HPC, a computer which hosts part of the latest version of Ponibooru Restored, has a program uptime of 33 days and 10 hours. As it's been up for more than a month, if it was an Invidious instance in a domain name, it could be listed at redirect.invidious.io (likely with a ~low health indicator) from what I remember. The other part of Ponibooru Restored is hosted in a different computer: LPC. Best ipfs program uptime in LPC: 60 days, but since that's lower propagation computer, it doesn't really count. Month ago, in LPC with >60 days up, the data was on an HDD, so slowness from mechanical storage drive + lower propagation. About the quickest I can do is ramdisk in HPC, or SSD in HPC, but I maybe won't pay for that (so HDD in HPC in the future, I hope). If I stick 3GB into HPC (higher propagation computer), that data being only Ponibooru Restored webpages, then I can host it all in that one computer with 800 MB of free RAM. HPC, in regards to the current daemon PID: ipfs-total-in=124 GB and ipfs-total-out=140 GB; LPC, regarding the same thing: TotalIn=18 GB and TotalOut=4.1 GB. HPC PID = old-3-08:34:45 + current-30-02:45:40; LPC PID = old-19-10:51:10 + current-9-03:35:34.
Other idea: use JS to bit-identically regenerate those ponibooru.org webpages. So it would load the data common among the webpages all of the time plus the unique data per-ID. However, that's maybe slower than static pages in this case, and it would mean making new data/files instead of just working with the already-existing "old" files from the 2010s. Latest version of Ponibooru Restored = all static data, but it is reliant upon a centralize server which I don't control for the image files. However, the image files should also all be in BitTorrent, and I do control those torrents to a degree.
Could TPA host it? I'm thinking no, for https://theponyarchive.com/ would have to put all of the HTMLs into web root for it to work. Or, theponyarchive.com could host the server-side version, but that uses crap like PHP, SQL, and whatever else gross server-side software. Maybe I'll ask users in the TPA Discord (gross) to seed it. The HTML files = all 100% done, so the only changes go toward the JS in the folders which is used to change the webpages. Therefore, all of the HTMLs would be the same from update to update, so they would all have the same IPFS CIDs. Only changes: the small amount of non-HTML files in the folder(s). Pin updates via "ipfs pin update --unpin=false <older-CID> <newer-CID>".
HTML CIDs would change if I set it to raw blocks or if I modify the HTML files themselves. More likely to do the former (--raw-leaves or something) than the latter. The default thing results in about 3 GB on disk, maybe with raw leaves it could be a significantly smaller size. Other thing (maybe): physically rearrange stuff with LPC so it's not so slow. (Other thing I could do: there's ssh server and ssh client, and maybe I don't need both in a computer if I only ssh into it and never use it to ssh into other computers; or that doesn't matter, so just use key-based and not password-based auth.) Screenshot of https://bafybeihvcxfqtlydt6le4zs2r7pyau6bc6vfi5l6xl2gkpfv5civh52cgu.ipfs2.eth.limo/23005.html
