Boosted by @GroupNebula563@mastodon.social
Replying to
@NanoRaptor@bitbang.social I mean, this is one of my more recent contributions to Internet culture!
Plus the variations:
DANA YES
and
DANA WHY
Remote
Dr. David McBride (dwm)
@dwm@mastodon.social
mastodon 4.8.0-nightly.2026-10-06I'm not that kind of doctor. Senior IT technical specialist and occasional patient magnet.
Currently hostmaster@durham.ac.uk; previously answered to root@doc.ic.ac.uk, unix-support@csx.cam.ac.uk, postmaster@cam.ac.uk, oxcert@infosec.ox.ac.uk, and root@stats.ox.ac.uk.
225 Followers
289 Following
42 Posts
Joined January 03, 2017
Pronouns:
he/him
Website:
Github:
The thought occurs that, now that the likes of Facebook, Twitter, Instagram etc. now predominately show you random shit off the Internet rather than what your actual friends you want to see are posting, it would be more accurate to call them antisocial media.
Friday this week (July 31st) is Sysadmin Appreciation Day.
https://en.wikipedia.org/wiki/System_Administrator_Appreciation_Day
#sysadminday
RE: @RinaVolpina@rubber.social
I'm just going to take a giant sip of coffee and read this email I've just received from Amazon about my account.
Replying to
@cR0w@infosec.exchange I remember looking at this a few months ago and thinking, "What is the point of this?"
https://mastodon.social/@dwm/116668003463453723
Quoting
I'm looking at https://dns-aid.org/ and https://datatracker.ietf.org/doc/draft-mozleywilliams-dnsop-dnsaid/ and not understanding why it doesn't make proper use of conventional DNS-SD SRV records, insisting on inventing their own, AI-agent-specific thing.
The ideas of DNS-based service discovery, central registries, and RPC are absolutely nothing new.
And not being able to indirectly point to other search domains seems like a serious omission.
I've had a look at this, and I think this is the wrong way to look at things.
rsync is load-bearing, mature, valuable software. It's featureful, it's complicated, it's useful, and it's everywhere.
And it's written in C.
And people have figured out that even relatively basic LLMs, hooked up to an appropriate harness, can be used to drive processes that once required lots of manual work.
Including vulnerability discovery.
Replying to
@Tvorsk@meow.social @NanoRaptor@bitbang.social
Not to be confused with the
DAAAA NA-NA-NA-NA-NAAA
DA NANA NA NA RAPTOOOR
[katamari drum solo]
This is fucking atrocious. UK folk, write to your MP, this is not okay.
Multitasking, v: The act of avoiding multiple sources of work at once.
I somehow missed these when they were first published.
https://www.rfc-editor.org/rfc/rfc8962.html
https://www.rfc-editor.org/rfc/rfc9948.html
I am also entertained that errata exist for these formal internet standards documents.
https://errata.rfc-editor.org/eid6516/
"It's a Superpower??" debuted today, and it's very, very good. I laughed out loud so much, in ways nothing has really prompted me to do in ages.
It's good. Go in cold. Trust me.
I am reminded of the time that, on the day I implemented DNSSEC validation on the university recursive nameservice, ac.uk broke its DNSSEC signatures.
Fortunately, I had a plan for that...
Replying to
@jwz@mastodon.social I am assuming it's 80/443 traffic that is the problem.
If so, I would try fronting Apache with nginx, and using it's GeoIP module to return your HTTP error response of choice to HK.
e.g.
- Move Apache to 127.0.0.1:8080;
- Install
nginx, have it listen on TCP 80/443; - Configure the GeoIP module as described at https://nginx.org/en/docs/http/ngx_http_geoip_module.html
- Match against those locations you wish to block and return your HTTP error response of choice.
- For everything else, reverse proxy to Apache.
RE: @mullvadnet@mastodon.online
I'm really not keen on this "papers, please" Internet environment that the UK is rapidly deploying.
Replying to
@jn9999@mastodon.social @portaloffreedom@social.linux.pizza If that is a concern, then more general network traffic anonymisation measures (e.g. VPNs, Tor, etc.) are necessary.
Replying to
@PeteLittle @losttourist @cstross @tek
But this should ensure our survival! Aliens now finally have a place with a lifeform that can usefully metabolise that toxic Marmite® and use that energy constructively, such as by making puns.
github.com's lack of reliability is a serious problem.
RE: @GossiTheDog@cyberplace.social
oh for fucks sake
crimes detected
Replying to
@cstross@wandering.shop ... which some autobots apparently think it's real: https://beach.city/@vikxin/115790210510364299
I'm looking at https://dns-aid.org/ and https://datatracker.ietf.org/doc/draft-mozleywilliams-dnsop-dnsaid/ and not understanding why it doesn't make proper use of conventional DNS-SD SRV records, insisting on inventing their own, AI-agent-specific thing.
The ideas of DNS-based service discovery, central registries, and RPC are absolutely nothing new.
And not being able to indirectly point to other search domains seems like a serious omission.
These are not serious people.
"Exclusive: Meta planning sweeping layoffs as AI costs mount"
> prepare for greater efficiency brought about by AI-assisted workers
[citation needed]
Replying to
@portaloffreedom@social.linux.pizza It starts at the root servers, and follows up where they point. Ultimately, each domain owner is responsible for serving answers from their authoritative DNS servers to systems that ask. A recursive nameserver follows the chain.
Sure, you can use 8.8.8.8 as your recursive server if you want. But you can also just run your own.
Replying to
@portaloffreedom@social.linux.pizza A recursive DNS resolver like BIND or Unbound will, for any arbitrary query, start at the public DNS root (for e.g. .com) and recursively ask the authoritative nameservers for each domain individually until you get the name you want.
This is how DNS services work. You don't have to use someone else's; you can use your own.
(The downside is that DNS queries go over the wire in clear, so someone snooping on your DNS traffic can see what questions and answers you're getting.)
Replying to
@gamingonlinux Yeah, trying to order, and Steam is saying there are problems proceeding with the transaction. Clearly under quite heavy load..
Replying to
@maltimore@social.tchncs.de @beasts@social.mythic-beasts.com are great.
And have recently reduced their prices: https://www.mythic-beasts.com/blog/2026/04/13/domain-price-reductions-2/
Replying to
@unixbhaskar@mastodon.social The criteria for evaluating applications is given in the linked job description; age is not a consideration.
Replying to
@jwz@mastodon.social (nginx also has a caching module, which could be useful to you depending on traffic patterns.)
Replying to
@portaloffreedom@social.linux.pizza I run my own, on my own local machine. apt install unbound pretty much covers it.
Replying to
I should add that you'd have me as a senior technical colleague. Hopefully that doesn't put anyone off. :)
https://www.bbc.co.uk/news/articles/cdx85vkk0gko
"You know, I'm so sick of Congress I could vomit."
— Joshua Lyman, "The West Wing"
@bjh21@wandering.shop This may be relevant to your interests.
As context, you may find BG's inaugural lecture informative/entertaining:
https://www.bennett.ox.ac.uk/blog/2023/11/ben-goldacres-inaugural-lecture/
I remember thinking that it's always a good sign when ~rja14 has nice things to say about your security design.
Bonus points if you spot me in the audience. ;-)
Replying to on mastodon.social
(This is not the same role that I previously advertised: https://mastodon.social/@dwm/115939807010381625)