<continued from #general>
Sven Mason [15:52] I have not used Digital Ocean before (where the forum is currently hosted), it seems like there is more admin involved than eg AWS, but it is cheaper. Maybe AWS would take less admin resources than Digital Ocean once it is setup? Also AWS being the larger perhaps there is a higher chance supporters will have experience working with it? (this is if we chose a self-hosted environment). We could stay with Digital Ocean if Max wanted and try to fix what is going wrong there, to do that we could discuss admin access protocol with @max?
Max Kaye [16:51]
Whoops - posted reply in #general
@sventhebarbarian - Auto updates actually stopped working - it's manual updates that are breaking
Sven Mason [17:36] @max agreed on DO and AWS, from what I read it seems DO is a little less managed than AWS (eg setting up databases) - but basically the same, both fall under the self-hosted umbrella. I naturally like to manage everything myself (I think Max you are the same) especially if it cuts costs, but I am learning as I have a lot going on it is perhaps not the best path. I guess the first step is to decide on what we are trying to achieve, if it is the development of projects I think something like Github works well.
Max Kaye [17:36] Discourse manages its own databases so not sure we can use Amazon's RDBS and that sort of thing I don't actually like to manage things much anymore - just too time consuming - but I dislike a lot of the control you lose with most managed services I know my way around servers though so tend to jump to self-management without thinking through it sometimes (because it seems easier at the time)
Sven Mason [17:37] But if we are talking about replacing the community forum (if that is our goal), I think we need to setup something like Discourse to be more user friendly for non-tech people. Same with the wiki I agree with your sentiments @max, that is why I was leaning towards self-hosting, but organizing a group of supporters we could lean on to manage things, this would mean we would need to open up admin control, but maybe we can do that and mitigate the dangers of reduced server security?
Max Kaye [17:40] Agreed (re replacing / fixing / upgrading forum) Server security isn't too hard - just a small number of trusted ppl
Sven Mason [17:41] Currently does the DO server only have the forum? Is there any other sensitive data on there? @max If it only has the forum would you be comfortable letting me (or another trusted dev) go in there and dig around? I am not too skilled on things like rwsy/redis so would not promise any quick results but if you pointed me to what conclusions you had come to so far I could tinker, or if any other devs chime in I could talk with you about allowing them in or assisting me isolate the issue
Max Kaye [17:45] Just the forum Yeah sure the rwsy/redis thing refers to a repository in the ubuntu PPA, the package is redis I'm not sure there's much we can do atm - the main area I'd explore next is upgrading the docker image redis uses from ubuntu 14.04 to ubuntu 16.04 while the server was running 14.04 I've upgraded it to 16.04 now
Sven Mason [17:47] is redis hooked into Discourse? is it doing some caching for the forum?
Max Kaye [17:47]
essentially the place the docker image wants to download the redis package from is gone
Sven Mason [17:48] I am I correct that the Discourse has its own image which is 16.04, but redis has another image at 14.04? old redis repository 14.04 was pointing to no longer exists?
Max Kaye [17:50] no - discourse is using 14.04 they're currently pulling redis from a ppa that has disappeared - so not actually related to 14.04/16.04 besides that a newer image probably pulls from a different repo
Sven Mason [17:50] redis and discourse tied tightly together, and in the same image?
Max Kaye [17:50] yup it's all one big integrated docker thing
Sven Mason [17:51] hmmm, wonder if the easiest track would be a new image, and export/import data or might be able to update the ppa, fear is it might create some sort of hell, dependency or other
Max Kaye [17:54]
can't update the PPA because its owned by someone
I have just brought the forums back up btw
just going to try the automated update again (since it's using a new docker thing it might be okay)
the issue is discourse is out of date, and we were getting the weird 500 errors (still not sure why)
anyway usually the first thing they tell you to do is
./launcher rebuild app which sets up a fresh instance and migrates things
Max Kaye [18:27] added this Plain Text snippet: discourse upgrade fail root@community:/var/discourse# ./launcher rebuild app Ensuring launcher is up to date Fetching origin Launcher is up-to-date
Sven Mason [18:38] https://askubuntu.com/questions/592998/redis-server-what-are-the-differences-between-rwky-redis-and-chris-lea-redis-s askubuntu.com redis-server - What are the differences between rwky/redis and chris-lea/redis-server? What are the differences between https://launchpad.net/~chris-lea/+archive/ubuntu/redis-server and https://launchpad.net/~rwky/+archive/ubuntu/redis? looks like rwky no longer exists - this is I guess what you figured and I assume its hardwired into the launcher (?) I need to read up more ..
Max Kaye [18:40] yeah - dw too much about that - trying to fix now think I mgiht have something that will work
Leo Treasure [19:54] joined #dev.
Sven Mason [20:23] @max Do you feel its worthwhile organizing an admin protocol now? Or a few trusted devs that can deal with problems like this? I don't think we need to wait for the SteerCo to work this out and I don't think messing with servers and generic software maintenance is using your time effectively - at the same time I think we should have a process to tackle these things efficiently. For example I guess we can setup user accounts on the DO server so all activity is logged, you could create an admin account for myself or any other trusted dev, we could log any activity and perhaps have an offline/backup procedure to protect against any rogue behavior. Anything that might add cost or is large scale needs to run past you (or SteerCo) first but general maintenance could be handled without you needing to commit yourself.
Andrew Goodman [11:26] joined #dev along with 2 others.
Quinton Reeves [17:40] @sventhebarbarian I'm not against self-hosting, just after ten years of it being an issue over-and-over-and-over again I gave up. It doesn't help that in my case, in an open source setting, income is pretty much zero so I have to improvise with things that cost nothing, such as GitHub. We pretty much use it for everything we can because it is reliable and is free for public repos. Plus it always fell to me to fix things when they broke because whoever set them up originally is long gone. Part of the problem with the long game. I'll admit I had some push back from my community over moving discussions to GitHub, which is why I started working on the JS api to make it friendlier and look like part of our website. Because the other issue over the years is that I could never get the various pieces of software themed to look like the rest of our content, it breaks the feeling of continuity imo. @max With regard to the docs, what I've done recently is deprecate our wiki in favour of a repository with markdown files. I then use a continuous integration build to convert that into pages on our Jekyll site repo. Because it is a repository, it's easy to give specific permissions to have a wider range of community contributors than you would, say, a source repo. I'm pretty new around here, so forgive me while I figure out how everything works. I'm just offering a foss developer point of view at the moment.
Max Kaye [18:05] @Quin - just had a look at redeclipse.net - I really like the github integration for discussions.
Also, I feel like emulating a wiki via a github repo is actually quite nice, and helps keep things pretty manageable (e.g. not open editing but still easy to contribute) Also - no worries about the new thing - if there's one thing we need more of it's foss thinking around this sort of stuff. I haven't done much leading communities like this before so there's been a lot of trial and error. I'm also quick to jump to a "build stuff" mindset when approaching these problems I like the idea of Wiki repo -> CD (Jekyll) -> Static Site, and fits with our existing site (which is jekyll too - github.com/voteflux/vote-flux-v2) Also discussion via JS forum hosted on github is probably pretty good - we can filter via tags and it means even if something breaks on our frontend the github backend is still there Also github has reasonable support for quoting and things
Quinton Reeves [18:14] It's all a work in progress, part of the complete overhaul we're doing for the v2.0 release which upgrades our underlying rendering engine. Trying to make things look more cohesive than a bunch of Franken things duct taped together lol. And yeah, GitHub integrates nicely across itself. There's apps for things like in issue polls etc as well. I always felt the issue tracker and discussion forum separation to be a duplication of effort.
Quinton Reeves [18:22]
Being a developer myself I can relate to the tendency to build your own things. Open source helps a lot in that regard because it lets the people who are good at those things specialise in them (by making libraries or engines etc) while you focus on what you're good at. This is why community building is important there.
The guy that builds the game with me is a programming genius, but doesn't have the creativity or people skills to make a successful game alone. I'm pretty much his opposite. (edited)
Probably the difference between frontend and backend development I guess
Max Kaye [19:31] @Quin - One thing I just noticed is that logging in to your forum requires granting access to read and write to all my repos - do you know if that's something you can turn off? (i.e. it isn't okay for me to log in atm with those permissions, and we probably would need to remove that to put Flux forums on github)
Sven Mason [19:32] I guess I look at this from a results and effective use of resources viewpoint: What are we trying to achieve with a wiki and a forum, who is the target market, what results are we wanting and are the costs worth those results (plus possibly how do we judge effectiveness) I am thinking a wiki will be more used by people who want to dig deeper into the philosophy and design of Flux, these type of people are probably already hooked so will probably use whatever tool we offer (eg github as an alternative to a wiki The forum I am not sure who our target is, but to be effective we should probably define it. Is it existing members wanting to nut out concepts, or is it casual supporters who might respond to a friendly location they can ask questions about the movement. I would approach each of these situations differently. (edited)
Sven Mason [19:41] If existing members then maybe a stand alone forum is not even necessary, something like Slack or Github or Groups could do the job. If casual supporters then it is really important to have a clean functioning landing site so their first impressions are not of something failing or substandard. If unable to effectively maintain a forum perhaps it might serve us better to not have one, we could work out other ways to take care of casual supporters (eg the Flux website, putting more energy into static explanatory pages, perhaps an email address or some other communication system if they have questions. (edited)
Sven Mason [19:54] I would be supportive of a forum is a warm soft way for newly initiated people to start reading and asking about Flux. My focus is how can we gain the most support from a wider audience, my fear (perhaps unjustified) is if we only have tech minded landing sites non-tech minded people will feel uncomfortable to join and not become supporters/voters.