While I do enjoy Twitter, I believe it's important to "own my platform." As such, Twitter's not material to the functioning of my business.
But I do talk to a lot of folks here, and a "subscribe" button for @LastWeekinAWS in my profile can't hurt anything...
I started by emailing @revue and checking their Terms of Service. As of today, there's nothing against using their sign-up function and exporting the list to another platform unless I'm directly monetizing the subscribers via subscriptions.
I am not.
(Pro tip, don't build things that are going to get yanked away when one of your vendors finds out about it. Ask rather than hoping to sneak by.)
Revue has an API that lets me do all kinds of things with it. getrevue.co/api
A few tests with curl later, and I knew it could work. I set this up via Zapier.
("Why not a Lambda function?" Because I presumably might want someone else to maintain this.)
Since @revue forces a double-opt in confirmation process I can bypass the one I have in @ConvertKit; Zapier now takes new subscribers and injects them into a slightly modified onboarding flow.
To be clear: I would cheerfully pay @revue money for this, but the only way to do that right now would be to charge subscribers to the list. I'm... not interested in doing that.
There are only a couple of rough patches on the @revue side; I imagine those will get sanded down with time. DMs are open if anyone on that team is curious.
I don't manage unsubscribes in Revue, because only transactional messages go to subscribers from there, and only when they first sign up. If folks are already subscribers, nothing changes in the existing newsletter. It's kinda elegant!
All in all, not bad for a couple hours of work. Onward and upward?
• • •
Missing some Tweet in this thread? You can try to
force a refresh
One of the hands down most sobering conversations I’ve ever had was with a bunch of Very Savvy Investment Bankers about what exactly a total failure of us-east-1 would look like economically.
The *best case* outcomes closely resembled a global depression.
That’s comfortably within their probability models for “a US civil war.” They’ve drawn up maps that show likely sides for such an event and they plan accordingly.
These people get paid significantly more than most engineers do to consider risk.
This is why you’ll not find even the most die-hard all-in cloud customer who’s publicly traded who doesn’t have “rehydrate the business” level backups either on-prem or in another provider.
And now, a thread of Ancient Sysadmin Wisdom: an incomplete list of things we have learned from decades of outages.
"It's always DNS." Yup. Everything relies upon DNS, those relationships are non-obvious, and some things like to cache well beyond your TTL.
"If an outage lasts more than ten minutes, it's likely to last for hours." Yup. Usually related to electric power, but this is a good rule of thumb for "do we activate our DR plan" decisions.
So today I'm going to store 1GB of data in @awscloud's S3 and serve it out to the internet. The storage charge is 2.3¢ per month the tier 1 regions.
Someone on the internet grabs that 1GB of data once. I'm paying 9¢ to send it to them. You read that right; just shy of four months' of storage charges to send it to the internet once.
I might get yelled at for this thread, but we'll give it a shot.
I'm not sure anyone needs to hear it as much as I needed to hear it myself a decade and change ago.
If you work in tech, either as an employee or as a consultant, most people you encounter *will not understand what you do*. "Something to do with the computers" is the best you can hope for.
They may be vaguely aware of a few additional facts. Such as "the company claims that people are their most important asset but pay the people who work on the computers three times what they pay the people who work in HR."