Time Nick Message 01:01 katp32 it's me again... is it possible to get minetest to bind to multiple addresses, rather than just one or all. specifically, I have an ipv4 address and an ipv6 address, on the same interface.. 01:05 MTDiscord From what I've heard so far it seems as though no, it's not, or if it's possible by some built-in mechanism, it's not obvious how or is difficult to make it work. 01:05 MTDiscord But it's possible to use iptables/firewalls to create a redirect for it. 01:05 katp32 argh :/ 01:06 katp32 alright, *also*, does the server listing request need to come from the same IP as the actual server 01:06 MTDiscord One big limitation is that whatever address family you use to bind is the address family that is used to contact the serverlist, so if you want a publicly accessible server that's only reachable via an overlay network or VPN that runs on v6, but you can only get outbound to the serverlist on v4, then you're forced to bind to v4 and firewall redirect, as far as I can tell. 01:06 katp32 ie is it okay for the server to connect to the serverlist via IP A while clients connect to the server through IP B 01:07 MTDiscord It doesn't need to reach the serverlist via the same IP (NAT and VPNs are fine) but it will try to use the same address family. 01:07 MTDiscord You can actually advertise whatever "address" you want, and that's actually probably advisable, i.e. use a DNS name instead of a static IP address 01:07 MTDiscord even if your IP address is actually static for now, using a DNS name would probably make it easier to migrate later if the need/opportunity arises. 01:08 katp32 yeah, already using a DNS name in fact, just wanted to double check 01:08 katp32 (although for a different reason involving IPv4/IPv6 shenanigans) 01:09 katp32 anyways, that's a pretty major limitation.. 01:11 MTDiscord It's annoying for sure. I've got a few people who want to host servers behind a restrictive NAT that doesn't allow them to port-forward, so I've had to teach myself iptables and wireguard to open ports for them on a VPS. 01:11 MTDiscord Seems ironic you'd have to jump through nearly the same hoops just because of IPv4 vs v6 shenans, but I guess that's the way it is for right now. 01:12 katp32 hm? 01:16 katp32 actually, it's IPv4 that's more annoying in this case; essentially, all my minetest traffic is being routed through my VPS, which then routes it through a VPN to my actual server since the VPS is primarily just for http, email, irc, etc; the VPS has an IPv6 /64 block routed to it, so I can actually just assign every device on the VPN its own globally routable IPv6 address, and so, no NAT is 01:16 katp32 required (as is the point of IPv6). so, my actual minetest server machine does have its own unique IPv6 address and routing just happens automagically. of course, the remaining problem is that the VPS still only has one IPv4 address, so I still need NAT for IPv4. thus, I've got a somewhat odd setup in that the VPS and the minetest server both have the same IPv4 address, but different IPv6 01:16 katp32 addresses. that means I can't just use my main domain name for the minetest server, since it would have the correct IPv4 and the wrong IPv6, so the minetest server has its own subdomain with the IP addresses set correctly 01:18 katp32 anyways, the issue in this case is that as mentioned, minetest traffic is all being routed through the VPN, however all of the other traffic on that machine is still being routed through a regular connection (I don't want to strain the VPS with even more traffic). that means that the request to the serverlist would come from a different IP. the other issue is that the regular connection on that 01:18 katp32 machine is ipv4-only, which is the actual problem now.. 01:19 katp32 I *can* just create a route that routes all ipv6 traffic through the VPN, but since applications default to ipv6 (and rightly so), that effectively means *all* traffic is going to be going through the VPN, which I don't want. so now I have the problem of "how do I route traffic to the serverlist through one interface while keeping the routing for everything else the way it is".. 01:19 katp32 that's beyond minetest tech support though :P 01:27 MTDiscord I think, though I am not certain, that the MT server list should be reachable by both IPv4 and IPv6. I think it's just a matter of making sure that whatever address family you bind on has an unbroken route of some kind to the internet for that address family. If you can reach the internet then you should be able to reach the serverlist, and if you can reach the serverlist you should be able to announce and get your server into the list. 01:27 MTDiscord The address of the server can be, I think, more or less anything you want, and the serverlist will verify it independently and neither rely solely on your announcement's content, nor rely on the reverse path the announcement took to get there. 04:03 erlehmann to recap: i have tested it and PR #10570 https://github.com/minetest/minetest/pull/10570 should be REJECTED in its current state. minetest on that branch has NOTICEABLE LAG in hotbar inventory item rendering on my already slow computer, where it did not have that with 5.5-dev. the PR description claims it would speedup item rendering – if it does not do that on a slow computer, the whole thing is wrong. 04:03 ShadowBot https://github.com/minetest/minetest/issues/10570 -- Item render-to-texture: Texture modifier, speedup by appgurueu 04:03 erlehmann (for ppl here because #minetest-dev has no logs or does it?) 04:03 erlehmann maybe someone else here can test it 04:43 MTDiscord minetest-devs should have logs in the same way this does 04:44 MTDiscord https://irc.minetest.net/minetest-dev/ 04:48 erlehmann ah it is not in the title in #minetest-dev 04:49 erlehmann sorry 06:10 katp32 yes, it's me, again... my server will not start, complaining "Game specified in default_game is invalid" 06:11 katp32 game has been installed to ~/.minetest/games/ 06:11 katp32 getting rather frustrated at this point.. 06:17 erlehmann katp32 i can try to help 06:17 erlehmann katp32 did you compile minetest yourself? 06:17 katp32 it's gentoo, so sort of yes 06:25 erlehmann katp32, did you use -DRUN_IN_PLACE=1 ? 06:25 katp32 no 06:26 katp32 also, is default_game supposed to be set to the name of the game on disk or the name of the game in its game.conf 06:27 erlehmann katp32, minetest_game 06:27 erlehmann on disk 06:27 erlehmann i think? 06:27 erlehmann no idea 06:27 katp32 it's not minetest_game, it's mineclone2 06:27 katp32 I would assume it's the on disk name yeah; I've tried both though, and neither seems to work, so 06:38 katp32 ahhh, nevermind, my bad :P 06:38 katp32 was running as the wrong user 06:41 katp32 ...and now I'm reminded that minetest doesn't support copy paste so I have to log in as root just to enter my password ;-; 13:40 rubenwardy There are currently 1234 packages on ContentDB 13:55 MTDiscord congrats on achieving "test data" status :-D 14:02 appguru yeah, I noticed that too when running my script 14:02 appguru Sam has 1234 particles if rendered with my ghosts mod BTW 14:02 appguru 1234: The answer to life, the universe, and Minetest 14:03 appguru "Scanning package joe7575/techage_modpack: 42 of 1234" 14:03 sfan5 your script would not need to write anything to disk if it was in python 14:03 sfan5 would also be easier to parallelize 14:04 sfan5 (you can open in-memory ZIP files in python no problem, is what I'm saying) 14:07 MTDiscord yeah 14:07 MTDiscord but I much like Lua 14:07 MTDiscord I don't think there is much to be gained from parallelization 14:09 sfan5 less time spent waiting for the network? 14:11 MTDiscord I'm pretty sure the network is just a single instance 14:12 MTDiscord ContentDB is just a single instance* 14:12 sfan5 you can still minimize overhead 14:13 MTDiscord well yes 14:13 MTDiscord But I don't think I can increase download speeds by downloading multiple packages at once 14:13 MTDiscord And reading the file list of the Zip is certainly not the bottleneck 14:29 sfan5 also if you're only reading the file list you could get away with only downloading a small part of the zip 14:29 MTDiscord could you modify it to run server-side and let rubenwardy run it on the server instead? if that would be alright with him 14:29 sfan5 ^ that would of course be much better 14:29 MTDiscord would prevent the bandwidth-killing scraping 14:30 rubenwardy yeah, that would be nicer for my server 14:30 MTDiscord sfan5: no I can't, the file list comes at the end 14:30 sfan5 yes you can 14:31 sfan5 HTTP supports range headers 14:31 MTDiscord Can I tell "I want the last n bytes"? 14:31 sfan5 don't think so 14:31 MTDiscord well but I need that 14:31 sfan5 but you can check the file size first 14:31 MTDiscord how? 14:31 sfan5 HEAD request? 14:31 MTDiscord HTTP header? 14:32 rubenwardy hmmm, `node_memory_MemAvailable_bytes` shows 8GB still free 14:32 sfan5 I wasn't saying it's easy but it's technically possible 14:33 rubenwardy node_memory_MemFree_bytes shows almost 0MB 14:33 rubenwardy ContentDB has had a memory leak bug that I haven't been able to reproduce locally, perhaps it's in serving the files 14:34 appguru Possibly upstream? 14:34 rubenwardy upstream? 14:34 MTDiscord I'd assume he means upstream web server software 14:35 MTDiscord like, nginx or whatever 14:35 sfan5 do you serve the files directly from flask? 14:35 rubenwardy I do ever since I moved to docker, as nginx can't read the docker volume 14:35 rubenwardy I have an issue to set up a matching www-data group 14:36 rubenwardy and then reenable the nginx bypass for uploads 14:36 rubenwardy nginx already serves static files directly 14:36 rubenwardy uploads uses send_from_directory, perhaps that caches 14:36 MTDiscord rubenwardy, what do you run docker on top of, i.e. distro-wise? 14:36 rubenwardy Debian 14:36 rubenwardy is the host 14:36 MTDiscord Okay, I run docker on debian for a ton of my MT stuff and I keep having this annoying problem that causes shit to hang when I try to reboot. 14:37 rubenwardy the guest image is based on python:3.6 14:37 MTDiscord I was wondering if you ran into that too 14:37 MTDiscord Presumably you have to reboot every so often because the HOST OS also needs updates from time to time, even if it's very package-baren. 14:38 rubenwardy I also use gunicorn which may be interferring 14:38 rubenwardy updating the groups for /data/upload/ to include www-data, and then serving directly from nginx will help if that's the problem 14:53 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Dynamic_Add_Media v2 (#11550) 13bbfae0c https://git.io/JuEUu (152021-09-09T14:51:35Z) 21:56 katp32 hiii, it's me, again; I am unable to reliably get more than ~20fps in minetest. this is not a low end system. ??