so in a recent review of our personal budget we realized we're paying several major cloud providers for services we barely care about, most of which reflect old architectural decisions that we would do in cheaper and easier ways these days

for example, apparently in 2019 we thought it was reasonable to spin up a Google App Engine instance for the sole purpose of issuing a single http redirect

80 cents a day!!!!!! so unnecessary. so absurdly unnecessary

in the years since then we've set up a reverse proxy on our own hardware that does all the redirects we could ever want and all we pay for is bandwidth

and also the bandwidth is a much better rate than what Google charges

also apparently in 2019 we were big fans of www subdomains. we no longer feel that way lol

bah, lol, and we used bootstrap.css for this site, too. that's going to be a pain to rip out.

we also used sass but at least that one is no big deal to move off of (all the parts of it we care about are standard css now)

realistically we should turn the site into a static copy of itself as an intermediate step, so we can get off App Engine and stop the recurring billing. we don't have to fully redo the whole thing right now.
yay okay we made a static version of the site... now to set up the self-hosting for it... (this is mostly minor tweaks of existing config for the site we already host statically)
conveniently, the site has a few side domains which redirect onto the main domain, and one of the side domains we would actually rather use as the main domain. so we can test on that one first and then change the redirects to work in the other direction. neat.

plus we already have the ACME cert set up for that one

(what happened was we wanted a .org but somebody else had bought it and was using it for .... well, nothing that made any sense. so we bought the .com and used that for a few years, then the .org expired and we bought that as well and made it a redirect. .... and by the time we did that, we had our current infrastructure working, so we used that for the redirect. so, conveniently, the .org is already pointed to our "good" infra.)

hm the header image takes surprisingly long to load via the new hosting

oh well, we're gonna go ahead and redirect the other domains regardless

really, we want to move to hosting our own DNS - we did that at work recently as part of de-US-ifying our employer's infrastructure, not to be confused with our infrastructure - but for now we'll content ourselves to migrate it from one cloud provider to another

hm well, migrated the root domain very smoothly, forgot to do the www. subdomain in the same operation with it. oops.

so we're technically taking downtime right now but DNS propagation is masking it a bit. well, just a couple more minutes while the redeploy happens.

there. migration complete. if we said what domain it is here on fedi, thousands of servers would immediately begin hitting it, causing their respective recursive DNS servers to cache whichever copy of the records they have for another day or so. we'd rather wait an hour or two for DNS propagation before we do that, heh. but it's all looking great here.

a big thing we've been working hard on allowing ourselves to believe lately is that when it's a personal thing, downtime is okay

like. we're human (at least, the projection of our true form into this soap bubble that some call reality is human). we have a life, and sometimes that life requires us to do things like turn off all the circuit breakers for an hour or whatever. why should we want to hide that?

@ireneista back when i was doing the burning man thing on the regular, i would power off all my services and infra for the week. i called it "rebooting my exobrain". yes, my dns failed to resolve. no problem! i was off grid anyway.
@astrid oh what a cool ritual tbh
@ireneista plus like. knowing that infra is regularly put through a coldstart cycle, gives confidence that it will come up clean after unexpected downtime (and my inevitable benign neglect)