And a very important section, that does not surprise me at all:

The low success rate of applications has been put down, in part, to an increasingly number of speculative applications being submitted. Industry reports show a rise in so called “phantom projects” in these cases, developers submit multiple applications for many sites, with the expectation being that very few will connect. These speculative and duplicate applications have seen the connections queue grow, increasing the work needed to progress projects.

It becomes a sad self-fulfilling prophecy. Applications take a long time to process, so companies fling lots in parallel, then only use the first to get through.
Which means that applications take even longer to get through.

This is now complete. Things seem a little snappier…Though that may be because everyone is disconnected! Various bits of generic software have also been patched to latest.

Mike Ashley didn’t actually want a game shop, he wanted people spending £70 on each football/COD release, then buying that season’s team strip on the way out of the shop.

I will hold my hands up: I went from “we should do this soon” to “oh, that’s not too horrible” to “ok, let’s do it” in the space of about 30 minutes.

Really, I should have stickies the post for an hour or so!

Emperor is talking about some cool custom error pages, which much improve things going forward.

The newer versions are slower, though I don’t know exactly how much by.
I could definitely feel it when going between 19.4 and 19.5 mid upgrade (unless the quick switch caused the slowness!)

We’re now federating more (as the newer version supports more things between instances)

We’re also caching and resizing images too, so the server load is roughly 3 times higher.
(This may not continue).

We may do a restart once we’re confident that things are stable. As it’s possible that a little weirdness crept in!

Thanks for the post, it persuaded me to get off my bottom and add another one to the list.

I have a suspicion that it’s the classic IT issue “it was configured, but never actually did anything previously”.

This has now been completed :)

Other than me missing that Lemmy does not like jumping multiple versions in one go (which does make sense), and a few issues with files not deleting, we are now running 19.5.

You can now see a list of images you have uploaded in your profile (and delete them if required).

Any bugs/issues, post it in this thread.

I’m not sure if there is a way to quickly find this. The best bet is probably trying the bigger instances, and seeing if it’s accessible through them.

You’ve made me think now: it might be a nice project for a some instance admins to flag when an instance they had received posts from goes offline. (I should probably check what I have from feddit.de!)

For components and wires that are made to a spec, I feel far more comfortable buying from CPC or Mouser.
Amazon sellers just feel like a coin flip if the guy is going to ship you CCA 24 AWG instead of OFC 23, in the hope you don’t notice or bother complaining.

Yes. It’s a legitimate inserted banner that goes on every inbound. It just blew my mind a bit that the correct action was to click a link in an email!

This is the thing, the balance of anonymity and preventing people using that anonymity to be a tit.
In my opinion, one of the answers is keeping the signal-to-noise high: Make sure that there are enough sensible people in a community that if someone starts acting up, they’re alone. And then they can either correct their course, or get banned, ideally before the next moron shows up.

And part of the way of achieving that is raising the barrier to sign-up, if only a little, and rate limiting.

Just for context, the full database of feddit.uk compresses down to about 4GB. I am not sure what’s going to happen to the ghosts long term, but I don’t think storage will be a huge issue.

And the shoe will probably drop at some point. Something like “communities must have nitro to access posts from more than 6 months ago”.

Or that 50% of the users on the discord only went there to find one thing, and probably won’t ever interact again.
So it looks like a bigger community, while losing accessibility.

Why Discord took off as a medium to replace forums is beyond me

My theory is that it was used as the primary form of informal communication by groups doing something, then it felt like a community.
And since everyone was there…Why not put the documentation there? Sure, it’s not indexable, but the group is open-sign-up, right? Right?

Then a few years down the line, someone suggests switching to another primary storage location…Then faces huge amounts of push-back from people comfy sitting on discord.

Info is much appreciated!

Normally, if we had a report like this, we’d just shuffle forwards after running it in the test environment.
Unfortunately, as there is a DB engine change, I’m expecting all hell to break loose as soon as we start, so it’ll be a few days before I can put the time into it.

I can absolutely see that happening in vsphere.

I wasn’t aware there were big issues, thanks for letting me know.

The pictrs upgrade was expected to take a lot longer, so the plan was to put 19.4.1 on when released.
However, if there are issues at the moment, we can jump a little sooner. I just fancied avoiding some of the 19.4 bugs!

Friday, possibly? I don’t like to do upgrades unless I can clear several hours in case of shit hitting fan midway through!

Give it a few years, and that’ll be the price for a 4 pack of sausage rolls.