3w ago

Solarpunk instance going offline for awhile was like a true solarpunk experience

First of all, I love you, slrpnk admins. You handled it all like champions! Happy to have you back.

Someone said in one one of the matrix chats during the outage that the fact that this instance is selfhosted and went down is like true experience of a solarpunk world; there won't always be power and that's okay!

Just wanted to write it down in a post since that message was so spot on.

24 comments

But then Lemmygrad would have to shut down every day after working hours as in the socialist worker's paradise these things are taken very serious and no bureocrat will work even an minute longer than the official working hours. /s
Joking aside, yes... hosting on older hardware and on a regular fiber connection does mean that if hardware breaks it can take a while to replace.
We have a bit of redundancy with PSUs and so on, and this specific issue we had now could have been avoided (and will be in the future), but we do trade off some reliability for the fun of DIY self-hosting (and it is also much cheaper).
- Ha true!
  And also I love that, please keep it fun. I feel like it's truly a solarpunk instance that way :)
- I was wondering why you mentioned Lemmygrad so I did a quick research, damn, you've been beefing since 2022?
  
  Even before that 😅
  But it's fine, they can have their online safe-space for circle jerking about Stalin and Mao if they want 🤷
- I would assume they have a five-year hosting plan. ;)

That's how I got through it. Reminded myself that in a volenteer-based mutual aid society, 99.999% uptime garauntees aren't a thing and that would be good because it would apply to my work as well. Services go down, and as you shouldn't be addicted or otherwise dependant on such services, you can do something else with your time.
- 99.999% uptime guarantee is a silly thing, once you realize that all things that have to be just available all the time - some health or security or manufacturing lines - get 10 minutes downtimes easily with all these fancy nines. It's just some cliche to disempower us punks.

First of all, I love you, slrpnk admins.
First, yes, <3 to all.
Then:
Meh, I consider that resilience is not opposed to sustainability.
We don't have to prioritize it right now, and I will always be grateful of volunteers who do the best they can with what they have.
And to me the lesson was that several communities (french-speaking jlai.lu) still worked and I just used an alt there to continue the conversations I had, and that I could still access through them the past conversations in slrpnk.net. This is a testament to the resilience of the fediverse architecture.
I just disagree with the sentiment that we should somehow romanticize power outages. Some people need a reliable power sources to survive, and it is a totally preventable thing. We can route around them in a radically different way though.
But thinking "this could be down for a week with no warning" also implies that I need to keep a fallback mechanism if I am using it to organize any sort of event with people.
- But thinking “this could be down for a week with no warning” also implies that I need to keep a fallback mechanism if I am using it to organize any sort of event with people.
  That is probably a good idea regardless of the external service you use. The best is IMHO if you have a group of people for an event to host something yourself, but of course you are welcome to use slrpnk.net as long as you can accept the above mentioned risk.
  That said, overall our uptime is quite good, so it is nothing I think you need to worry about all that much.
  
  Really this is just a general remark, not something specially related to slrpnk.net and you are doing an awesome job, that I am grateful I don't have to do. Please don't see it in any form as a criticism of this server!
  I am just saying that we should not embrace downtimes as an unavoidable thing or a problem we need to live in order to have a sustainable world. Resilience is also an important aspect of solarpunk.

In addition to this excellent point, during the outage I saw several people asking which server to switch to from .ee, and the replies made me realize how much I love it here. My fellow Slrpnkians didn't get all bent out of shape and pissed about things the admin just simply couldn't control at the time. From what I saw (maybe it's different in private messages/behind the scenes, iunno) any people talking about the skrpnk outage from their alts were all being very reasonable, understanding, and just being generally decent netizens. I love it around here.
- That's the kind of commenting I saw from slrpnk users as well!
- I feel like downtimes are a badge of honor for self-hosting in some ways. Being more efficient and minimal means there will be slightly less redundancy and that can be a good thing. Perfect uptime to avoid lost revenue during downtime is a capitalist craze, and not how an ecological project should operate.
  
  Indeed. The instances that have solid uptime have, in most cases, sold their soul to the devil (aka Cloudflare, which is a centralised threat on the free world and all things good).
  
  Totally agree - there's something authentic about accepting occasional downtime as part of the self-hosted journey, plus it encourages redudancy planning like those portable LFP power stations you can compare on gearscouts.com which fit perfectly with solarpunk ideals of energy independence.

We appreciate the slrpnk admins <3 them all.
Down time is fine, I support the goals

"Well shit, a UFO just crashed in the middle of the town square. That'll take some time to sort out. Whatever. I'll fix up my bicycle and go see what's going on in the next town over."
(Edit: spent this time getting acquainted with PieFed through piefed.social. Seems like pretty nice software! Hope slrpnk will migrate, as they suggested.)

I just assumed it was Earth Week

Why stop with power? We should aim for water outages as well! Just when I was starting to think this might not be a pathological ideology...
Good luck to the admins though, I appreciate the effort to stay out of cloud. It's hard to maintain constant uptime when you don't have teams of people dedicated to keep stuff afloat.
Now when I think about it, is it even possible to run lemmy on a cluster, like a regular web app, or is it a monolith?
- Happy cakeday :)
  My home country has a general guideline for all the inhabitants: keep stocked up on emergency food water + some other stuff enough that you can survive 3 day power, water and other outages.
  Been living with that in mind for a long time!
  Never been under a true test though.
- Now when I think about it, is it even possible to run lemmy on a cluster, like a regular web app, or is it a monolith?
  There's a page for info on horizontal scaling Lemmy here: https://join-lemmy.org/docs/administration/horizontal_scaling.html
  It does seem to be out of date (for one, it mentions that pict-rs can't be horizontally scaled due to the sled database, but pict-rs now supports postgres so maybe that has changed.
  I know larger instances run a separate federation container and many instances (including mine) run multiple lemmy-ui containers for the front end, but as I understand it the Lemmy backend can only run as a single service.

Downtime by a freedom-respecting trully decentralised node like slrpnk.net really exposes how Lemmy clients leave a LOT to be desired.
The prospect of data loss is gutting. A proper client would be syncing threads of interest between the server and my PC, so during downtime I can still at least locally access past content. No proper clients exist for Lemmy.

24 comments