Time |
Nick |
Message |
00:00 |
|
diceLibrarian2 joined #minetest |
00:06 |
shaft |
Seems mesecons wire doesn't have any param2 data. It could have been implemented on top of mesecons like I said |
00:25 |
shaft |
Oh yes. Erle wrote a review below VoxeLibre complaining about past breakage in VoxeLibre |
00:25 |
shaft |
VL gets a good review from me and Mineclonia a bad one. |
00:39 |
|
shaft joined #minetest |
00:43 |
shaft |
Despite Erle favoring Mineclonia his mod https://content.luanti.org/packages/erlehmann/mcl_lever_status_indicator/ is definitely broken in Mineclonia now. Don't remember his irc alias, so i can't use the tell command. Someone should tell him. Maybe I'll do it when i see him |
00:56 |
|
diceLibrarian2 joined #minetest |
01:53 |
|
mazes_83 joined #minetest |
02:05 |
|
est joined #minetest |
02:07 |
|
mazes_83 joined #minetest |
02:11 |
pgimeno |
might take a while... [0126 23:35:21] * erle has quit (Quit: K-lined) |
02:56 |
|
Trifton joined #minetest |
03:01 |
|
SFENCE_arch joined #minetest |
03:19 |
cheapie |
pgimeno: "(Quit: K-lined)" is a fake kline, the real ones just say "(K-Lined)" |
03:39 |
|
mazes_83 joined #minetest |
03:48 |
|
mazes_83 joined #minetest |
03:49 |
|
SFENCE joined #minetest |
04:00 |
|
Thomas-S joined #minetest |
04:00 |
|
Thomas-S joined #minetest |
04:01 |
|
chilledfrogs joined #minetest |
04:05 |
|
liceDibrarian joined #minetest |
04:10 |
|
diceLibrarian2 joined #minetest |
04:26 |
|
SFENCE joined #minetest |
04:26 |
|
bdju joined #minetest |
04:49 |
|
liceDibrarian joined #minetest |
05:00 |
|
MTDiscord joined #minetest |
05:26 |
|
diemartin joined #minetest |
06:29 |
|
sparky4 joined #minetest |
06:54 |
|
sparky4 joined #minetest |
07:52 |
|
SFENCE_arch joined #minetest |
08:29 |
|
Guest45 joined #minetest |
08:58 |
|
tarsovbak joined #minetest |
09:01 |
|
gregon joined #minetest |
09:10 |
|
PoochInquisitor joined #minetest |
09:14 |
|
mazes_83 joined #minetest |
09:28 |
|
mazes_83 joined #minetest |
10:09 |
|
ireallyhateirc joined #minetest |
10:14 |
|
mazes_83 joined #minetest |
10:28 |
|
mrkubax10 joined #minetest |
10:29 |
|
mrkubax10 joined #minetest |
10:45 |
|
mazes_83 joined #minetest |
11:12 |
|
Desour joined #minetest |
11:50 |
|
mazes_83 joined #minetest |
12:31 |
|
PoochInquisitor joined #minetest |
13:34 |
pgimeno |
ah right, guess their fame made it believable |
14:12 |
|
ll0lll0 joined #minetest |
14:35 |
PoochInquisitor |
Hey peeps, remember: don't feed the trolls! |
14:53 |
|
SFENCE_arch joined #minetest |
15:15 |
|
fling joined #minetest |
15:20 |
|
Leopold joined #minetest |
15:24 |
|
jaca122 joined #minetest |
15:59 |
Zamolxis |
so I've built luantiserver and I'm having trouble locating the server in the server list.. i did set server_announce to true in both minetest.conf and world.mt, do I need to pass the --config to the world.mt or minetest.conf? I'm confused, I know I've did this in the past on a VPS and it just worked but can't recall how due to using containers back then |
15:59 |
Zamolxis |
I can connect to it perfectly using the subdomain |
16:03 |
celeron55 |
sounds like you're doing it somehow absolutely blind! have some documentation: https://dev.luanti.org/setting-up-a-server/ |
16:12 |
sfan5 |
you should see a line like "Announcing start to ..." in the output |
16:30 |
|
gregon joined #minetest |
16:34 |
|
tarsovbak joined #minetest |
16:51 |
Zamolxis |
Hi celeron55 thanks for the link, i followed that and this is why my question remained, is minetest.conf the one which should contain the data or world.mt int he world file? |
16:51 |
sfan5 |
minetest.conf is the config file where you need to put it |
16:51 |
sfan5 |
it's probably not reading the config file you think it is |
16:51 |
Zamolxis |
Looking through the journalctld logs after I left it running for some time I get: HTTPFetch for servers.luanti.org/announce returned response code 409 |
16:52 |
Zamolxis |
Request has been filed, but the previous request encountered the following error |
16:52 |
Zamolxis |
server did not respond to ping |
16:53 |
Zamolxis |
thanks sfan5 |
16:53 |
Zamolxis |
Weird, seems this provider is blocking ping requests |
16:54 |
MTDiscord |
<warr1024> Like 90% of the time, failing to make it onto the server list seems to be caused by the server list server not being able to ping your luanti server at the address+port that you told it. It's very often a firewall/forwarding sort of thing, especially since this is one of those "you have to know your own system" kind of things. |
16:55 |
sfan5 |
it should say "Server ... port ... did not respond to ping". does that match what you're expecting? |
16:56 |
Zamolxis |
Yes it says: ERROR[CurlFetch]: Server game.zendenpen.com port 30000 did not respond to ping (tried 2a06:1fc0:0:1::43d) - I just cut out those bits |
16:56 |
Zamolxis |
SO it's using ICMP ping? or just connecting to the ip& port? |
16:56 |
Warr1024 |
!up game.zendenpen.com 30000 |
16:56 |
MinetestBot |
game.zendenpen.com:30000 seems to be down (IPv6) |
16:56 |
Warr1024 |
it's a Luanti protocol ping |
16:56 |
Zamolxis |
because from what I'm seeing I suppose my provider is blocking ICMP |
16:57 |
Warr1024 |
ICMP is not involved, you only need to open/forward that one single UDP port |
16:57 |
sfan5 |
since you have ipv6 on your domain you will need to make sure it's 1) allowed through the firewall and 2) enabled for luanti |
16:57 |
Warr1024 |
though, if you're multi-homed (Especially v4 + v6) you need to make sure the port is open on whatever address families the clients will try to use |
16:58 |
Warr1024 |
IIRC Luanti can't listen on both v4 and v6 at the same time (OS dependent maybe?) so you might have to have a firewall handle forwarding from one to the other if you want to be able to use both... |
16:58 |
Zamolxis |
hmm let me recheck the ipv6 ip+port manually, cause when i go to the dns it works, and dig works |
16:58 |
Zamolxis |
Ah |
16:58 |
sfan5 |
Warr1024: it can. just not enabled by default |
16:58 |
Zamolxis |
how can it be enabled? |
16:58 |
sfan5 |
see ipv6_server in minetest.conf.example |
16:58 |
Warr1024 |
If you have both v4 and v6 registered then it's possible that the client is picking either one of them. |
16:59 |
Warr1024 |
sfan5: I thought enabling v6 disabled v4, so I guess that's news to me. |
16:59 |
Warr1024 |
Some OS's let you bind a single socket to a multi-address-family wildcard so you can accept both v4 and v6 on the same port ... others require a socket to bind to only a single AF so you would need 2 sockets to bind to v4 and v6, even if each is a wildcard. |
17:01 |
Zamolxis |
So I'm setting both ipv6_server and enable_ipv6 to true and see how it works:D |
17:01 |
Zamolxis |
!up game.zendenpen.com:30000 |
17:01 |
MinetestBot |
game.zendenpen.com:30000 is up (42ms) (IPv6) |
17:01 |
Zamolxis |
Nice, it works, Thanks guys |
17:01 |
|
SFENCE_arch joined #minetest |
17:01 |
Zamolxis |
Or galls.. |
17:04 |
sfan5 |
this again makes me wonder why we don't enable ipv6_server by default |
17:05 |
Zamolxis |
so the solution is to enable ipv6 in minetest OR to disable the subdomain to use ipv6 or use a reverse proxy (never did this for UDP!) |
17:05 |
Zamolxis |
Yeah, cause it may cause some headache for newbies |
17:07 |
MTDiscord |
<warr1024> The only reason not to enable ipv6 by default is if it would cause issues on certain system configurations, like crashing on a system that doesn't have v6 enabled. If that's not an issue and we can make it Just Work automatically, then yeah, we should go for it. |
17:08 |
MTDiscord |
<warr1024> tbh you might just have to enable it in dev and see what kind of weird behavior you dredge up from the field when people try it out... 🤔 |
17:08 |
Zamolxis |
just wondering, is it normal for CPU to be between 20 to 30% just at idle on the server? |
17:09 |
sfan5 |
my personal opinion is that if you run a system with ipv6 hard-disabled in 2024 you deserve having to mess with one or the other application |
17:09 |
MTDiscord |
<warr1024> Is that 20% of all cores, or 20% of one core? |
17:09 |
sfan5 |
(note that hard-disabled is not the same as not configuring it, and you have to go out of your way to do so) |
17:09 |
MTDiscord |
<rollerozxa> the separate enable_ipv6 engine setting is already enabled by default fwiw |
17:09 |
MTDiscord |
<warr1024> "in 2024" as if IPv6 is any closer to being a reality than it was in 2014 😆 |
17:10 |
|
ireallyhateirc joined #minetest |
17:10 |
MTDiscord |
<warr1024> (we haven't even graduated to IPv5 😏) |
17:11 |
SwissalpS |
:D |
17:11 |
MTDiscord |
<rollerozxa> it's 2025 and IPv6 broadband is still practically nowhere to be seen in my country 🥲 |
17:12 |
SwissalpS |
yep |
17:13 |
MTDiscord |
<warr1024> One of the inherent and surprising challenges of layered architectures is that every time you try to make a thing better, some jackass will come along and find a way to tunnel an implementation of the better thing over the worse thing, and any attempt to improve the underlying architecture will have to "compete" against that. So instead of rolling out IPv6 you find that people are just using IPv6 over a VPN over websockets over HTTP |
17:13 |
MTDiscord |
over TLS over IPv4 😳 |
17:15 |
Zamolxis |
Warr1024: from what I'm seeing in htop it's over 1 cpu |
17:15 |
SwissalpS |
may/11 |
17:15 |
MTDiscord |
<warr1024> Then that's not great |
17:15 |
SwissalpS |
oops |
17:16 |
MTDiscord |
<warr1024> Mods have a pretty broad license to consume resources, and not all of them are careful about it, so pretty much any sufficiently complex mod could cause the main thread to remain very busy even if nothing is loaded. |
17:17 |
MTDiscord |
<warr1024> Also, if you have any forceloading enabled, or if players were recently on so that things are still active for a little while, then even without especially poor decision-making, any kind of mod functionality could consume extra resources. |
17:21 |
celeron55 |
if you just start the server, it shouldn't consume much cpu, but of course if the cpu is underpowered to a large enough extent then even that might as well be 100% |
17:21 |
sfan5 |
the only general answer is: it depends |
17:22 |
celeron55 |
a cheap vps can be surprisingly wimpy when compared to even a laptop |
17:23 |
Zamolxis |
aha, interesting, I remember 5 years ago I tried running it in a raaspberrypi and it just would lag out 2-3 players , and indeed I think the dedicated CPU VPS is wimpy |
17:23 |
Zamolxis |
I remember migrating to linode with 1 core and it would fly :)) now on a different provider (cheaper) and this may be it |
17:24 |
Zamolxis |
I tried not enabling too many mods, plus I'm using repixture, not mineclonia or mesecons |
17:24 |
Zamolxis |
Any way to see CPU breakdown by mod? |
17:24 |
celeron55 |
anyway running on an underpowred server is a common thing to do with luanti and there are guides for doing that |
17:25 |
Zamolxis |
well the server does have 4 cores so I think it should be able to handle at least 5 people at a time |
17:29 |
|
gregon joined #minetest |
17:31 |
MTDiscord |
<warr1024> In my experience, cheap VPS doesn't so much give you a "weak" CPU as it gives you a CPU whose effective performance can vary a lot over time. It might look like a powerful CPU and be able to meet some good benchmarks, but you could also see a lot of "VM steal" time happen suddenly, or it just stalls for several hundred milliseconds every so often. |
17:32 |
MTDiscord |
<bastrabun> Always depends on who shares the same host machine with you |
17:32 |
MTDiscord |
<warr1024> More cores won't automatically get you more players. No matter how many players there are, there's still only one world, and its loop needs to fit inside one core. I've rarely ever seen the server process use more than about 1.5 cores. |
17:32 |
|
SFENCE joined #minetest |
17:33 |
MTDiscord |
<bastrabun> If I paste a screenshot, will that magically work in IRC? |
17:33 |
Zamolxis |
OK so not really multithreaded, more like 1 core usage till it fills up? |
17:34 |
Zamolxis |
If you post a url link we should probably be able to click and see it |
17:34 |
|
est31 joined #minetest |
17:35 |
Zamolxis |
I was really curious to how deadlocks are handled in lua mods.. sometimes I see some lua code and I wonder, won't this freak things up? |
17:35 |
Zamolxis |
or has no one experinced such events?:) |
17:36 |
MTDiscord |
<bastrabun> What's deadlocks in this context? infinite loops? |
17:37 |
MTDiscord |
<warr1024> It's very multithreaded, it's just not very balanced. |
17:37 |
MTDiscord |
<warr1024> Deadlocks and whatnot are not an issue in mod code, because Lua itself can only run in one thread. |
17:37 |
MTDiscord |
<bastrabun> In MT you don't need multithreading to mess up stuff. You can mess up exceptionally well in singlethreading here |
17:38 |
|
Verticen joined #minetest |
17:38 |
Zamolxis |
This explains why I saw some very bad coding choices and wondered, won't this crash my freaking server? so if it's singlethreaded then it's probably event driven so first in first out |
17:39 |
Zamolxis |
especially looking at money mods, markets etc, i wonder if 2 people try to do the same action at once won't that affect anything , guess It's a surprise suprise kind of situation |
17:39 |
MTDiscord |
<warr1024> Threads exist for things like IO handling and stuff, but most of the core logic runs in the main thread. Almost all of the Lua code has to run in one main thread, with exceptions (like mapgen or workers) being very obvious when you encounter them (but not used nearly as often as you might expect). |
17:39 |
|
SFENCE joined #minetest |
17:41 |
MTDiscord |
<warr1024> Any sequential code in Lua can be assumed to run atomically (outside of crashes). Concurrency needs to be cooperative, so if it's not yielding to another procedure, no other procedure will run. Async workers can do work concurrently, but can't access the central game state concurrently. |
17:43 |
Zamolxis |
so animalworld with repixture is hardcore, most animals kill you almost instantly even with a good armor and they have like 65 HP while the main user has 20 ... probably not a good mod choice lol. |
17:44 |
Zamolxis |
I remember a few years ago HP was limited to 20 for players, I imagine this might be fixed by now? |
17:44 |
Zamolxis |
thanks for the explanations Warr1024 |
17:46 |
MTDiscord |
<warr1024> tbf the HP limitation is only relevant anyway if you use a pure engine health implementation. NodeCore has an 8hp system, but it's floating point so it's easier to scale to pretty much anything ... though I've been thinking for a while to change the overall scale because of how falling damage works. |
17:48 |
|
diemartin joined #minetest |
17:56 |
|
jemius joined #minetest |
18:07 |
|
Glaedr joined #minetest |
18:19 |
|
jaca122 joined #minetest |
18:26 |
|
Talkless joined #minetest |
18:52 |
|
SFENCE joined #minetest |
18:52 |
|
Thermoriax joined #minetest |
19:26 |
|
SFENCE joined #minetest |
19:33 |
|
SFENCE joined #minetest |
19:35 |
|
SFENCE joined #minetest |
19:41 |
|
SFENCE joined #minetest |
19:56 |
|
SFENCE joined #minetest |
20:10 |
|
SFENCE joined #minetest |
20:19 |
|
SFENCE joined #minetest |
21:53 |
|
turtleman joined #minetest |
21:54 |
|
YuGiOhJCJ joined #minetest |
22:02 |
|
mazes_83 joined #minetest |
22:04 |
|
Kimapr_ joined #minetest |
23:33 |
|
panwolfram joined #minetest |