**** BEGIN LOGGING AT Wed Dec 16 02:59:56 2020 Dec 16 04:09:53 jow: is luci login sysauth cookie the same as ubus session id? Dec 16 09:39:40 rr123: yes Dec 16 10:15:16 could packages be hosted on github/sf as an alternative to downloads.openwrt.org? Dec 16 10:15:59 is there a particular motivation for that? Dec 16 10:18:35 karlp: yes, multiple occasions I have had issues downloading packages from downloads.openwrt.org but GH/SF always works. and since it's a transient issue, I forget about it till it happens again. and I'm not the only one, here's another person facing the same issue - https://www.reddit.com/r/openwrt/comments/kd0548/knockd_on_openwrt_19074/gfygst6?utm_source=share&utm_medium=web2x&context=3 Dec 16 10:19:03 currently packages from GH load fine, but "curl -k -vv https://downloads.openwrt.org/releases/18.06.5/targets/x86/64/packages/Packages.gz" gives "curl: (35) ssl_handshake returned - mbedTLS: (-0x0050) NET - Connection was reset by peer" Dec 16 10:19:37 but I can download it fine in a browser behind the same openwrt router :/ Dec 16 10:20:00 doesn't really sound like a problem with the host hten :) Dec 16 10:22:46 I've never seen opkg croak with a download off github, just downloads.openwrt.org - https://bpa.st/XEMA Dec 16 10:25:14 I'm not convinced that there's any evidence that "just use this other place" will somehow make that magically different. Dec 16 10:25:25 I've cetainly had no end of problems with sf downloads personally Dec 16 10:26:16 well, if my recollection serves, opkg has had many issues downloading from openwrt.org over the years. ye old sf or recently? Dec 16 10:27:26 since package Makefiles and issues are being served by GH, I assume it can't hurt to have another hosting provider with a different cdn Dec 16 10:29:41 Anyone else seen an incompatibility between the cmake bump & ccache? It tries to pass an unsupported option to ccache and hence thinks the compiler's broken. Might just be a macos thing, or a ccache thing, or cmake thing, either way have fallen into problem and trying stuff to get out :-) Dec 16 10:39:59 yep, zlib and ccache are unhappy with the cmake bump. Dec 16 10:41:40 is this a cmake bump on macos? from what to what? Dec 16 11:00:19 no, cmake bump in master Dec 16 11:03:13 cmake and ccache has been a quite painful experience Dec 16 11:03:39 it does this wird compiler_arg1 thing and it is next to impossible to funnel flags through it properly Dec 16 11:04:08 that's why we ended up introducing the ccache_cc and ccache_cxx wrappers so that cmake has only one executable to work with Dec 16 11:04:21 but its chicken/egg and generally brittle, probably broke now in the process Dec 16 11:08:21 That's fine, I've never really timed how much difference it makes for me anyway and I'd only enabled it a few weeks ago.. no idea how popular or otherwise it is Dec 16 11:38:51 Regarding qos, there are too many options on openwrt, which one is the best? sqm-script, qos-script or gargoyle-qos Dec 16 11:41:32 Albert: sqm for generic usecases Dec 16 11:42:27 sqm + piece of cake. Dec 16 11:43:52 and you want to set your download and upload to about 5% less than your actual achievable values. Dec 16 11:45:43 make sure it's assigned to your wan, not lan interface. Dec 16 11:46:59 jow: Is this worth a backport? Seems important enough: http://lists.busybox.net/pipermail/busybox/2020-December/088362.html Dec 16 11:48:40 Why choose sqm? Dec 16 11:54:25 Albert: Because cake has the best all-around performance, is simple to configure and has tons of scientific research behind it. Dec 16 11:54:46 karlp: https://dev.archive.openwrt.org/ticket/19621 Dec 16 11:56:03 Strykar: what am I mean tto be reading there? Dec 16 11:57:19 if you're saying "somethings wrong with curl" then ok, sure, but that's pretty unrelated to "we should move package hosting to sourceforge or github" ? Dec 16 11:58:33 I didn't suggest move, merely mirror so there's more than once source Dec 16 11:59:11 mirror from github to sourceforge probably not the best option. mirror from github to gitlab, much better :) Dec 16 12:00:21 ultimately both could be mirrors of an openwrt-run gitlab instance. Dec 16 12:12:09 Strykar: there are plenty of mirrors for downloads.openwrt.org Dec 16 12:14:06 take http://rap.mirror.cyberbits.eu/openwrt/ for example Dec 16 12:14:18 rsalvaterra sqm be accelerated based on services, such as http, facebook or game Dec 16 12:15:14 That being said, I've never personally experienced issues with downloads.openwrt.org (except for the server-transition some weeks ago) Dec 16 12:17:02 why would anyone even consider using sourceforge. they fail more frequent, have lots of dead mirrors and don't care at'all Dec 16 12:17:57 couldn't download some sources that is needed to build openwrt from sourceforge and had to use look around for 3rd party that might have it Dec 16 12:20:08 I agree. sourceforge is offensive to use. Dec 16 15:57:40 hmm. lldpd doesn't seem to attach to lan ifaces. And when I force it too, it doesn't seem to receive anything Dec 16 16:13:32 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_tegra.html has been updated. (0% images and 97.1% packages reproducible in our current test framework.) Dec 16 16:41:28 f00b4r0: could you try to see with strace what it is doing? Dec 16 16:41:55 damex: on that device, strace isn't an option I'm afraid Dec 16 16:42:46 the most annoying part is that it doesn't hook to the lan device by default, despite the init script telling it too (from what I understand) Dec 16 16:43:08 hooroay, snapshots got lldpd ! :) Dec 16 16:43:37 f00b4r0: how do you set it up to bind for specific interface? Dec 16 16:43:58 I don't. By default it's supposed to do the right thing, reading the openwrt-provided init script Dec 16 18:04:33 f00b4r0: how do you check that it does not hook on lan by default? https://gist.github.com/damex/4d2687c5056569f368d4861b15db9c97 this is the default config i have on mikrotik hex-s that runs last snapshot build. lldpd listen only bridge ('br-lan', which is 'lan' in openwrt configuration) Dec 16 18:06:40 damex: for you it works. It randomly fails for me. lldpcli show config will typically display only "lo" in interface pattern Dec 16 18:07:15 at first I thought it was a race with the daemon starting before the lan interface was up, because restarting it manually "fixed" it. But it's not the case, even when restarting sometimes it won't hook. Dec 16 18:07:20 I'm testing on 19.07 though. Dec 16 18:23:22 jow: is uhttpd-mod-lua's only purpose to speed up lua processing? it does bring in about 2MB RAM usage, luci does not need it either, any other reason to use it? i can do lua cgi just fine without it, as expected Dec 16 18:58:31 any idea if mt7621 could offload ipv6 forwarding/nat? Dec 16 18:58:51 it does offload for v4 just fine (and works well with DSA) Dec 16 19:02:20 damex: https://bugs.openwrt.org/index.php?do=details&task_id=1916 seems to not work in ipv6 Dec 16 19:02:40 in openwrt Dec 16 19:04:19 jow: sorry, what was the conclusion about the correct way to avoid filling /var/log/messages with “Permission denied” messages from Bind? Dec 16 19:04:29 (besides not running a name server locally…) Dec 16 19:05:17 SwedeMike: thanks, 18.06 use swconffig and it was updated to work with dsa recently. Dec 16 19:06:16 i guess i need to test it myself to verify that it works (or not) Dec 16 19:06:44 damex: ipv6 hwnat offload is spotty in commercial mt7621 offerings as well, so there must be some gotcha Dec 16 19:14:26 jow: nice review on the iperf3 stuff, by the way. some really good comments. Dec 16 19:33:37 damex: got my er4, very happy with it so far. haven't used sfp yet Dec 16 19:33:52 thanks again for adding support :) Dec 16 19:34:14 Borromini: thanks for the feedback :) Dec 16 19:35:24 kudos to damex for that! Dec 16 19:36:57 Borromini: what kind of performance are you seeing? Dec 16 19:37:01 funny how everything seems worse when using ipv6 ;P Dec 16 19:37:19 :P Dec 16 19:37:19 f00b4r0: depends on platform, it works great on my APU2 Dec 16 19:38:05 SwedeMike: I was being sarcastic: 25 years since RFC and it's still not broadly deployed and iffy. Dec 16 19:38:18 it's semi-broadly deployed Dec 16 19:38:32 f00b4r0: I've had IPv6 home for 10+ years now, lately it's even native Dec 16 19:38:42 in countries where one large enough ISP pays enough attention to it, most other ISPs kinda followed Dec 16 19:39:00 mobile, sure. Enterprises, mostly. End-user "landlines", still leaves a lot to be desired. Dec 16 19:39:04 f00b4r0: https://www.google.com/intl/en/ipv6/statistics.html we're now over 30% of eyeballs with IPv6. Dec 16 19:39:19 I had IPv6 at home 18 years ago Dec 16 19:39:28 f00b4r0: enterprise is the worst. Residential broadband and mobile is best. Dec 16 19:39:46 good times =) Dec 16 19:39:53 if I enable ipv6 on my link (provided I find the correct parameters to do so, my ISP doesn't advertise), first thing that'll happen is I'll lose reverse DNS. Dec 16 19:40:08 which has all sorts of "interesting" consequences. Dec 16 19:40:20 f00b4r0: IPv6 works out of the box with openwrt defaults on my ISP. Dec 16 19:40:36 good for you :) Dec 16 19:40:59 that's how it works on most ISPs Dec 16 19:41:05 sorry for you having a bad one Dec 16 19:41:38 *shrug* Dec 16 19:41:55 how would I boot an AMD x86 coreboot SoC kernel with iomem=relaxed Dec 16 19:42:10 Strykar: give it some xanax Dec 16 19:42:15 SwedeMike: it's beefier than my connection atm, which is 100/30 :P Dec 16 19:42:15 lol Dec 16 19:42:24 sorry, bad joke Dec 16 19:42:27 stintel: all I have is some weed sadly Dec 16 19:42:32 f00b4r0: what ISP are you with? Dec 16 19:42:38 you're in france right? Dec 16 19:42:40 Borromini: K-net. Dec 16 19:42:47 aka Kwaoo Dec 16 19:42:55 Strykar: I believe there is a .config option for it Dec 16 19:43:00 let me find it Dec 16 19:43:11 stintel: so short of recompiling the kernel, I can't? Dec 16 19:43:22 f00b4r0: fiber and no ipv6? Dec 16 19:43:28 ~ Dec 16 19:43:35 pretty much yeah Dec 16 19:43:42 that sounds weird. Dec 16 19:43:44 although I can't say it's bothered me much. Dec 16 19:43:44 stintel: its CONFIG_IO_STRICT_DEVMEM and CONFIG_STRICT_DEVMEM Dec 16 19:43:46 https://stats.labs.apnic.net/IPv6/FR France is still doing kind of ok. KWAOO K-NET SARL is only at 16% though Dec 16 19:44:13 SwedeMike: I'm willing to bet that 90% of these numbers are mobile Dec 16 19:44:29 all mobile providers have now switched to v6 native, AIUI Dec 16 19:44:40 no, it's Free and Orange Dec 16 19:44:47 zorun: ? Dec 16 19:44:52 f00b4r0: did you bet without even looking at the link? Dec 16 19:44:59 Strykar: on x86 you might be able to edit the grub cmdline by interrupting the boot process, if that's an option Dec 16 19:45:13 Free have been doing ipv6 for ages, and Orange enabled it massively a few years ago Dec 16 19:45:20 SwedeMike: I looked at the link Dec 16 19:45:24 zorun: not in Belgium they didn't :( Dec 16 19:45:38 stintel: grub on openwrt? Dec 16 19:45:46 I didn't scroll down, but I see the top names as mobile providers, which tends to confirm Dec 16 19:45:48 sales folks couldn't even answer my question about Orange+IPv6 Dec 16 19:45:51 Strykar: on x86, yes Dec 16 19:46:05 zorun: amusingly, Free was late to ipv6 on mobile Dec 16 19:46:45 SFR seems nowhere to be seen. How unsurprising. Dec 16 19:46:49 they were more like IPvwhaaaaaat? Dec 16 19:47:21 good thing my parents told me to join the sales meeting lol Dec 16 19:47:28 zorun: also, until recently, Free was v6-in-v4, which sucked. Dec 16 19:47:34 sounded great in theory Dec 16 19:48:06 f00b4r0: yup, now it's v4-in-v6 ;) Dec 16 19:48:16 6RD was ok for initial deployment, after a while, no. Dec 16 19:48:45 zorun: what did they deploy? Dec 16 19:49:07 for v4-AaaS Dec 16 19:49:14 4rd Dec 16 19:49:49 really? Dec 16 19:50:04 f00b4r0: btw, mobile ipv6 in France is nowhere near the level of fixed access Dec 16 19:50:14 zorun: ? Dec 16 19:50:17 stintel: I see /boot/grub/grub.cfg but opkg does not list a "grub-mkconfig", there's a "grub2-editenv" Dec 16 19:50:34 (trivia: Free invented 6rd) Dec 16 19:50:36 that list gives Orange, Free and Bouygues at the top, it's broadband Dec 16 19:50:45 it's not Dec 16 19:50:47 except maybe for bouygues Dec 16 19:50:51 Bouygues mobile is 100% v6 Dec 16 19:50:58 Strykar: just edit it manually Dec 16 19:51:00 they've been among the first: they ran out of IPs Dec 16 19:51:01 did they finally enable it by defualt? Dec 16 19:51:14 it's been on for ages. Where have you been? ;P Dec 16 19:51:15 they ran out of Dec 16 19:51:17 Strykar: everything normally happens during image build (which happens on different hardware) Dec 16 19:51:17 they ran out of *private Dec 16 19:51:22 they ran out of *private* IPs ;) Dec 16 19:51:47 oh, ok, didn't follow, at the beginning it was a manual process Dec 16 19:52:12 "BOUYGTEL-ISP". Keyword "TEL" Dec 16 19:52:16 it's the mobile arm Dec 16 19:52:31 how you know my password?! Dec 16 19:52:36 comcast initially ran out of RFC1918 space for management of their cable modems, that was their initial driver for IPv6 deployment.. for management. Dec 16 19:52:36 lol Dec 16 19:53:21 SwedeMike: yeah usually it ends up biting them in the ass. Dec 16 19:53:36 Free is using shared v4 nowadays. Which is an abomination Dec 16 19:54:16 SwedeMike: it was a problem for management but not for deployment?! Dec 16 19:54:16 doing v4-as-a-service over IPv6 with A+P is the least bad way to do things Dec 16 19:54:37 I have to confess though, all the talks about "migration" don't make sense to me. You can't have both system coexist and hope that it'll be fine Dec 16 19:54:39 stintel: they deployed it as well to customers. Dec 16 19:55:01 SwedeMike: yeah but I mean ... they had more modems than RFC1918 would allow for? Dec 16 19:55:12 and at the same time it's not a problem for public address space? Dec 16 19:55:16 f00b4r0: the end-game will probably be a NAT64 gateway at the corner of the network or something, for residual IPv4 connectivity to laggards Dec 16 19:55:38 stintel: I don't know their IPv4 situation. Dec 16 19:55:50 for public IPs Dec 16 19:56:08 either way I think I've had enough internet for today Dec 16 19:56:28 i remember reading an article that put in plain words why this whole migration process is doomed. I can't remember if it was written by ESR or some other figure Dec 16 19:57:41 lots of haters. Dec 16 19:58:22 it wasn't hate. It was actually a technically sound argument for why a transition will bog the switch down. Dec 16 19:59:26 did the author have a better suggestion how to do things? Dec 16 19:59:51 has anyone seen IPv6 inside bigger companies as the main way to address internal hosts? Dec 16 19:59:57 i'll see if I can find it back Dec 16 20:00:06 Hauke: hehe Dec 16 20:00:28 SwedeMike: istr the bottomline was, ipv6 is stillborn Dec 16 20:00:51 f00b4r0: sigh. Dec 16 20:01:27 I've yet to see anyone come up with a better suggestion Dec 16 20:01:33 lots of people with strong opinions on IPv6 that never bothered to deploy it. Dec 16 20:01:46 IPv6 supports all of the transition methods that can work... and the ones it doesn't support don't work Dec 16 20:03:25 there is more work to be done, but IPv6 works well today and is deployable by people who want to. Dec 16 20:03:58 SwedeMike: the fact that it's been 25 years and its still not even past a third of traffic is rather telling, tbh Dec 16 20:04:41 it's telling, but what it tells is that people can't handle long-term planning and are scared of change Dec 16 20:04:47 it doesn't actually say that the design is wrong Dec 16 20:04:56 f00b4r0: it's telling, but telling different things to different people Dec 16 20:05:08 I'm sure. Dec 16 20:05:20 if you can suggest an alternative design that would be easier to deploy, then I'm all ears Dec 16 20:05:52 so long as a) it works, b) it's not already a thing you can do in v6 Dec 16 20:06:30 IPv6 is difefernt to IPv4 and all the things that are needed work with with IPv4, the new features people are not aware of Dec 16 20:07:15 or they actively don't want them because they modelled their entire operational process on how things work in IPv4 Dec 16 20:07:35 SwedeMike: yes Dec 16 20:10:41 ignorance is bliss, should just pull the plug on ipv4 Dec 16 20:11:02 i remember a key argument was that ipv6 isn't backward compatible, and ipv4 isn't forward compatible. Which kinda breaks rule #1 of a transition path Dec 16 20:12:11 f00b4r0: when did you get into networking? What decade? Dec 16 20:12:12 v6 /is/ backwards compatible though Dec 16 20:12:42 v4 isn't forwards compatible, but that's not a problem with v6, it's a problem with v4 -- the only way to fix it would be to roll out a replacement protocol for v4, and that's exactly what v6 is Dec 16 20:13:06 Dagger: it's not. You can't receive a v4 packet on a v6 stack and expect it to work. Dec 16 20:13:22 f00b4r0: when I got into networking in the 90ties it was very common to run multiple networking protocols on the same wire Dec 16 20:13:46 f00b4r0: it seems to be only people who started in networking after ~2000 who are baffled by running multiple protocols at once Dec 16 20:13:57 you can do that. that's what ::ffff:0:0/96 is for Dec 16 20:14:30 Dagger: host A is v4 only, wants to talk to host B which is v6 only. Doesn't work. Dec 16 20:14:39 I think I've found the article I had in mind Dec 16 20:14:43 you can also connect outwards to v4 addresses, either with ::ffff:0:0/96, NAT64 or dual-stack Dec 16 20:14:47 if this is the djb article... -.- Dec 16 20:14:53 written by one Geoff Huston, Chief Scientist at APNIC Dec 16 20:15:01 http://www.circleid.com/posts/is_the_transition_to_ipv6_a_market_failure/ Dec 16 20:15:12 you'd expect him to know what he's talking about. Dec 16 20:15:35 f00b4r0: yes, I know him, I've had several conversations with him face to face. He also likes to be alarmist and absolutist to make a point. Dec 16 20:15:57 I'm not 100% sure it's the piece I remember, but it develops the same arguments Dec 16 20:16:08 f00b4r0: okay, that's because v4 doesn't handle longer addresses, which is a problem with v4 Dec 16 20:16:10 f00b4r0: and I've told him people like you take his statements at face value, takeaway that IPv6 sucks, and never do it. Dec 16 20:16:41 Dagger: the argument was back/forward compat. That's the point. Dec 16 20:17:06 f00b4r0: how much IPv6 deployment have you done? Dec 16 20:17:20 SwedeMike: well, you don't have to convince me. I'm a measly lowlife end user. But I'm afraid he has a good analysis of what's going on. Dec 16 20:17:35 f00b4r0: you can work around it with NAT46 or a reverse proxy, or by giving the v6 host a v4 address. all of which v6 supports Dec 16 20:17:49 Dagger: *sigh* Dec 16 20:17:55 you fail to see the point. Dec 16 20:18:15 well the point I'm trying to make is that v4 isn't forwards compatible with longer addresses, and *nothing can ever fix that* Dec 16 20:18:22 dual-stack isn't backward compatibility. It's two incompatible systems living side by side. Dec 16 20:18:36 v4 not being forwards compatible isn't a design failure on the part of v6. it's a design failure on the part of v4 Dec 16 20:19:01 i don't disagree. I'm saying it might explain why the "transition" isn't working so well. Dec 16 20:20:18 "pull the plug" see how fast the transtion goes Dec 16 20:21:02 *nod*. If anything, I suspect dual-stack is providing _less_ incentives to move forward Dec 16 20:22:03 sure, so that's what the long transition is telling you: that v4 screwed up. if you have any great suggestions for how to fix that then make them, but otherwise let's not use it as an excuse to keep not doing v6 Dec 16 20:22:07 for any ISP deploying IPv6 to residential users, they get more than half the traffic over IPv6 because all the major CDNs running video traffic are IPv6 enabled. Dec 16 20:22:36 IPv4 was way too forward looking. It should have stayed at 24 bit address or something so it needed to be replaced earlier Dec 16 20:22:52 heh Dec 16 20:22:59 it's an experiment that escaped the lab, never designed to run at world scale Dec 16 20:24:10 SwedeMike: OTOH that probably would've given us a 64-bit v6 which would probably lead to needing NAT. so this might be better in the long run Dec 16 20:24:30 https://www.networkworld.com/article/2227543/software-why-ipv6-vint-cerf-keeps-blaming-himself.html Dec 16 20:24:41 SwedeMike: 6rd, which some ISPs were doing here for a while, made things worse imo: by switching to ipv6, the end user ended up with longer hops to reach anything, poor latency, etc Dec 16 20:24:52 "21:22 < SwedeMike> it's an experiment that escaped the lab, never designed to run at world scale" Dec 16 20:24:54 not to mention basic stuff like reverse dns not working Dec 16 20:25:00 so true for aall kinds of techs Dec 16 20:25:03 (still the case with most ISPs here, AFAICT) Dec 16 20:25:41 :c Dec 16 20:25:48 f00b4r0: I gave up on relying on reverse-DNS 20 years ago, what are you using it for? Dec 16 20:26:15 geolocation Dec 16 20:26:40 SwedeMike: *I* am not using it for much besides aesthetics. But some services do rely on rlookup match to allow incoming connections Dec 16 20:26:44 light bulb at living roon, light buld in kitchen Dec 16 20:27:33 I run a moderately loaded postfix MTA, and that simple check rids me of 90% spam; for instance. Dec 16 20:29:16 yeah, that's probably one of the very few remaining use-cases, and not applicable to wider IPv4/IPv6 deployment. Anti-spam needs to adapt its reputation system to something better. Dec 16 20:30:06 I use it to look up the hostname for a given IP. why is that not a valid use-case? :( Dec 16 20:30:27 SwedeMike: are you suggesting reverse is entirely unnecessary, at the end of the day? Dec 16 20:31:07 f00b4r0: I doubt we'll see increased use of it, rather the opposite Dec 16 20:31:16 I'm afraid you're right ;P Dec 16 20:32:16 change starts with you, demand it Dec 16 20:32:19 Dagger: because most people do not care about reverse-DNS, and there still isn't generated reverse for IPv6 for residential for instance. Dec 16 20:33:54 theres no "generated" reverse dns for ipv4 in residential areas either AFAIK Dec 16 20:35:04 grift: ? Dec 16 20:35:20 all FR ISPs provide a generic v4 reverse. Some even offer custom Dec 16 20:36:09 look i am willing to bet my left nut that this isp doesnt even offer more than one ip address Dec 16 20:36:59 grift: for residential? Of course not Dec 16 20:37:08 which 8 port switch is supported by the realtek target? Dec 16 20:37:13 "ofcource" ? Dec 16 20:37:30 Hauke: zyxel GS1900 iirc Dec 16 20:37:32 Hauke: there are many ? Dec 16 20:37:36 ISP-generated reverse is bleh anyway. the ISP has no idea what my rDNS entries need to be, so they need to delegate the zone to me rather than invent crap Dec 16 20:37:46 https://biot.com/switches/+ Dec 16 20:37:50 https://biot.com/switches/ Dec 16 20:37:58 without the + sorry Dec 16 20:38:18 thats the thing with this tech and most other techs, its misunderstood Dec 16 20:38:26 people are used to one ip address Dec 16 20:38:31 thats not the norm Dec 16 20:38:34 but if this tickles your pickle, please join this community Dec 16 20:38:55 grift: my ex uni owns a full class B Dec 16 20:38:58 (I mean, it's fine for them to do that if the customer configures them to do that, I guess) Dec 16 20:39:09 all devices had public IPs, even though not reachable from the outside. Dec 16 20:39:15 Such luxury ;^) Dec 16 20:39:43 thats the norm ipv4 was just a fluke Dec 16 20:40:14 incidentally, that class B is probably worth $$$, but they can't sell it Dec 16 20:40:22 when hat was invented (the need for it was identified) that when ip failed Dec 16 20:40:32 stintel: thanks+ Dec 16 20:40:55 Hauke: welcome ;) Dec 16 20:41:30 this industry is all about excuses, just say we failed, either we start over or you can get back to sending your pmail with pigeon Dec 16 20:42:13 i bet people will not choose the pigeon Dec 16 20:42:33 f00b4r0: my ex uni also had a class B, but all reachable from the internet at least in the computer sience department Dec 16 20:42:54 when I was in the wifi network everyone could acces the webserver running on my laptop ;-) Dec 16 20:43:00 i dont have a job, but i have 65K routable ip addresses Dec 16 20:43:15 1 IPv4 is 20$ to 30$ Dec 16 20:43:25 Hauke: neat. When I was there, I had an entire C to play with. Pays off to befriend the sysadmin and run cool IT projects ;-) Dec 16 20:44:02 Hauke: my understanding is that contiguous blocks fetch a premium? Dec 16 20:44:14 yes I think so Dec 16 20:44:26 so a class b is probably more the 30$ per IP Dec 16 20:44:31 makes sense Dec 16 20:44:59 at least some are more expensive than others and to me only bigegr blocks look better Dec 16 20:45:06 so c.2M$ give or take. not bad Dec 16 20:46:17 just checked, they still own it. Dec 16 20:46:38 i guess that is the solution raise prices of ip4 address 2000% Dec 16 20:46:52 lets see how fast the !@#$#$W transtion Dec 16 20:46:57 meanwhile that rtl port looks promising. I'll take my GS1900-8 for a spin, I guess Dec 16 20:57:26 sorry but thats how it works, these people think in terms of profit and just make it unafordable (add a new "obstruction" tax) Dec 16 20:57:59 f00b4r0: i bought a GS1900-10HP because of it :) Dec 16 20:58:04 and running it now :^) Dec 16 21:03:51 re ipv6 is awesome, "I turned off ipv6 (no reason, just did) and things broke" https://github.com/eclipse/mosquitto/issues/1955 Dec 16 21:05:36 if IPv6 breaks things you need to fix them, not ignore them Dec 16 21:05:55 ip6 isnt awesome -- ip4 has been inadequate for decades Dec 16 21:05:57 in 20 years, IPv4 will be obsolete technology Dec 16 21:06:08 this is the other way, everything's fine, they tried turnint it off, and somethign stopped Dec 16 21:06:18 I suspect their "off switch" isn't actually appropraite. Dec 16 21:06:38 same with DAC , that stuff was invented in labs as well, they didnt foresee this Dec 16 21:07:08 my colleagues say that v6 will be (and is) the ghetto internet, for people too poor to have a v4 ip... Dec 16 21:07:27 but I didn't like that view of the world :| Dec 16 21:07:27 stintel: proximus broke my brother's ipv6 :P Dec 16 21:07:45 Borromini: proximus broke me 2nd sim. and I live 2000km away Dec 16 21:07:51 karlp: if our leader make it Dec 16 21:08:13 grift: pick a new leader Dec 16 21:08:17 thats the this when this tech was academic they academics sold out Dec 16 21:08:29 stintel: aren't they lovely :P Dec 16 21:08:40 they basically sold "the company" Dec 16 21:08:41 Borromini: no they aren't Dec 16 21:08:53 hence the smiley ;) Dec 16 21:08:57 Borromini: spare me the rethorical ;) Dec 16 21:09:34 completely off topic, but I'm very impressed at JLCPCB's order progress tracking Dec 16 21:09:51 anyway this tech was never designed to be used on suck a scale but it was sold as such nonetheless Dec 16 21:09:59 karlp: yeah, I'd say quite a few applicatiosn today assume there is basic IPv6 support enabled in the OS and it should only be turned off per interface, not completely Dec 16 21:10:26 like being able to bind to :: to get all interfaces instead of 127.0.0.1 Dec 16 21:15:04 just to make it clear i am talking about ipv4 here. its the same challenge with many subsystems like also "access control" where a tech was designed to be used in much smaller scale (namelu a few elites) but then it was "sold" and now that tech that was ment to be used by few academic is used by the whole world Dec 16 21:15:35 and ofcourse that doesnt "fit" Dec 16 21:15:52 why would anyone try to defend that? Dec 16 21:16:36 just tell them like it is, if youre still on ip4 and if youre still relying of DAC (etc) then youre hopelessly living in the past Dec 16 21:16:53 * stintel applauds grift Dec 16 21:17:46 there is no point in defending ipv4. unfortunately roughly half of this world accepts v4 only as an internet connection. we need to change this Dec 16 21:18:12 right raise taxes Dec 16 21:18:21 if your connection is v4 only, you do not have access to the "entire" internet Dec 16 21:18:25 just please, remember that collective punishment is NOT the way to go abotu it. Dec 16 21:18:37 so you shouldn't be paying for that kind of connection Dec 16 21:18:56 and any !@#$% isp telling you that any prefix lower than /48 is reasonable tell them to !@#$ OFF Dec 16 21:19:42 karlp i dont understand Dec 16 21:20:01 its obvious were dealing with intergarten material here Dec 16 21:20:23 its the only way, brute force Dec 16 21:21:47 ynezz: a few days ago i became aware of this in the context of adding support for bullet-ac: https://github.com/true-systems/ubnt-openwrt-flashing ... did you give any thought to trying to disable the rsa key check in the ubiquiti u-boot? Dec 16 21:22:04 thing with humanity is that one will always try to control one-other, you you have to do this collectively Dec 16 21:22:36 no, this IT attitude that you can hammer users to make them apply pressure on isp's is batshit crazy, and incredibly arrogant, patronizing and ... just WRONG. Dec 16 21:23:03 isp is just another business Dec 16 21:23:10 its all about profit Dec 16 21:23:23 we've seen this fail abyssmally with spam black lists Dec 16 21:23:31 all we've actualyl done is push to more monopolies Dec 16 21:23:44 while spamhaus strokes their beards and thinks they're making the world a better place Dec 16 21:23:55 we still have massive chunks of the world with no _real_ alternative ISP Dec 16 21:24:07 and IT peopel going "urh urh, the free market will provide yuou with a better isp!" Dec 16 21:24:09 no they won't Dec 16 21:24:11 it's a joke Dec 16 21:24:17 collective punishment is WRONG Dec 16 21:24:20 i dont see the connection, just pull the ip4 plug, raise taxes on ip4 address 5000% Dec 16 21:24:45 in otherwords, exactly what I said earlier, v6 becomes ghetto internet for the poor Dec 16 21:25:03 for the rich you mean Dec 16 21:25:21 well theres more poor than rich anyway so ... Dec 16 21:25:45 eventually the remaining ip4 elites will come to ip6 Dec 16 21:26:06 v4 will be the ghetto internet for the rich Dec 16 21:26:16 which makes no sense, but... nevertheless Dec 16 21:26:18 good Dec 16 21:26:40 i dont see a problem with that Dec 16 21:26:53 rich people are aleady on v4, they'll stay there, you want to mak ev4 more expensive, so the rich stay there, and no-one else can join them? Dec 16 21:27:10 theyll isolate themselves Dec 16 21:27:16 and then disconnect ipv4 from the internet you mean? :P Dec 16 21:27:21 make it ipv6 only <3 Dec 16 21:27:27 ye send it to marse with elon Dec 16 21:27:28 you were just earlier saying that it's business? Dec 16 21:27:29 I want v6 deployment. I don't care what people do with v4 after that Dec 16 21:27:35 and then we solved the bigger problen Dec 16 21:27:37 business wants to sell to people with money... Dec 16 21:27:39 making v4 expensive seems to encourage people to do v6 deployments Dec 16 21:27:43 who are demonstrably on v4 Dec 16 21:30:06 business doesnt want to sell to people with money, they want profix they want to grow Dec 16 21:30:32 and the profix is with the poor minority Dec 16 21:31:08 but sure let them choose Dec 16 21:31:44 i mean its a simple calculation: Dec 16 21:32:04 i get 1 ip4 address and i get 64K ip addresses Dec 16 21:32:07 hey, look as long as we're not doing collective punishment, I'm ok with whatever :) Dec 16 21:33:16 anyone with half a brain (that means 0.01 percent of our population) will that well if i can get 64K ip addresses for the price of 1 then that sounds more attractive Dec 16 21:34:03 point i am trying to make is that the masses need direction Dec 16 21:34:18 they dont want discretion , they cant handle that Dec 16 21:37:39 its not punishment either. its just a choice. i think NAT is punishment Dec 16 21:38:30 just pull the plug and everyone will benefit Dec 16 21:38:32 64k subnets, presumably. 64k v6 addresses is an abnormally and unjustifiably small allocation Dec 16 21:38:59 and yes, NAT is collective punishment, but everybody seems to love the extra effort/cost/breakage that it brings :/ Dec 16 21:39:10 right so next thing to do is put jail sentences on anyone claining that /48 is too much Dec 16 21:39:30 every person on earth needs to get atleast /48 Dec 16 21:43:50 but dont worry about me, theyre in the process of bringing water to the stock exchange. so by that time youll have other things to worry about Dec 16 21:46:39 i remember me 2 decades ago screaming about roof tops that everyone needs atleast 512/512KiB , they pronounced me "nuts" . two decades later (and i still done have a job or income) i am on 100/100Mbits with 64K p6 addresses Dec 16 21:47:30 grift: at that time I was working for a swedish startup that did ETTH and 10/10 to customers. Dec 16 21:47:56 was a lot of buzz here around then about that Dec 16 22:51:57 find staging_dir/target-powerpc_464fp_musl/ -xtype l | wc -l Dec 16 22:51:58 37 Dec 16 22:52:02 what the hell Dec 16 23:32:15 build #233 of sunxi/cortexa53 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/openwrt-19.07/images/builders/sunxi%2Fcortexa53/builds/233 Dec 17 00:40:44 build #227 of at91/sam9x is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/openwrt-19.07/images/builders/at91%2Fsam9x/builds/227 Dec 17 00:50:54 build #228 of x86/64 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/openwrt-19.07/images/builders/x86%2F64/builds/228 Dec 17 02:43:57 Hauke: I think now you removed support for gcc5, doesn't that mean we have to apply ynezz patches for gcc6+? **** ENDING LOGGING AT Thu Dec 17 02:59:57 2020