Hey all, apologies for the downtime. Now that enforcement of TX HB20 has been temporarily stayed pending further circuit court action, I’m comfortable putting things back online. If Texas’ blatantly unconstitutional bullshit survives its inevitable Supreme Court challenge, then I’ll be taking down the CoG forums permanently and looking to have one of y’all host them, if you’d like.
The liability risk as a TX resident is simply too high. Even though the CoG doesn’t come anywhere even remotely close to the 50 million user requirement, there’s nothing stopping a SLAPP-happy idiot from firing off a suit because I moderated a post in a way they didn’t like, and I can’t afford to take a case through to discovery to prove we’re below the 50M threshold.
So, I suppose where I’m going with this is if anyone not in Texas wants to spin up some alternate Discourse hosting, I’ll happily tarball up the forums and send them your way.
Have migrated us over to some temporary hosting so I can free up my dedi box and recoup a bit of monthly spend. This is an AWS EC2 t3a.medium instance and it’s a bit pokey, but it ought to hold us for a few months while I figure out a more permanent hosting setup.
If anyone is getting intermittent weird-ass HTTP 500 errors, try clearing your cache. Discourse does not like what I just did to it, and if you’ve got any stale discourse javascript hanging around in cache, Weird Stuff™ will happen.
Yep, she or anyone else is welcome. There’s some cost to it—there’s managed Discourse hosting direct from the Discourse guys, and that’s probably going to net out at $100/mo. Alternately, this EC2 virtual box seems to be holding up okay for now (weird server errors notwithstanding—I think those’ll clear up over the next few days), and it costs about $380/year for a reserved instance, plus another $10-ish per month for bandwidth and storage.
I’m sure there are other options for considerably cheaper hosting on Digital Ocean, and I leave it to you guys to eventually decide what to do. This EC2 instance is paid for through November, so we’ve got some time—circuit court actions notwithstanding.
Hm, no, the weird-ass errors are persisting. Something’s not right.
I basically forklifted Discourse from one host to another one and it definitely is not happy about being disturbed. This whole thing is a really creaky Docker image at this point, and it might be worth putting in the effort to actually rebuild Discourse and then import a forum backup, which is what I did for the last host move.
If the freaky server errors keep up through tomorrow, that’s what I’ll do. I don’t think it’ll take too long.
Google OAuth2 appears to be broken with a redirect URI mismatch error. I was able to get around it by having the forum email me a login link, but the “Log in via Google” functionality wouldn’t work at all for me.
Google OAuth2 appears to be broken with a redirect URI mismatch error.
Hopefully once I figure out whatever’s wrong, that fixes itself. It’s been a loooooooong time since I screwed with the oauth stuff and i’d have to go back and read the docs again to figure out troubleshooting.
Things seem to work OK if you can log in, but coming back some period of time later after being logged out results in Weirdness™. I’m having a hard time figuring out why from checking logs—everything seems OK. I’ll give the image another rebuild and see if it helps. If not, then either today or tomorrow I’ll re-set up the forum the “correct” way (which is to create a brand new Discourse instance, then load a backup of the forum you want to restore) and see if that fixes things.
Apologies to all about the problems. The HB20 stuff prompted me to take things offline, but this re-hosting has been a while coming. I can’t swing the $190/mo that my current dedicated server is costing, so I’m trying to shift things over to ec2 hosting. I had this t3a.medium reserved instance not doing anything from some experimenting I was doing last year, and it’s prepaid until Nov, so this was a good use for it.
If the problem was the thing that I just fixed, then it was because I copied over the old server’s nginx configuration without actually looking at why I had the old server configured like it was. Because I’m proxying Discourse behind nginx and doing the TLS at the nginx level, nginx needs to append a few headers to the requests it sends along to Discourse, both to let it know that the original connection was an HTTPS connection and also to pass along the requesting visitors’ IP addresses. Failing to do this can result in (among other things) CSRF errors. On the old server, I had HAProxy doing the SSL/TLS termination and all the header business, and nginx was just the traffic cop, so I’d pulled all of that out of the old nginx config. It was all still there, just commented out, lol.
I went through several reactions after the site went offline:
One of my favorite sites just shut down due to politics. That’s a natural topic for the Politics is Stupid thread. Oh, wait…
Did I cause this to happen because politics is where I focused a lot of my posts? (After thinking about it for a while, I realized the law is so broadly-written it would have caught a lot more sites than it intended to.)
I wrote a lot of good stuff that I didn’t have personal copies of. How do I get it back? It looked like it would have been difficult to recover much.
If this ever does come back online, how would I know? Randomly check every so often? (Which is what I did when I spotted a few articles about the law being pushed back.)
Once I’m actually moved and in a real house, I’ll consider hosting the server. That’s not going to be before the end of the year, though. I’m leaving my current place in about three weeks, but I’ll be living rough until the house is done.
