Even people with money don't keep their money in money since they're not retarded.
I bet more millionaires than you think live in their overdraft.
I bet more millionaires than you think live in their overdraft.
All we wanted was a hollowed out side of a mountain shaped like a skull with lava coming out of the eyes so we could store retirement documents, and we can't even have that one thing!
I feel like if the conservatives were smart they'd be pointing out that this is the natural consequences of acting like a bunch of little mean girl bitches for the past 10 years.
Hey, Liberals and NDP -- I think there's someone in Nunavut you hate that you haven't called "Canada's Donald Trump" yet, better go off and do that.
Hey, Liberals and NDP -- I think there's someone in Nunavut you hate that you haven't called "Canada's Donald Trump" yet, better go off and do that.
[Admin mode] I was getting pretty annoyed that we weren't getting streaming updates and I wasn't seeing notifications without a refresh, so I looked into it and found something that needed to be cleaned up in the reverse proxy config to fix streaming. Looks like it's working now.
Behind the scenes, we're fully on http3 now, which I probably talked about before but I still think it's a really cool protocol -- it's done using UDP, and it takes way less time to send content because it doesn't establish a connection so there's a bunch fewer hops.
Behind the scenes, we're fully on http3 now, which I probably talked about before but I still think it's a really cool protocol -- it's done using UDP, and it takes way less time to send content because it doesn't establish a connection so there's a bunch fewer hops.
It's the same mechanism, you're basically leapfrogging the galvanic series by putting something with a better potential for ions to chip away at instead of your iron. With a sacrificial anode you're basically making a battery, but you could just as easily use a power supply and do the same thing.
Both forms of corrosion protection do behave differently depending on water type including water hardness.
Both forms of corrosion protection do behave differently depending on water type including water hardness.
You know, that gets me to thinking: Why have a sacrificial anode when they could just throw a 3V power supply in there and have it last forever?
It's always got power, it's attached to the house...
It's always got power, it's attached to the house...
The name "Kier Starmer" in England is a troll name. His parents obviously hate him and wanted to take revenge by having a name with 3 Rs in a land where people both don't say Rs or randomly add them to sentences.
One of the things I've had to do for my next book is imagine how schools could be better.
One of the core improvements we'd have to see is people have to accept that school has to be hard, and people are going to flunk out or drop out if they don't make the grade. School needs to be hard, and it needs to be structured to produce great men with skills, not people with pieces of paper saying they have skills.
I don't just mean university, either. The fact that you need a university degree for jobs that don't have anything to do with a university education is caused by wasting kids time and everyone's money at the elementary and high school level.
People focus on AI, but reality is there's a sort of fantasy where westerners aren't competing with the rest of the world. You can hire people from parts of Africa or Asia for much less than the cost of a GPU, and that's the lowest common denominator entry level employees need to compete with.
Because we've mismanaged global civilization so badly, we're already on a path to global population collapse as bad as the black death. Things will be much harder than today in a few decades as a result, and nobody seems to realize it yet. The positive part of demographic collapse is that there's a potential for great improvements to the power and material conditions of the working class. It happened in the past after the black death and the world wars, it could happen again. When there's so many fewer people there's more opportunities for who's left -- AI or not.
One of the core improvements we'd have to see is people have to accept that school has to be hard, and people are going to flunk out or drop out if they don't make the grade. School needs to be hard, and it needs to be structured to produce great men with skills, not people with pieces of paper saying they have skills.
I don't just mean university, either. The fact that you need a university degree for jobs that don't have anything to do with a university education is caused by wasting kids time and everyone's money at the elementary and high school level.
People focus on AI, but reality is there's a sort of fantasy where westerners aren't competing with the rest of the world. You can hire people from parts of Africa or Asia for much less than the cost of a GPU, and that's the lowest common denominator entry level employees need to compete with.
Because we've mismanaged global civilization so badly, we're already on a path to global population collapse as bad as the black death. Things will be much harder than today in a few decades as a result, and nobody seems to realize it yet. The positive part of demographic collapse is that there's a potential for great improvements to the power and material conditions of the working class. It happened in the past after the black death and the world wars, it could happen again. When there's so many fewer people there's more opportunities for who's left -- AI or not.
When does the "companies that don't make anything good regardless of AI" bubble pop?
I don't think anyone I like calling themselves "we" should, in a "I like you don't come to school tomorrow" sort of way.
[Admin mode] This is a log I was writing as I continued through.
We've finally at long last made it to the new server! (lol when I wrote that line I was so naive)
One thing I learned is that pg_repack will totally fill up your storage if it fails (as mine did during the time period of crashing all the time) -- hundreds of gigabytes of old tables that didn't do anything. It massively increased the time I took to transfer the database for no good reason. For anyone else running an instance, it's probably something to be aware of.
According to documentation, it can be cleaned up with:
\c pleroma
DROP EXTENSION pg_repack CASCADE ;
CREATE EXTENSION pg_repack;
In the case of my database, I got well over 100GB of drive space back immediately for no good reason. In terms of restoring the backup I made, it ended up sucking up huge amounts of time on dead databases.
So I wrote the above 7 hours ago. It turns out the restore isn't a linear process!!
It's a never-ending process.... I understand now why I failed on the previous process, I couldn't have actually completed the steps I'm waiting for -- 12 hours after I started.
It's a substantial upgrade in some ways. The SSD was SATA before, now it's nvme. The original CPU was a Intel(R) Core(TM) i5-4570TE CPU @ 2.70GHz with hyperthreading disabled. The social container I have only has 2 of the 4 cores now, but it's on a AMD Ryzen Embedded R1505G with Radeon Vega Gfx
(The rest of the day passed) Holy moly, 20ish hours in?
(Several more hours in) I ended up calling it a failure 24 hours in, and went with a new way of looking at things: Instead, I upgraded the postgresql 15 to postgresql 16 and plan to just move the binary folders over.
This seemed like a great idea for the first hour... But it turns out slow machines are slow, so it took quite a while to migrate. Still probably the right idea.
Eventually the upgrade did finish, then I was able to just tar up the postgresql 16 folder and ftp it over to the new server.
Thankfully, this time it did in fact successfully transfer. I had one problem where it seemed the user didn't get created properly so I set the password and database permissions. Next, I had a quick issue where pleroma was exposing itself to the old IP address, but that was one line change in the config. Finally, after what felt like days without FBXL Social, things were back up.
One thing not related to the technical side of things, there were a few times where I had a thought and went "Oh, that's clever I should post that on -- oh nevermind I sure hope postgresql hurries up!"
So a few points afterwards:
1. Proxmox is really nice. Other than constantly whining about not having a subscription, it's really nice.
2. We're now doing automated backups to network attached storage, which is also really nice.
3. It's all just containers, so if hardware fails, I can fire up the same container on another proxmox machine which is (you guessed it) really nice. (I was going to try for High Availaibilty, but you need
4. Containers are really light, so I'm able to have individual containers for individual services which is (find another description bro) really nice.
5. Migrating large postgresql databases is friggin slow!
6. Using straight pg_dump to create a backup of your database is actually stupid, because my backup was 200GB. Once I used -FC the size went down like 75%.
7. pg_repack helps improve size and performance of postgres databases, but if it fails half way through you end up with potentially huge databases that don't do anything! That was the final straw that stopped me from the original migration. The server took a full day on re-indexing one table (I think activity visibility) and I realized the repack tables would probably be just as long or longer.
8. I should have cleaned up my database before trying to migrate in the first place.
One thing that's really funny -- the server that ran my reverse proxy, my nextcloud, my main website, the fbxl website, and fbxl social all at once now just runs a couple small things, and now it's sitting at 0.04 load. That machine crashing (ostensibly because it couldn't turbo anymore) was the thing that began this whole ordeal, and now it's basically idle.
Next for me will be taking a lot of my now idle or removed boxes and making them into tiny proxmox nodes so I can do all kinds of neat things on the fringes from one centrally managed system. No downtime required since nothing active will go down.
Still 0 fans in my entire empire of dirt.
We've finally at long last made it to the new server! (lol when I wrote that line I was so naive)
One thing I learned is that pg_repack will totally fill up your storage if it fails (as mine did during the time period of crashing all the time) -- hundreds of gigabytes of old tables that didn't do anything. It massively increased the time I took to transfer the database for no good reason. For anyone else running an instance, it's probably something to be aware of.
According to documentation, it can be cleaned up with:
\c pleroma
DROP EXTENSION pg_repack CASCADE ;
CREATE EXTENSION pg_repack;
In the case of my database, I got well over 100GB of drive space back immediately for no good reason. In terms of restoring the backup I made, it ended up sucking up huge amounts of time on dead databases.
So I wrote the above 7 hours ago. It turns out the restore isn't a linear process!!
It's a never-ending process.... I understand now why I failed on the previous process, I couldn't have actually completed the steps I'm waiting for -- 12 hours after I started.
It's a substantial upgrade in some ways. The SSD was SATA before, now it's nvme. The original CPU was a Intel(R) Core(TM) i5-4570TE CPU @ 2.70GHz with hyperthreading disabled. The social container I have only has 2 of the 4 cores now, but it's on a AMD Ryzen Embedded R1505G with Radeon Vega Gfx
(The rest of the day passed) Holy moly, 20ish hours in?
(Several more hours in) I ended up calling it a failure 24 hours in, and went with a new way of looking at things: Instead, I upgraded the postgresql 15 to postgresql 16 and plan to just move the binary folders over.
This seemed like a great idea for the first hour... But it turns out slow machines are slow, so it took quite a while to migrate. Still probably the right idea.
Eventually the upgrade did finish, then I was able to just tar up the postgresql 16 folder and ftp it over to the new server.
Thankfully, this time it did in fact successfully transfer. I had one problem where it seemed the user didn't get created properly so I set the password and database permissions. Next, I had a quick issue where pleroma was exposing itself to the old IP address, but that was one line change in the config. Finally, after what felt like days without FBXL Social, things were back up.
One thing not related to the technical side of things, there were a few times where I had a thought and went "Oh, that's clever I should post that on -- oh nevermind I sure hope postgresql hurries up!"
So a few points afterwards:
1. Proxmox is really nice. Other than constantly whining about not having a subscription, it's really nice.
2. We're now doing automated backups to network attached storage, which is also really nice.
3. It's all just containers, so if hardware fails, I can fire up the same container on another proxmox machine which is (you guessed it) really nice. (I was going to try for High Availaibilty, but you need
4. Containers are really light, so I'm able to have individual containers for individual services which is (find another description bro) really nice.
5. Migrating large postgresql databases is friggin slow!
6. Using straight pg_dump to create a backup of your database is actually stupid, because my backup was 200GB. Once I used -FC the size went down like 75%.
7. pg_repack helps improve size and performance of postgres databases, but if it fails half way through you end up with potentially huge databases that don't do anything! That was the final straw that stopped me from the original migration. The server took a full day on re-indexing one table (I think activity visibility) and I realized the repack tables would probably be just as long or longer.
8. I should have cleaned up my database before trying to migrate in the first place.
One thing that's really funny -- the server that ran my reverse proxy, my nextcloud, my main website, the fbxl website, and fbxl social all at once now just runs a couple small things, and now it's sitting at 0.04 load. That machine crashing (ostensibly because it couldn't turbo anymore) was the thing that began this whole ordeal, and now it's basically idle.
Next for me will be taking a lot of my now idle or removed boxes and making them into tiny proxmox nodes so I can do all kinds of neat things on the fringes from one centrally managed system. No downtime required since nothing active will go down.
Still 0 fans in my entire empire of dirt.
To be fair, it has to end eventually -- either because it is ended intentionally or because the American empire drowns in debt.
I should be clearer: he seems like Harris in that he’s totally unlikable, lacks any real qualifications, and was solely installed by party apparatus rather than any sort of excitement.
Carney really is looking like Canada's Kamala Harris. Hopefully we get a similar result to America's Kamala Harris.
[admin mode] so remember when I said "should be 2 hours"... 3 days ago?
The lie detector determined that was a lie.
I'll post a postmortem later today, but bottom line is we've finally successfully migrated fbxl social to the new hardware and software platform.
Glad it's done, I don't want to do that again any time soon. Thankfully I shouldn't have to.
The lie detector determined that was a lie.
I'll post a postmortem later today, but bottom line is we've finally successfully migrated fbxl social to the new hardware and software platform.
Glad it's done, I don't want to do that again any time soon. Thankfully I shouldn't have to.