One of the websites that went down on the morning of 30 October was the high-traffic center-left news website Huffington Post, owned by AOL.

The site was hosted at one of the several data centers that went down that morning. This was ISP Datagram’s facility at 33 Whitehall St. on Manhattan Island in New York City.

The facility’s basement was completely flooded with water amid a power outage, so its backup generators had to be switched off to cut all power at the building out to prevent a fire.

Along with HuffPo, this data center hosted web properties of Gawker Media, including popular gossip blogs Gawker and Gizmodo, among others. While Gawker sites stayed down through Wednesday, company posting content on stripped-down “back-up” sites, HuffPo came back online during the day on Tuesday.

Its CTO John Pavley gave an interview to a Huffington Post reporter about what had happened.

“It was really a difficult night,” he said in a video interview about Monday night. “Most of the technology team that we have in New York was awake all night long.”

Turns out only portions of the main HuffPo website were hosted at Datagram’s data center, albeit portions crucial enough to make the entire site inaccessible. Its data center infrastructure is geographically dispersed, including the primary Manhattan facility, a failover site in Newark, New Jersey, and a widely dispersed private cloud operated by its parent company AOL.

At first, the team migrated the crucial portions of the website to Newark, but something happened with the back-up site, he said without clarifying what had gone wrong. “But then we had an issue there and then we started really scrambling.”

They put up a temporary site and got to work restoring the functionality, while a more limited temporary site was put up.