New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Pruning unused accounts #18
Comments
@globz, I think it's fair to say it's too early for us to know how we'll handle users who haven't updated their tilde.club home page yet. Some of them probably haven't gotten to that part of this experiment yet — they might be messing around with shell scripting, learning their way around a command prompt for the first time, interacting with other users via chat as they get their sea legs... and for others this might be a one-and-done phenomenon. But my guess is that it's going to take us a bit of time to figure this all out. |
Other systems I've been on have had a "grim reaper" that expires unpaid, unauthenticated, unbefrriended accounts after some period of inactivity. You might make use of the Unix groups system to automatically move people between categories and apply policy based on that. |
@delfuego Yeah I totally understand because there's no previous reference on how this community will turn out or if it will survive at all, it is an experiment after all! I think what's important to know first, is if you guys want to grow and increase the size of the community, all on a single machine? everyone into tiny clusters of 600+ members? @vielmetti Yeah the "grim reaper" is what I had in mind. I would like to help when the time comes for now I am going back on the #waitinglist it was nice talking to you guys. |
I'm also inclined to say that 1 yr is a good time for a grim reaper to reap, since people might take multiple months to come back to tilde.club. |
There's also a question of whether there's any concrete gains other than neatness-for-neatness'-sake in actually removing/disabling dormant accounts vs. just not counting them as active; they won't be cluttering up "recently active" lists, they won't be using up increasing disk space, or for that matter much space in the first place, they won't be eating up CPU or memory footprint with processes, etc. If the primary motivation for cleaning out dormant accounts is to make figurative user-count space for new active users, I think it'd be sufficient to just develop a reasonable heuristic for counting how many users are active right now and operating on the deficit between that number and the target number of active bodies, rather than being literal about e.g. counting user directories in the filesystem. |
I've started logging the output of "uptime" hourly, and a week of that should get a reasonable maxusers estimate. |
Hello! I'm jonbell on the system and I've been chatting with @joshmillard and others about this topic. Agree that a grim reaper isn't necessary, it's more about avoiding the "Peter Principle" of communities, where things get bigger until they suck. Several of us have mentioned this private but I'd like to formalize the proposal here:
For example, let's say on Friday we run a script that shows 140 people have been active in the last week, and our magic number is 150. So we add the next 10 people on the waiting list. Maybe the next week, 210 people are active. Great! No one's added from the waiting list that week. I also like the idea that this happens at the same time each week, because that can lead to a neat cultural thing. Let's say each Sunday afternoon new people show up. Wouldn't be long until someone makes a "new users this week" script, a welcoming committee forms, etc :) |
+1 to @ienjoy for a brilliant algorithm |
@ienjoy I really love your approach, it seems to be the best way to keep a steady growth without disturbing inactive accounts. |
Thanks, all! Not my idea, actually. My approach was dumb: @joshmillard (and others in various wall texts, emails, etc) was the one who came up with this proposal. And it is pretty great, isn't it? :) |
@ienjoy Your approach does make sense though, eventually if the demand for new accounts is too high it might be worth it to expend into clusters. Lets say each cluster can house 600 users that would be great and we would be interconnected to each others. I guess in the long term each clusters would have their own community with different variations of what's going on inside the box but we would all share the same goals of maintaining tilde.club alive and keeping it a great place to learn and share information. |
Data point: There are 5,000 people on the waitlist and that was yesterday
|
Sorry early evening. It was like 2000 in the morning. I stopped looking at
|
Yeah, I think there's two issues here that glance up against each other:
So talking about throttling makes sense to the part of my brain interested in avoiding all that scaling/overwhelming madness, but it doesn't help the bulk of the waitlisted. Seems like there's a few different answers to that issue: A. Eh, forget about throttling, let's just go crazy and throw the doors wide. All of those answers have different potential problems; (D) is in a sense the least problematic if we handwave away a lot of potential questions about inter-server communication/interaction/integration and if someone or a team of someones is actually up to tackling that task and can handle the nitty gritty mostly independent of what's happening at the same level on Tilde A. But if the folks in charge of Tilde B are also substantially the folks in charge of Tilde A, what we've actually done is something on the spectrum between (A) and (B) except also with the complication of multiple servers in play. It's not really a load off for the folks organizing/sysoping. Which is up to those folks, and isn't necessarily a bad thing, and may be a good way to kickstart some experimentation with a federated collection of tildes that's been talked about in abstract some since necessity and invention are reportedly an Item. Thinking out loud. It's easy for me to say the more the merrier, spin up Tilde B and let another few hundred people get jamming over there, but then I haven't been doing any of the work and don't have any concrete vision for how any of that inter-server management would work right now. |
@ftrain: @joshmillard: So I like the idea that it takes forever to get in. Not just for scaling, but it makes it really mean something. Meaning people that show up will want to make the most of it. Meaning fewer tourists and more engaged citizens. And please understand, I'm not doing this to feel superior to everyone on the outside. I genuinely think it could be the lifeblood of the community. It would be so sad to have a bunch of people show up but not actually engage. It happens all over the place. But with this waitlist, and the enthusiasm that scarcity stirs up, we may be able to dodge the curse of novelty things like this :) |
@joshmillard Yeah if you guys really feel the need to scale then I agree with the idea of "Tilde B" being managed by a new group a sysops in conjunction with the current sysops on tilde.club to maintain the same standards and of course some divergence will occur at some point and it is to be expected. Will it end up splitting the community? is it bad? who knows... @ienjoy I really understand your point and this is what scares me the most, see I am not even in tilde.club atm and I envy all of you for being part of this magical 90's throwback club. I believe the waiting list adds a meaning to this community and will promote a lot of engaging activities within the community once new bloods gets in but it might also do the opposite and kill the interest of the thousands of people on the waiting list..think about it what if you get an email in 5 months saying "Hey you just got in" then you might just don't care and give up because things move so fast nowadays. The buzz is right now, should it be Throttled to the point of letting the interest died down or add a new Tilde and let another 600 users inside the club? What could be throttled is the rate of new Tildes showing up. When a tilde is at maximum capacity, just create a new Tilde and add 600 users in one go. |
Yeah, I totally get it. I think we're seeing the same problem and trying to address it in exactly different ways :) I'm positive demand will die down, and I'm positive too many people will kill the community. So my "let the waiting list stand" suggestion super sucks for people that are enthusiastic today, on October 11th. But I think it could do a ton to meter out the enthusiasm over time, in a sane way. I'm tired of the "big spike of traffic and everyone leaves" pattern. I'd love to try a new way this time. |
Other clubs are coming online of their own volition. If those clubs want
|
So if I want to create a club on my own would you guys help out? I would need my own sysops team? I just think there's a lot of mix feelings right now, I am really up for it and I would be glad to help. I just don't want to kill the current vision. I am all good waiting for my spot to be open, novelty dying down is a huge problem and I agree that it could benefit the community by throttling the waiting list. Sent from my iPhone
|
there are a bunch of other sites with free shell accounts, like http://grex.org . |
@globz Now that you mention it... If you were to peel off and make your own system (or just go to https://bleepbloop.club who already started one) you know what would be awesome? Thinking through what it means to be on one of the other servers. Draw up a vision of how it might work. I'd love the ability to go "visit" other servers ... somehow? But if you step up and start describing how something like that might work, I'd be happy to help with the vision. I'm not great with unix, but I am a designer and I love this sort of challenge. |
@ienjoy I will definitely think about it, I really want to get involve, what I don't want is to hurt the community. I would be glad to talk about this vision with you. Did you guys sent people over to bleepbloop.club? If I create a new club I really want to be interconnected and visiting other clubs would actually be a thing. Think about it this way, a new club is only an extension of tilde.club with a different vibe, we still share the same vision and features. I will step up, just need more time to get a clear vision. Sent from my iPhone
|
I didn't ask anyone for permission to start tilde.club :). I can't see how We'll need to figure out a secure approach for those servers to Helping users on the waitlist find servers remains an open problem. Maybe If resources can be discovered with publicly available text files in ini |
I'll just weigh in to say that I'm now 100% confident that we know the way to have multiple IRC servers talk to each other (I mean, this is unsurprising, since this is a core feature of IRC); I worked it all out and documented it last night. The only thing missing is the focused way we so this -- as in, some semi-formal rules that make the IRC network further the tilde ethic and not turn into a shit sandwich. (Like, the last thing I think we want is for an IRC server to come online into the tilde network which also peers with open-access servers that AREN'T in the tilde network.) |
Okay sounds good to me, I guess I will start a new box very shortly. If this is the way you guys wanna scale I have no problem with that. @ftrain I agree that each boxes should serve a different mission but still comply with the global vision and share our findings and best practice when it comes to setting up a box. Did you send over some people to bleepbloop or it isn't ready yet? |
You are implying a level of strategic foresight that I sadly lack. |
Well do you need help? Do you need a dispatcher? Just send some work over my way and I will gladly help you guys out. |
i put you on the box. get situated and then let's talk. i would love a |
@ftrain thanks a lot Paul! I will create my account today, kinda busy right now! Lets talk for sure I will gladly help. Sent from my iPhone
|
No pressure on tilde.club
|
this kinda feels like the beginning of a decentralized network where people share interests and documents..... |
I mean just as tilde.club is just Unix, that network of tilde.clubs is just
|
there was a tongue in cheek. |
do we have any way of keeping track of what tilde.clubs are up and running? |
I started a manual list: see #46 for more info |
Awesome @pfhawkins, I was just updating the FAQ and I referenced the other tildes. Here's the pull request, please take a look because I know you're also working on faq stuff: |
I like the algorithm that @ienjoy suggested. I don’t, however, like the idea of any sort of “official” tilde servers, as that would (partially) invalidate the others and also just make tilde.club seem horrible and elitist. A lot of the beauty of this system is that it’s decentralised and unofficial. From what I can tell, @ftrain agrees. @vimes1984 and @pfhawkins: I think the ~faq account is a good way of keeping information on the site itself, but what if there were a convention that servers reserved ~meta or ~root (or a similar name) for admins, keeping inter-server data (IRC stuff, statistics on number of users and open spots, etc.) and general information for users in that account? Alternatively, this information could be kept in /usr/share or something like that. @globz: if you’re to be a dispatcher, such an account/information could be to your advantage. I’ll try to send a link to this discussion to all the admins of tilde servers (That’s the best generic name I’ve come up with yet. [EDIT: I’m now calling them tildeboxes.]) if I can find the right email addresses. EDIT: I addressed some of my concerns by creating @tildecabal. |
@ke7ofi +1 I've been thinking about the best way to formalize this. It'd be nice to write something up that explains that tilde.club isn't the "parent" server, it's just the first what we hope will be many servers. Not sure where to put that though - maybe a file underneath the ~faq directory. |
Ah ha, I have now. |
I think the original reason for this has been sufficiently addressed. Closing. |
Hi everyone,
I am still on the waiting list but here's an idea that might be useful...
How do you guys handle unused accounts?
ie; "Just log in with your secure internet shell to change this file!"
Maybe it might be useful to create a cron job which delete those accounts after x days to make more room for new people?
Is it currently capped at 615 members? I could host a new machine so we can expand into "clusters" sharing the same domain name.
I would like to help out and I can create a script for this particular matter if needed.
waitinglist
The text was updated successfully, but these errors were encountered: