I’m looking for input on how well Piefed responds to occasional downtime of up to a few days, specifically regarding how federation recovers after such an event.
Basically looking to know how amenable it is to selfhosting via reverse proxy from dodgy setups which are the best some of us proles can manage. I try to run some services where I can but my situation inevitably results in an event or three per year, on average, with some days down.
PieFed will retry the sending of every federation activity for 4 days before it starts dropping things. So if you’re offline for 3 days then it’ll be ok. But at the moment 95% of the federation traffic is sent by Lemmy servers so their retry window is the more relevant one. I don’t know what theirs is but I think it’s much longer than PieFed.
If you compare https://lemmy.fediverse.observer/list with https://piefed.fediverse.observer/list and look at the uptime columns, you’ll see Lemmy and PieFed are roughly equal. PieFed has slightly better uptime on average but instances are younger and have much less users so… grain of salt. The point is they’re in the same ballpark. Hardware and network failures are going to be the thing that has the biggest influence.
This is very useful info, thanks. I am definitely interested in PieFed for other reasons too so I will be giving it a look :)
If I could make one suggestion, it would be to extend the retry time, perhaps with a backoff in time between retries. Just speaking as somebody whose main issue with occasional outages is having something break when I’m having a brain fog episode and literally cannot rally the mental resources to fix it right away.
Yes, it has a backoff. I agree 4 days is pretty short, when I was coding it I had no idea what impact it would have on performance so was pretty conservative but things have been going fine so I’ll extend it and see how things go.
Wow, this is super cool to hear, thank you!
I see you’re looking at options to leave SDF too 🙂
I left SDF and switched to Piefed a few weeks ago myself. It hasn’t gone down on me yet, so I can’t answer your question. But it seems pretty solid.
It does have small federation quirks if you’re used to Lemmy, but nothing major.
What’s going on with sdf?
Some downtime recently, and recurrent issues accepting media uploads. The Lemmy instance is a bit sickly and so far it’s been difficult for them to fix.
Some downtime recently
That’s the understatement of the year 🙂
I’ve been on SDF for two years and I can’t remember it functioning well or fast for more than a month total - when it doesn’t simply go tits up for days at a time. And I paid for it, so I’m not super-happy.
My communities are still there, but I’m still evaluating Piefed with a view to moving them away from SDF permanently.
It’s too bad because SDF has two major advantages over all the other Lemmy instances:
- They virtually don’t defederate with any other instance unless they absolutely have to, so YOU get to choose which YOU want to block.
- I kind of like the SDF philosophy of an old-style Unix ecosystem that you pay for because you use it. The problem is, you pay for it and they don’t deliver. So, kind of a letdown.
But in fairness, I only have a beef with the Lemmy server and whoever is asleep at the wheel administering it. I have no issues with other SDF Fediverse services, and they’ve always answered my emails - if they don’t answer their internal IRC thing, they do answer emails. If Lemmy wasn’t the main thing I’m interested in in the Fediverse, I’d be pretty happy with SDF I guess…
FWIW I’ve only noticed one or two instances where it was down for what seemed like a day or so, but I only ramped up my threadiverse use in maybe the past 6 weeks.
My guess is that whoever runs the instance doesn’t actually use it personally, so they don’t really notice when stuff breaks.
And yeah, the minimal defederation policy was a big selling point.
Oh wow, is it true that you only have a single admin (the sidebar area of your instance says that at least), and that the last post or comment from them was TWO YEARS AGO? (https://lemmy.sdf.org/u/SDF) And at the time they were welcoming people to Lemmy 0.18 https://lemmy.sdf.org/post/347031/1001724? If so you may have another Kbin+Ernst situation going on with an abandoned Lemmy instance, and in that case if you can’t raise them on Matrix or Discord or wherever else you know you can reach them (I did not see any such link in the sidebar though), I would absolutely jump ship ASAP before it goes down.
However, that simply can’t be true since your instance says it is running 0.19.11 now so it must have been upgraded sometime since then, albeit unannounced afaict. Maybe the database is corrupted and the timing of posts are off, maybe this is expected behavior for Lemmy when the last post is two years ago and the admin has not communicated since (at least from that account, but then why go ninja to hide their identity?), or some other funky thing going on.
Regardless it does not sound great. PieFed is great though, so if you go that route then welcome:-).
Yeah, somebody eventually fixes stuff if people complain in [email protected] but never responds to any of the threads in there … weird.
I did just create a new community there today to scratch a personal itch, and there are a couple fairly active communities hosted there, so … guess we’ll see. SDF.org has been around since before HTTP was invented, so the organization isn’t going anywhere. I’m a little less sure about the Lemmy instance.
I’m not specifically looking for options in that regard (yet), but I’m mulling the idea of running my own server with closed signups just to have a place to potentially slurp in RSS feeds and such for people who want to subscribe to them. I haven’t touched PieFed yet but I’m definitely interested in it because I’m more familiar with the Python ecosystem than Rust.
Piefed doesn’t currently natively support setting up an rss feed to populate a community (though there is an issue for it on the codeberg, so hopefully soon). However, we do have an api at this point that is feature complete enough to cook something like this up without too much trouble.
As for self-hosting piefed, it is pretty comparable to lemmy. They both are dockerized and I have a personal instance of both of them. Resource-wise, for a personal instance, I haven’t really noticed a difference, though my VMs are well enough provisioned that I doubt I would run into any bottlenecks for a 1-user instance.
I’m more familiar with the Python ecosystem than Rust
This is basically how I have ended up becoming a piefed contributor. There were features that weren’t in lemmy that I was frustrated by, but don’t know a thing about rust. So, I just started implementing things in python…and have just kept doing it.
I haven’t written a lick of code in years, so I figure it might be good to start where I have some familiarity … and I would prefer Python over PHP in that case, those being my two best-known languages!
We are on the verge of adding a RSS feed plugin… https://codeberg.org/rimu/pyfedi/issues/1279
Interest increasing!
