Amazon, Paypal, Spotify, “Le Monde”… What we know about the global internet failure which has temporarily paralyzed many sites

A US company with many websites among its customers recognized a quickly resolved problem on Tuesday, ruling out a computer attack.


 

The incident is already resolved, but you may have experienced it. Many online services and media sites were made inaccessible for about an hour, Tuesday June 8 at midday, due to a technical incident.

Fastly, a company whose services allow these sites to gain speed, admitted to having encountered a problem, ruling out the hypothesis of an external attack. Franceinfo summarizes what we know about this incident.

Who was affected?

Of an unusual magnitude, the blackout affected a wide variety of sites, which it is impossible to list as a whole. At least one network of public services has been affected, that of the United Kingdom, where all sites in the gov.uk domain (equivalent to French gouv.fr) have remained inaccessible, paralyzing a number of online procedures. In the United States, the White House site was also concerned.

Amazon and the online payment service Paypal were also unusable, as were the streaming services Spotify and Twitch, and the social network Reddit. Twitter remained online, but the display of emoticons no longer worked.

If franceinfo.fr escaped the blackout, this is not the case with many other media sites such as The world and the other sites of his press group (L’Obs or Telerama, especially). But also the New York Times and CNN in the United States, Guardian and the Financial Times UK, The world in Spain, Corriere della Sera and Italy…

Where did the problem come from?

All affected pages displayed a message indicating a “503 error”. This means that the server on which the site is hosted is not able to display it, because it is off, in maintenance or overloaded, explains the site of Mozilla, the creator of the Firefox browser.

Eyes quickly turned to an American company, Fastly. While she has not, for the moment, taken responsibility for the problems observed, her Twitter account and her page dedicated to incidents have both mentioned a problem that she now says she has. “Identified” and resolved.

On its site (in English), Fastly presents itself as a service allowing its customers to speed up the display of their pages and to welcome more visitors.

Do we know if it is an accidental breakdown or an attack?

To designate the problems encountered on Tuesday, the company mentioned a “Impact on the performance of [son] service de CDN”, acronym which designates a content distribution network.

Concretely, Fastly does not host the sites of its customers, but copies of these sites, in geographically distributed servers, allowing nearby Internet users to access them more quickly and to avoid the worries in the event of peak connections. It is this service, obviously popular with merchant sites, streaming services and the media, that has been affected.

On Twitter, Fastly explained that a bad “Service configuration” had “Triggered disturbances” and that his “deactivation” had made it possible to restore his services. A vague description, but which indicates an internal error in the company, making it possible a priori to rule out the scenario of a computer attack or a material incident (such as the fire which affected the OVH host in March in Strasbourg ).

How long have these sites remained inaccessible?

The first message reporting a problem appeared on the Fastly site at 11:58 am (UK time). The company explained that they identified the problem and applied a fix at 12:36 p.m., and presented it as “resolved” at 14:41. While warning that the return to normal could be gradual among its customers.

The downdetector site, on which Internet users can alert when a site is not working, shows that the reports started around 11:45 am, that they reached their peak around noon, and that the situation had returned to normal around 1 pm.





 

Other news