Time Nick Message 01:12 lhofhansl Not that I can see in the code. (PRAGMA optimize) 07:23 sfan5 re animateJoints: I vaguely remember this being discussed some time, but I forgot whether someone actually tested it and found it was broken or that was just a theory 09:38 [MTMatrix] also in my settings set synchronous = 0, but in the dump database the pragma doesn't exists 09:39 [MTMatrix] I just reacreated database manually with some pragmas, nice workaround 13:18 MTDiscord Merging #13779 in 10 min 13:18 ShadowBot https://github.com/minetest/minetest/issues/13779 -- Fix that negative integer values for float settings don't get a ".0" suffix by grorp 13:40 MTDiscord Done. 13:50 erle question to devs: has the 10 min thing ever resulted in something being not merged? it seems awfully low (i mean what if someone is not glued to their screen at all times) 13:50 erle or am i misunderstanding its purpose? 16:20 sfan5 cdb downloads really need any kind of progress indication, even if it's just some stuff printed to console 16:21 sfan5 I just waited 3-4 minutes for some unknown amount of data to be downloaded, unpacked size is 160 MB at least. My internet could manage more than 10 times that in the same time at least. 16:23 MTDiscord sfan5: what were you downloading, dreambuilder? 16:23 sfan5 mesecraft 16:29 sfan5 https://0x0.st/HpRf.png this might explain why I waited so long 16:29 sfan5 (it was even lower before) 16:30 rubenwardy is that mine or you end 16:31 sfan5 yours 16:34 MTDiscord uh yeah, downloading from CDB seems to be kinda slow 16:34 MTDiscord https://cdn.discordapp.com/attachments/747163566800633906/1148657367894995035/image.png 16:34 sfan5 consider doing some bandwidth limiting (QoS) for downloads if you don't already 16:36 rubenwardy nginx config: https://gist.github.com/rubenwardy/05b9eb2ed4990ff8d955b2bd2ff26fd3 16:38 rubenwardy so do I want limit_rate ? 16:38 sfan5 as an addition to limit_req, maybe 16:39 sfan5 import question first is: where's the current bottleneck? disk, network or cpu? 16:39 sfan5 important* 16:44 rubenwardy htop: https://i.rubenwardy.com/ZEYT4.png 16:47 sfan5 you can enable "detailed cpu time" to easily see when processes are spending time waiting for io (in gray here): https://0x0.st/HpRv.jpg 16:48 sfan5 judging by what you showed it's probably an io bottleneck 16:50 rubenwardy iftop: https://i.rubenwardy.com/yEzhh.png 16:53 rubenwardy netcup claims 2.5 Gbps network card https://www.netcup.eu/vserver/#root-server-details 16:58 rubenwardy a lot of the top users in ifconf are aws and linode instances 17:02 rubenwardy *iftop 17:02 rubenwardy PMed an image with more info from iftop, the top users are on aws and linode so maybe abusing it or scraping? 17:19 rubenwardy ok turns out it wasn't sorted, probably just crawlers 17:59 [MTMatrix] How specify two or more bind interfaces? enp1,enp2 is valid or no supported multibinding? 18:02 sfan5 not supported 18:05 [MTMatrix] eh 18:08 sfan5 no idea who wrote that description anyway, bind_address does not specify an interface 18:08 sfan5 it specifies an IP 18:10 [MTMatrix] hm but tun0 works fine, but one problem, httpapi also binds on this interface and it can't fetch some content from lo interface 👀 18:11 sfan5 it works for curl but not MT's networking code 18:11 erle i have a suggestion to speed up CDB if it is upstream-bandwith-limited: grab some torrent library and use cdb as a webseed. 18:11 rubenwardy no 18:11 erle ok 19:37 MTDiscord sfan5: #13550 has two approvals now. You self-requested a review on that PR. Do you still want to review it or can it be merged? 19:37 ShadowBot https://github.com/minetest/minetest/issues/13550 -- Improve UX when no game exists and drop `default_game` by rollerozxa 20:06 [MTMatrix] minetest can handle more than 1000 players? 20:08 erle localhost i'll doubt any amount of 200 it until someone shows a server like that. 20:08 erle over 200 i mean 20:08 erle you have one? 20:09 erle and even if the engine can handle that (which i do not know), i have enough experience with badly written mods that send random crap to everyone scaling with the amount of players 20:10 erle or have some globalstep that does player stuff 20:10 erle localhost so, show and tell? 20:11 [MTMatrix] server have super simple 1 block (singlenode) world, it can help improve expirence? :D 20:11 erle i don't think it's really about the world in any case 20:12 erle i mean, you can make a world that is laggy for 5 players pretty easily: fill chests with recursive shulkers or so 20:13 erle localhost, if this is a practical question, i am interested. if it is a theoretical, i am not. are you running a server and try to up the user count? 20:13 ROllerozxa I don't think even voxelmanip classic would be able to handle 1000 players even if it was hosted on a beefy dedi server 20:14 erle the only tried-and-tested way i know to make a server handle more players bandwith-wise is to shave bytes off here and there (often many times, with each single attempt looking like it would not improve much) 20:14 [MTMatrix] also, voxelmanipulator via lua scripts have native speed as native writen mapgen? 20:15 erle computation-wise, it's about not being stupid 20:15 erle like not accidentally quadratic or so 20:16 erle luk3yx you made this python library for minetest clients, right? did you run any load tests with it? 20:17 [MTMatrix] wut!? python library for minetest, whoa 20:19 [MTMatrix] In this case, minetest more CPU or RAM like? 20:21 erle what do you mean 20:21 erle minetest can fill your RAM, regardless of how much you have, if you do something stupid with it 20:22 [MTMatrix] of cource, i think, minetest eat less resources than minecraft java stuff 20:22 erle an example: cora had an implementation of fire spread that worked eXaCtLy LiKe MiNeCrAfT. if you ask the engine to run a counter per node for the age of the fire and the fire node copies itself, filling the world with more counters … eventually you run out of RAM, regardless of how much you have. 20:23 erle cora fixed it by making the fire die probabilistically 20:23 erle like, each tick the fire had a chance to die – this does not require a counter (and thus a bit of RAM) for every fire node 20:24 erle if you have a minetest engine game you *probably* want to find all these little stupid things before even thinking about the engine stuff 20:31 [MTMatrix] or cource, stupid things like create tnt world is huge pain.... 20:33 luk3yx erle: I don't think I wrote a client library? I made something once to try and detect servers lying about players on the server list but it was very simple (no reliable packet handling) and only tries to log in rather than joining 20:34 luk3yx I'm sure someone made a Go library at some point 20:34 luk3yx https://github.com/minetest-go/minetest_client 20:34 erle luk3yx then i confused you with someone 20:35 erle the go library is from anon5 i think? 20:36 luk3yx I think BuckarooBanzai wrote the one I linked 20:38 [MTMatrix] oh it mean bot client for minetest already exists? 👀 >_< 20:38 [MTMatrix] bots everywhere 20:38 [MTMatrix] eh 20:41 erle localhost how about you use that and test and then report your findings 20:42 [MTMatrix] but still thanks! 20:43 [MTMatrix] and this is true, mine **test** 20:43 [MTMatrix] :D 21:48 sfan5 @grorp: I will take a look at it tomorrow