Security.ubuntu.com mirrors under DDoS?

41 points by argulane ↗ HN
security.ubuntu.com have been slow and unresponsive for this whole morning. Is anyone else experiencing the same slowness?

20 comments

[ 4.9 ms ] story [ 49.4 ms ] thread
I'm experiencing the same issues with both security.u.c and archive.u.c , status dashboard shows everything operational https://status.canonical.com/
(comment deleted)
Opens up just fine for me.
It does not work for me from Europe, where are you located?
I have some CI pipelines failing due to failing connections to archive.ubuntu.com
I added something like this to my Dockerfiles:

    # make it pick a mirror close to me
    RUN echo 'deb mirror://mirrors.ubuntu.com/mirrors.txt jammy main restricted universe multiverse' >> /etc/apt/sources.list
I am new to Linux but I ran "sudo apt update && apt upgrade" as I usually do and it's VERY slow on this line:

Get:1 http://security.ubuntu.com/ubuntu jammy-security/main amd64 thunderbird amd64 1:102.7.1+build2-0ubuntu0.22.04.1 [62,4 MB]

It's at 23.9 kB/s and taking forever! First I thought it was my mirrors (I'm on Linux Mint) and I changed them to like 5 different. But it's definitely security.ubuntu.com that's being slow!

Can't update as package downloads are super slow
My approach to avoiding downtime from slow upstream repos: I run an apt-cacher-ng service in my k8s cluster and have the containers in the cluster build with apt.mydomain instead of the upstream sources. Even under normal circumstances, this makes apt operations blazing fast.
https://status.canonical.com is green across the board, and yet archive.ubuntu.com it has been failing to load packages for a while now.

Not a massive problem for us... can imagine it might be a problem for many though (service agreement or no) - it does annoy me when status pages don't show an accurate status though. Makes me feel like the point of them is just to pretend they care.

Yeah, I have a project that I've been wanting to add a status page to for a while. I haven't done it yet because I want to display accurate information about the inner workings of the system, which will take actual time and effort.

Meanwhile, large organizations seem to almost universally just have a static page with a bunch of green checks that don't change no matter what.

Maybe I should just serve the status page from within the app, and if you can't reach it that means something is broken! At least then it's providing some information about the state of the system.

It’s finally showing “Major Outage” now, “Archive servers overloaded”.
My build pipelines have been failing the last 7 hours because of this.. Amazing that it's still green on the status page. Twitter reports a lot of affected users as well.
DDoS is a total guess. It does seem unresponsive, but I wouldn't jump to terrorism given no more information.

Most mirrors work, so you should be able recover by updating your `sources.list`. Of course this is a huge pain if you have many machines without centralized configuration management, or if you have tenants with virtual machines.