed that its worst outage in four years was caused by an internal configuration management problem.
The social networking giant suffered a 2.5-hour global outage earlier today.
It said the outage was caused by "an unfortunate handling of an error condition."
"An automated system for verifying configuration values [also] ended up causing much more damage than it fixed," the company said in an outage report.
The only way to stop the problems occurring was "quite painful - we had to... turn off the site", Facebook said.
"Once the databases had recovered and the root cause had been fixed, we slowly allowed more people back onto the site," it said.
Arbor Networks used its ATLAS tool to map global Facebook traffic for 80 ISPs, showing a severe traffic fall as the site was taken out.