OVHcloud cut off from the world (in IPv4): feedback on a global failure

OVHcloud coupé du monde (en IPv4) : retour sur une panne à l'échelle mondiale

Sébastien Gavois

By Sébastien Gavois

Wednesday 13 October 2021 to 19:

This morning, some discovered that many sites and services were offline. Demonstration of the strong presence of OVHcloud on the web in many countries. A human error which led to an IPv4 routing failure which occurs a few days before the Roubaisien IPO.

This morning, at 9 a.m. 12 (Paris time), social networks are agitated because a large number of sites are inaccessible: National Assembly, Arte, Arrêt sur Images , the government’s open data site… They would be too numerous to name them all. Their common point is quickly found: “ the entire OVH network is unavailable ” , recognizes the host on its site dedicated to support … which was also inaccessible during the outage.

An error in United States drops the whole network 2021

Around 10 h, Octave Klaba intervened on Twitter to give some details: “ Following a human error during the reconfiguration of the network on our DC to VH (US-EST), we have a concern on the whole backbone. We will isolate the DC VH then fix the conf ”. He explains that “ in recent days, the intensity of DDoS attacks has increased a lot “. To cope with this, the host has decided to increase its processing capacity “ by adding new infrastructures in DC VH (US-EST)


We know the following: “ a bad configuration of the router caused the failure ”. An official press release containing these elements was subsequently posted online. The post mortem has just been published. Analysis.

When you are told to switch to IPv6

Back to top button