the one nice thing about aws at least at work is when i run into weird unexplainable problems like this the answer is always “email support” and then I just wait response. lol.
the lambda might need permissions? I have this jury rigged route53 dns solution for one of my endpoints and it uses a lot of cloudwatch alarms + lambdas and configuring proper permissions for them was a PITA.
I have multiple topics where Discourse marks all the posts as read when they aren’t, and the number of them seems to be increasing. They’re grayed out on Latest, no unread posts indicator on Categories, and when I click them I go to the very last post even if there are dozens of previous ones I haven’t seen yet.
I’ve already tried clearing cache & cookies, logging out and back in, etc. This info is stored on the server. Maybe the tubes are clogged and somebody needs to rejigger the whatsit?
No longer can I start utube vid at any point. This is a calamity
Ok thx I’ll have a look today.
idk if you did anything, but the problem seems to have resolved.
Correction: The problem has resolved in previous threads, and manifested in new ones. I guess I should feel special that this only seem to be happening to me.
FWIW I’m having this problem in exactly one thread that I know of (the Ukraine thread). May be happening with others but I just don’t notice because they bump more infrequently.
I tried switching that thread from Tracking to Normal and then back again just in case that would jostle something, no dice.
It’s still ongoing for me too. The two topics I’m aware of are Ukraine and the 2022 midterms.
it’s the ruskis
Glad you mentioned it, just realized the midterm thread is doing it for me too. Dunno if that will help the powers that be with troubleshooting.
Wonder if it has anything to do with derails being separated out? And the software having trouble with users being tracked to both the new thread and the original? Totally spitballing there.
Do you read the forum with multiple devices?
Couple of different PC’s for the most part, rarely on mobile. Doesn’t seem like any correlation as to where I last browsed it from or anything, and all other tracked threads behave fine.
Yeah, I’ve tried troubleshooting from multiple devices. Whatever is happening is happening on the server.
There used to be a way I could select + all below and replies when I wanted to excise posts to a different thread with mod controls. I can’t seem to do that anymore on mobile. Any ideas?
Just noticed that this issue is also happening for me, and also with the main Ukraine thread and the midterms thread. Haven’t confirmed any others but it’s definitely happening with those two. Sneaky little bug.
I think I can recreate this -
I noticed a similar thing when I have multiple tabs or browsers open in the same thread. so one tab will be in a thread, and since the thread responses load dynamically, they may get registered as “read” when you view the topic’s title in another tab/browser.
If you have no other tabs or browsers open while this happens, maybe it’s some stale authenticated session lingering around - I would first try logging out all devices from the security tab under your user profile and logging back in to see if this fixes anything.
yep, just reproduced it in this way.
Interesting, I browse on both mobile and computer but am pretty good about closing tabs. Also it seems like there must be something else going on since it’s happening with the same threads for all of us, no way that’s a coincidence. Strange.
@spidercrab just wanted to follow up on a previous convo we had about the cost of UP’s cloud stuff. I tried to price it in a way that could be cheaper for you, but no matter how I did it the volume alone will run you $40-ish a month depending on how many iops you require on the disk (i have no idea what you need but I imagine during high traffic news events this would need to be fairly high). I expect you maybe don’t need data stored on an attached disk and it could be cheaper to use some other storage like S3, but I haven’t looked at anything like that.
The 8gb/4vcpu calculations I was looking at ran between $60-$100 a month. I suspect you could get it lower with clever usage of spot instances but that definitely isn’t worth your time and is almost certainly a massive chore.
I’d look into different provider’s cloud pricing though - I didn’t look at anything other than aws, which is usually cheapest.
disclaimer: I’m not a professional cloud savings guy, and I’m sorry this news isn’t great, but from an initial glance it appears that’s gonna be a decent price for you for what you have. AWS could probably come close I suspect but not as much better as I expected. In my own stuff, I was using a way smaller disk size and not as big of a machine. Apologies.
this is not specifically at you spidercrab, just whatever random technical admin that may come across this. wrt something like s3 as storage, this guide kind of gets at what my idea was:
I don’t know how much of your file system usage is due to image uploads or just DB writes (i suspect images take up most of it) using something simple like this could probably let you ditch that attached disk or whatever they call it. To know whether that’d be a benefit, you’d have to get an idea of how much your disk space is used by image uploads vs whatever else discourse is writing to disk, which I imagine is just the database (shouldn’t be that large honestly).
A huge benefit to this approach is not only is it way cheaper than these attached volumes usually (depending on how much access and speed is needed), but that s3 volumes don’t have a fixed size so you’d never need to worry ever again really about the server just running out of disk space.