Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Tuesday June 08, @02:39PM   Printer-friendly [Skip to comment(s)]

Websites begin to work again after major breakage:

A major outage has affected a number of major websites including Amazon, Reddit and Twitch.

The UK government website - gov.uk - was also down as were the Financial Times, the Guardian and the New York Times.

Cloud computing provider Fastly, which underpins a lot of major websites, said it was behind the problems.

The firm said there were issues with its global content delivery network (CDN) and was implementing a fix.

In a statement, it said: "We identified a service configuration that triggered disruption across our POPs (points of presence) globally and have disabled that configuration.

"Our global network is coming back online."

[...] Fastly runs what is known as an "edge cloud", which is designed to speed up loading times for websites, as well as protect them from denial-of-service attacks and help them when traffic is peaking.

It currently looks as if the problems were localised, meaning specific locations across Europe and the US were affected.

Also at c|net


Original Submission

Display Options Threshold/Breakthrough Reply to Article Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 5, Insightful) by fustakrakich on Tuesday June 08, @03:09PM (10 children)

    by fustakrakich (6150) on Tuesday June 08, @03:09PM (#1143152) Journal

    The "cloud" was supposed to mean content is distributed (torrent fashion) over various servers around the globe, not a single company that can blow things up on a whim.

    --
    Ok, we paid the ransom. Do I get my dog back? REDЯUM
    • (Score: 4, Insightful) by Tork on Tuesday June 08, @03:32PM (9 children)

      by Tork (3914) on Tuesday June 08, @03:32PM (#1143162)
      It's not like the whole internet went down. *Any* service has a single point of failure when you go high enough up the chain.
      --
      Slashdolt Logic: "23 year old jokes about sharks and lasers are +5, Funny." 💩
      • (Score: 3, Informative) by Socrastotle on Tuesday June 08, @05:01PM (4 children)

        by Socrastotle (13446) on Tuesday June 08, @05:01PM (#1143199) Journal

        This isn't true. Well implemented decentralized services, like Bitcoin, have no single point of failure. Even if the entire internet was destroyed, the more robust decentralized services would continue to work with minimal to no modification on the resultant intranets that would rapidly spring up.

        • (Score: 2) by maxwell demon on Tuesday June 08, @05:43PM (3 children)

          by maxwell demon (1608) on Tuesday June 08, @05:43PM (#1143217) Journal

          I'm pretty sure a humanity-destroying asteroid impact would also terminate the Bitcoin network. Not that anyone would still care.

          --
          The Tao of math: The numbers you can count are not the real numbers.
          • (Score: 1, Funny) by Anonymous Coward on Tuesday June 08, @06:59PM (2 children)

            by Anonymous Coward on Tuesday June 08, @06:59PM (#1143243)

            Yea, nobody would care about bitcoin. But if an asteroid destroyed all smart phones, THAT would be the end of the world.

            • (Score: 2) by looorg on Tuesday June 08, @07:24PM (1 child)

              by looorg (578) on Tuesday June 08, @07:24PM (#1143250)

              But if an asteroid destroyed all smart phones, THAT would be the end of the world.

              It would have to be a pretty big asteroid. Humans, and/or the planet would not survive the event. That said it would be a fantastic and wonderful thing if all smartphones just selfbricked and we had to move on. We would all be better off.

              • (Score: 0) by Anonymous Coward on Tuesday June 08, @10:30PM

                by Anonymous Coward on Tuesday June 08, @10:30PM (#1143300)

                the "pigs" would be responsible for the asteroid, just was well there are people starting to fight [gadgethacks.com] back.

      • (Score: 2, Insightful) by fustakrakich on Tuesday June 08, @05:02PM (1 child)

        by fustakrakich (6150) on Tuesday June 08, @05:02PM (#1143201) Journal

        It was only a test, to demonstrate how easy it is to knock you offline. But the fault here is concentrating all the hosting in one place, exactly the opposite of how to run a reliable network.

        --
        Ok, we paid the ransom. Do I get my dog back? REDЯUM
        • (Score: 2) by Tork on Tuesday June 08, @05:41PM

          by Tork (3914) on Tuesday June 08, @05:41PM (#1143215)
          Is it common when hooking up with a service like this to get a list of other clients that service has? "We turned this service down because topographically it doesn't make sense."
          --
          Slashdolt Logic: "23 year old jokes about sharks and lasers are +5, Funny." 💩
      • (Score: 4, Informative) by PinkyGigglebrain on Tuesday June 08, @05:50PM

        by PinkyGigglebrain (4458) on Tuesday June 08, @05:50PM (#1143221)

        *Any* service has a single point of failure when you go high enough up the chain.

        So very true [xkcd.com]

        --
        "Beware those who would deny you Knowledge, For in their hearts they dream themselves your Master."
      • (Score: 2) by stormreaver on Wednesday June 09, @12:47PM

        by stormreaver (5101) on Wednesday June 09, @12:47PM (#1143498)

        *Any* service has a single point of failure when you go high enough up the chain.

        Exactly, which is what makes cloud computing so pointless. All you're doing is moving the point of failure from something you can control and fix to something you can't control and can't fix. The one and only thing that makes it in any way, shape, or form not 100% stupid to host your services on someone else's servers is the outrageous price of bandwidth. Remove that obstacle, and it becomes completely idiotic to not host your own stuff.

        When uncapped fiber to the premises takes over (we're starting to see it happen in some locations, such as my own), the whole cloud computer insanity will become obsolete. This whole thing has already played out in the past, which is how the PC revolution took off so fast: the failings of central control encouraged decentralization. This has happened before, and it will happen again.

  • (Score: 3, Funny) by iWantToKeepAnon on Tuesday June 08, @03:14PM

    by iWantToKeepAnon (686) Subscriber Badge on Tuesday June 08, @03:14PM (#1143155) Homepage Journal

    Distributed, robust, high availability, redundant systems ... till they aren't. : /

    --
    "Happy families are all alike; every unhappy family is unhappy in its own way." -- Anna Karenina by Leo Tolstoy
  • (Score: 5, Funny) by Anonymous Coward on Tuesday June 08, @03:15PM

    by Anonymous Coward on Tuesday June 08, @03:15PM (#1143156)

    None of the videos on Soylent would play this morning. None of them.

  • (Score: 3, Insightful) by gtomorrow on Tuesday June 08, @03:21PM (3 children)

    by gtomorrow (2230) on Tuesday June 08, @03:21PM (#1143157) Journal

    So much for "route around damage."

    • (Score: 3, Informative) by DeathMonkey on Tuesday June 08, @03:51PM

      by DeathMonkey (1380) on Tuesday June 08, @03:51PM (#1143170) Journal

      That saying is about TCP/IP, which continued to route packets just fine. It does not apply to any random service built like a seven layer dip on top of it.

    • (Score: 2) by maxwell demon on Tuesday June 08, @05:37PM (1 child)

      by maxwell demon (1608) on Tuesday June 08, @05:37PM (#1143212) Journal

      By definition, you only can route around damage on the way to your destination, not when your destination itself is damaged.

      --
      The Tao of math: The numbers you can count are not the real numbers.
      • (Score: 2) by gtomorrow on Tuesday June 08, @07:47PM

        by gtomorrow (2230) on Tuesday June 08, @07:47PM (#1143257) Journal

        I beg to differ. My (hypothetical) destination was the Financial Times|the Guardian|the New York Times|other and not Fastly (what a lousy name, under any circumstance). Ergo, "not redundant enough" or else "putting all your eggs in one basket", which is truly surprising (to me) for online entities listed above, especially so for www.gov.uk.

  • (Score: 1) by shrewdsheep on Tuesday June 08, @03:37PM (1 child)

    by shrewdsheep (5215) Subscriber Badge on Tuesday June 08, @03:37PM (#1143163)

    ... of this article is that there is only a single point to mention (see comments above).

    • (Score: 0) by Anonymous Coward on Tuesday June 08, @03:47PM

      by Anonymous Coward on Tuesday June 08, @03:47PM (#1143169)

      My ISP just shit itself a few minutes ago, so there's at least a second point of failure.

  • (Score: 2) by looorg on Tuesday June 08, @04:07PM

    by looorg (578) on Tuesday June 08, @04:07PM (#1143178)

    Didn't even notice. So nothing of value was lost during that hour. I'm sure there is something something here about all the eggs in one basket or that if you rely on others to run your business. Lessons, not, to be learned.

  • (Score: 5, Informative) by Hartree on Tuesday June 08, @04:17PM (1 child)

    by Hartree (195) on Tuesday June 08, @04:17PM (#1143183)

    The Cloud:

    https://xkcd.com/908/ [xkcd.com]

  • (Score: 1) by Rocky Mudbutt on Tuesday June 08, @07:55PM

    by Rocky Mudbutt (4659) on Tuesday June 08, @07:55PM (#1143260) Homepage Journal

    It's always DNS.

    --
    Ethics II Axiom 2. "Man thinks." B. Spinoza
  • (Score: 2) by looorg on Wednesday June 09, @11:01AM (2 children)

    by looorg (578) on Wednesday June 09, @11:01AM (#1143483)

    Apparently Fastly are now throwing some unnamed (or unknown) customer under the bus as the one responsible for the outage.

    https://www.fastly.com/blog/summary-of-june-8-outage [fastly.com]

    • (Score: 0) by Anonymous Coward on Wednesday June 09, @01:15PM (1 child)

      by Anonymous Coward on Wednesday June 09, @01:15PM (#1143505)

      They describe it as a software bug they themselves introduced, which was triggered by a valid configuration change by a customer. Perhaps they changed the text since you read it, but right now they are clearly not blaming the customer.

      • (Score: 2) by looorg on Wednesday June 09, @02:33PM

        by looorg (578) on Wednesday June 09, @02:33PM (#1143526)

        Fine. I think they might have changed it a bit then cause when I read it this morning (CET) it was more or less that while it was a software bug somehow a user managed to push their configuration onto the entire network and that made a lot of the other customers sites unreachable. They didn't want to name that customer. Not that I believe that it was malicious intent on the part of the user/customer but they still at that time assigned blame there instead of taking responsibility for their own software bugs. That said it took a long time before it happened, they pushed their "bug" out to the customers in may and it didn't trigger until yesterday. So if it was a known software bug one wonders why they didn't bother fixing it for almost an entire month.

(1)