08.09.20
It Looks Like Red Hat’s (IBM) Fedora Project May be ‘Outsourced’ to Amazon’s Datacentres
Amazon datacentres in Ashburn (source: Wikileaks)
Summary: In “seeking a more modern and cost effective location” for Fedora Infrastructure it seems to have been decided, privately, that Amazon (AWS) would be the new home of this project; but there’s sufficient obfuscation surrounding the matter and many people seem to be totally unaware
OVERNIGHT we’ve spent several hours researching this and we also asked some Fedora people, who don’t seem to know exactly what’s going on. Surely some people higher up the corporate chain have a better and clearer idea, but they’re not likely to speak about it because IBM pays their salary and there’s fear of further layoffs. A lot of the finer details will be available in IRC logs (frustratingly enough, almost all the coding/code hosting was outsourced to Microsoft), but the gist of it all is this: while examining about a dozen services of the Fedora Project we’ve occasionally stumbled upon Amazon addresses. And then there’s this:
dig copr.fedorainfracloud.org ; <<>> DiG 9.11.5-P4-5.1+deb10u1-Debian <<>> copr.fedorainfracloud.org ;; global options: +cmd ;; Got answer: ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 34423 ;; flags: qr rd ra; QUERY: 1, ANSWER: 2, AUTHORITY: 0, ADDITIONAL: 1 ;; OPT PSEUDOSECTION: ; EDNS: version: 0, flags:; udp: 512 ;; QUESTION SECTION: ;copr.fedorainfracloud.org. IN A ;; ANSWER SECTION: copr.fedorainfracloud.org. 300 IN CNAME copr-fe.aws.fedoraproject.org. copr-fe.aws.fedoraproject.org. 939 IN A 3.81.0.123
This probably merits a little bit of context. A few hours ago Fedora Infrastructure had another very major OpenShift failure (not the first such incident over the past few days). Well, it wasn’t an issue until lately…
This so-called ‘cloud computing’ software is Red Hat’s brand (with GUIs and everything) of something that’s otherwise widely available as Free software but harder to use.
“Fedora Infrastructure Status: Major service disruption,” said the notice, reading out the extent of the impact: “New status major: issues with openshift cluster for services: Ipsilon, Badges, Blockerbugs, Package Updates Manager, Fedora Infrastructure Cloud, COPR Build System, Documentation website, Fedora elections, Account System, Fedora Messaging Bus, Fedora Calendar, Fedora pastebin service, The Koji Buildsystem, Koschei Continuous Integration, Kerberos, Mailing Lists, Module Build Service, Mirror List, Mirror Manager, Fedora Packages App, Pagure, Fedora People, Package maintainers git repositories, Fedora Container Registry, ABRT Server, Fedora websites, Fedora Wiki, Zodbot IRC bot”
“…while examining about a dozen services of the Fedora Project we’ve occasionally stumbled upon Amazon addresses.”And later on it said “New status good: Everything seems to be working.”
“We are now operating from the new datacentre and are bringing back capacity as time allows. The known issues are documented here in our wiki,” it adds. If one goes to the corresponding status page for pertinent services (“Fedora Project Datacentre Move”) it is linking to this wiki page, which in turn says: “This is a page to track known issues after the Fedora Infrastructure Datacenter move in 2020 from phx2 to iad2 and rdu2.”
Those are not descriptive names, as there’s not even a domain name/suffix. We tried tracing back the IP addresses of these services to identity where at least some these services were being moved to. “I think it’s more Red Hat management pressure to rely on Github and Google,” Ryan said, but it seems to be related to neither. “You know,” Ryan continued, “the infrastructure self-hosting was VERY reliable for many years before that.” He mentioned “GMail, Google Drive, calendar stuff” as examples, citing the anonymous post that said about Jim Whitehurst: “He didn’t use our products (used a Mac, along with other members of the executive management team), allowed RH to dump open source solutions for our own business to move things to Google’s services (which is a huge message to our very clients of “we don’t value or trust open source, so why should you?”)”
“A few hours ago Fedora Infrastructure had another very major OpenShift failure (not the first such incident over the past few days).”The migration pages are edited mostly by Kevin Fenzi (IBM/Red Hat), who wrote about it less than a year ago in the devel-announce mailing list. No replies to it as of this date.
Is Copr on AWS now? Seems so. If the new DC of Fedora is Amazon-owned and Amazon-controlled, there may be something to conclude. Depending on one’s opinion regarding Amazon and AWS, which turns Free/libre things into proprietary.
This message from last year says the hosting of various services is moving to the city/area in VA where AWS has many datacentres that already route some Fedora traffic. While we don’t know for sure what’s going on, as we need to better understand what plans are made in the boardroom (behind closed doors), it seems likely there’s a move to Amazon.
Four years ago The Atlantic published “Where Are Amazon’s Data Centers?” and it took note of Ashburn and Sterling. About a year later Wikileaks revealed that both places have lots of datacentres in them, mostly Ashburn (shown above is our screenshot of these).
Going back to Kevin, less than a year ago he wrote: “Fedora Infrastructure currently has the majority of its hardware in a datacenter in Arizona, USA. Red Hat leases this space for use by a number of teams, including Fedora.”
“This message from last year says the hosting of various services is moving to the city/area in VA where AWS has many datacentres that already route some Fedora traffic.”We can confirm, based on lots of checks, that many services still link to that datacentre.
“However,” Kevin continued, “they’ve been seeking a more modern and cost effective location for some time and have decided on one: So, we will be migrating to a new datacenter located in Ashburn, Virginia in 2020.”
Oh, Ashburn. Where Amazon has a densely-distributed cluster of datacentres.
Is Red Hat moving to ‘the cloud’ and, if so, why not its own? Does IBM feel like it needs to outsource that to a competitor/partner?
The details on this are still not certain and I could not get anybody from Fedora (in IRC) to confirm or deny. Quite frankly, many of them may not be aware. █
Content is available under CC-BY-SA