In Review, November '22
I don’t know how regularly I’ll do posts like this, but November was a big month for the fediverse and for sunny.garden, so it seemed appropriate.
sunny.garden went online April 27, 2022, and stayed invite-only until about a month later when I listed on joinmastodon.org. Some days there were no registrations, and some days there were several, but on average I’d get about 1 per day.
By late October, the server had grown to approximately 150 users total, with maybe 50 active that month. Then Elon’s purchase of Twitter completed, and sunny.garden started getting 20 signups a day, and the user count doubled over the weekend. Then doubled again in a couple of days, getting 100-200 signups per day.
Each registration takes at least a minute to review and approve, so 200 per day was too many. I started opening registrations for only 1-2 hours a day and closing them again after 20-40 registrations. Even just doing this was quite a bit of work, plus it was a hassle managing invites for people to come and join their friends on the server.
With no sign of the Twitter exodus slowing down in the short term, I eventually decided to delist from joinmastodon.org. This way I can keep registrations open all the time, I don’t need to deal with invite links, and it also greatly reduces the number of people choosing to join the server mostly at random, after being forced to choose some server from the list.
This was the admin dashboard at the beginning of November…
And a month later at the beginning of December…
Signups have come way down, back to 1-2 per day on average, and I’m happy with that.
If people want to join their friends here, or find the server organically and it’s appealing to them, that’s great, sign on up! But I’m not interested in growing any more for growth’s sake.
I manage the Mastodon installation myself on a VPS hosted at OVH. I had started with 2 CPU cores, 4 GB RAM, and 80 GB of disk space, which was much more than the server needed until October, but I figured it would be nice to have the room to grow. This cost about $30/mo in total.
With the influx of new users, I did a couple of upgrades to the hosting plan, increasing to 4 CPU cores and 8 GB RAM. I also moved storage off the server’s local drive and onto OVH’s object storage – essentially infinite space, billed by the GB for storage and bandwidth.
Enough people asked if they could donate to help support the rising server costs, so I set up a ko-fi account to accept donations.
Thank you, to everyone that has donated! There’s no obligation or expectation that you donate in order to use the server, and I am still able to cover the costs myself if needed.
Now that it’s been about a month since I started accepting donations, here’s a breakdown of the costs for November:
Expand text version
|Totals for November
Assuming that we don’t grow that much more, these costs should stay relatively stable, although storage costs will grow slowly over time.
What the heck is a Test Server
If you’re wondering what a Test Server is, it’s a mostly complete copy of the main server, which I use to test things like doing the upgrade of Mastodon from v3.5 to v4.0, or moving storage from the local drive onto object storage. That way I can be sure that I know it’s going to work properly before doing the same thing on the live system everyone is using, and hopefully reduce the risk of any unexpected downtime.
You might also reasonably wonder why it costs twice as much as the Main Server and this is mostly because OVH’s billing is confusing and I wasn’t paying close enough attention. I paid for the test server at hourly rates, which are double the cost of paying by month. I wasn’t sure if I was going to want to keep it around all the time, but because I did, the charge was pretty high. It has been useful enough that I’m going to keep it around, and have now switched it to monthly billing. It should be half the cost going forward.
The real cost to me personally is the time and attention needed to both operate the server and deal with moderation, from onboarding people to dealing with user reports.
The first week or two of November was exhausting, handling 700 or so new users, plus working through the upgrades. This is the main reason I’m not interested in growing for growth’s sake. I want this to continue being a hobby that I enjoy, and not a business that I have to manage. I think I’m comfortable with the current size of the server, and if things stay stable or grow slowly that will be great.
I’m very happy with the wide variety of folks that have decided to sign up on sunny.garden with all their various interests, hobbies, and backgrounds. I feel vindicated in my choices of branding and target audience in attracting the sort of people I want to be around!