**** BEGIN LOGGING AT Mon Jul 10 03:00:03 2017 Jul 10 17:30:41 timeless: wb! Jul 10 17:30:48 thanks Jul 10 17:31:38 I wonder how networking works at all in containers Jul 10 17:32:13 in the end you need virtual NIC (or bridge) just like on any other virtualization Jul 10 17:32:54 unless you have a dedicated physical NIC for each container Jul 10 17:33:27 right? Jul 10 17:33:42 there are choices Jul 10 17:33:50 you can assign a device straight into a container Jul 10 17:33:56 you can let a container use NAT Jul 10 17:34:09 or you can set up a bridge of a real device and then allow access to the bridge Jul 10 17:35:18 * DocScrutinizer05 needs to sort his mind re bridge vs NAT Jul 10 17:35:39 for more fun, my kernel is 2.6 Jul 10 17:35:39 iirc a bridge is simply a sw-switch, right? Jul 10 17:36:19 at least on my router, it seems to be Jul 10 17:36:23 that might be a reasonable way to look at it Jul 10 17:36:58 roughly, a bridge allows multiple devices to claim to have distinct ip addresses together all of which can route out using the underlying device Jul 10 17:37:36 thought the distinct IP addr would turn it into sth involving NAT Jul 10 17:37:57 unless you do full routing Jul 10 17:38:01 the key difference is that w/ nat, you generally can't reach the connected devices from the host Jul 10 17:39:45 on my router br0 IFs all have same network and thus IP Jul 10 17:39:54 NAT is functionally equivalent to a program acting as a client and making all outbound requests on behalf of all the things that it's natting Jul 10 17:40:50 http://wstaw.org/m/2017/07/10/plasma-desktophU8317.png Jul 10 17:41:44 for more fun, this computer has two bond interfaces Jul 10 17:41:58 iirc eth0+eth1 = bond0; eth2+eth3 = bond1 Jul 10 17:42:48 bond is channel bundling for bandwidth? Jul 10 17:43:29 and reliability in case one of the cards/cables fails Jul 10 17:43:35 I heard from a sysadmin friend that systemd killed all the bond groups in their corporate infra Jul 10 17:43:55 joy, i can't wait for that Jul 10 17:44:15 systemd/udev are replacing the eth{randomnumber} system w/ a different means of assigning names Jul 10 17:44:39 SLE Jul 10 17:44:47 11? Jul 10 17:44:53 sle? 11? Jul 10 17:45:03 Suse Linux Enterprise 11 Jul 10 17:45:10 i'm on rhel6 :) Jul 10 17:45:31 with Service Level agreement and on-site suse dude who's cursing whole day long Jul 10 17:46:20 systemd is real fun, the larger the infra the more the fun Jul 10 17:46:21 i'm surprised they didn't test their rollout a couple of systems at a time Jul 10 17:46:46 anyway, what i really want is a vbox style network Jul 10 17:46:49 https://www.thomas-krenn.com/en/wiki/Network_Configuration_in_VirtualBox#Host-only_Networking Jul 10 17:47:42 no idea about LXC here Jul 10 17:47:49 neither about docker Jul 10 17:47:58 we've occasionally tried docker Jul 10 17:48:07 we decided it was a PITA and totally unusable Jul 10 17:48:18 i personally decided that their OpenSource system was *awful* Jul 10 17:48:32 they were really really really bad at accepting PR requests via github Jul 10 17:48:41 seems Poettering likes it, so it must be awful Jul 10 17:49:06 in theory, it's a good idea Jul 10 17:49:25 but their docs, their PR system, their people... all lead me to not wanting to touch it w/ a 10' pole Jul 10 17:50:22 the idea is great. and i think that docker has a proper local network bridge Jul 10 17:50:39 I only ever looked into Xen a tiny bit, with maemo infra. And the usual "everything for testing" on my personal workstation Jul 10 17:51:18 i've used xen a bit, i think kvm a bit, vbox a bunch, occasionally vmware / fusion (iirc qnx/bb10 favored them) Jul 10 17:51:23 I think a decent full virtualization is better than that para(?) stuff Jul 10 17:51:46 oh, and i used solaris zones while hosting maemo stuff :) Jul 10 17:52:01 huh? Jul 10 17:52:04 solaris zones were awesome. lxc is only now barely close to where solaris zones were 10 years ago :( Jul 10 17:52:17 my cross reference was done using a lot of magic Jul 10 17:52:27 opensolaris host os, nexenta guest zone Jul 10 17:52:31 aaah the awesome MXR(?) Jul 10 17:52:39 nexenta is a debian solaris userspace Jul 10 17:52:48 yep Jul 10 17:52:58 so i used that to get the sources from .deb's Jul 10 17:52:58 a pity that vanished Jul 10 17:53:03 yeah, sorry about that Jul 10 17:53:28 couldn't have done H-E-N without Jul 10 17:53:43 it might be possible to talk to http://timeless.justdave.net/ (say "host: mxr.maemo.org") Jul 10 17:54:17 oh, no, because there's too much magic involved Jul 10 17:54:25 we should port that stuff to maemo infra Jul 10 17:54:27 dreamhost doesn't know about it Jul 10 17:54:37 i'm pretty sure the data is still in timeless.justdave.net Jul 10 17:55:12 if you look at http://timeless.justdave.net/mxr-test/mxr-test/source/ Jul 10 17:55:13 warfare: xes_^: any chance for that? Jul 10 17:55:19 you can see that the configs are mostly there Jul 10 17:55:54 timeless: given you get access on a VM, would you be able and willing to port stuff to there? Jul 10 17:55:56 anyway, solaris zones are basically proper lightweight virtualization Jul 10 17:56:13 DocScrutinizer05: eventually, maybe. not anytime soon Jul 10 17:56:27 right now i need to focus on fixing these two lxcs Jul 10 17:56:31 sure, it's dead since3 what? 5 years now? Jul 10 17:56:39 something like that... Jul 10 17:57:05 anyway... lxc hell Jul 10 17:57:23 at large or just networking? Jul 10 17:57:30 just networking Jul 10 17:57:44 roughly i'm hosting a `pydio` instance (that i don't trust much) in an lxc Jul 10 17:58:07 never heard of Jul 10 17:58:16 pydio.com/ Jul 10 17:58:25 and i just need to let the host talk to the pydio instance so that the internet can access it Jul 10 17:58:31 it's basically a file sharing system Jul 10 17:58:39 w/ semi-shiny ui Jul 10 17:58:53 i don't really care. mostly it's a web server and scary [hence lxc] Jul 10 17:59:07 in theory, this would be a good chore for docker Jul 10 17:59:33 I'm not an expert on lxcs but we do use them at work Jul 10 17:59:53 (for bulk of the stuff anyway, not all of it) Jul 10 18:00:10 i had this lxc working, but apparently not enough of it was set to automatically start after a reboot :-( Jul 10 18:00:16 afk Jul 10 18:00:34 I handle them with proxmox Jul 10 18:00:45 makes it nice and clicky mostly Jul 10 18:01:45 * timeless googles Jul 10 18:02:55 joga: did anyone at your company worry about AGPLv3? Jul 10 18:03:20 ...not to my knowledge Jul 10 18:03:29 ... did anyone look at it? :) Jul 10 18:04:06 well I didn't at least :) Jul 10 18:05:08 some of these licenses (and i can't remember which) have cascading impact on things that run nearby Jul 10 18:05:47 imagine a "GPL-like license" on a "linux kernel" that required all userspace to be governed by that those same license terms Jul 10 18:06:02 sadly, i can't remember which licenses have that property Jul 10 18:06:24 you mean I couldn't host a webapp on a container in proxmox? Jul 10 18:06:33 i'm not saying that Jul 10 18:06:42 i don't remember which license was risky Jul 10 18:07:03 but in theory a poorly designed (or intentionally designed) license could require the associated webapp to also be open sourced Jul 10 18:07:42 well, I assume it's fine since they also offer paid support (we have community licenses for now) etc., but not gonna research it on my N900 right now :) Jul 10 18:07:57 :) Jul 10 18:08:16 http://elinux.org/Legal_Issues#GPL.2FLGPL_license_version_.28v3.29_issues Jul 10 18:09:37 that page is actually pretty good. see also the "Use of kernel header files in user-space" Jul 10 18:10:29 essentially the linux kernel developers have a carve-out designed to make sure that non GPL stuff is allowed to run w/o being tainted by GPL. But conceptually, w/o such carve outs, it could be pretty easy to consider "linux + all running applications on it" as a "derivative work" of "linux" Jul 10 18:10:54 and "derivative work" is where open source licenses typically bind their magic Jul 10 18:11:17 ah well, we're not distributing the systems per se, just using it as an alternative to vsphere et al Jul 10 18:11:46 so, v3 and agpl and friends were specifically to catch "apps in the cloud" Jul 10 18:11:57 such that "accessing an app" qualified as "distribution" Jul 10 18:12:17 if it's totally in house, you don't have a problem, because none of your people would exercise the clauses Jul 10 18:12:30 in my case, the applications i'm hosting aren't only for my employer Jul 10 18:12:36 which means i'm always slightly more cautious Jul 10 18:13:18 well, basically the only layer outsiders access is some nginx instance Jul 10 18:13:30 right Jul 10 18:13:44 it's just always very hard to tell what a given license does and how it impacts other things Jul 10 18:13:52 especially if one hasn't played w/ it in a while Jul 10 18:13:53 I suppose so Jul 10 18:14:00 i haven't looked at agpl/agpl3 in perhaps 10 years Jul 10 18:14:24 i do remember that some license near that family was designed to catch more things and have more cascades Jul 10 18:14:35 but i could easily be recalling the wrong license :) Jul 10 18:14:45 and i certainly don't remember the particulars :) Jul 10 18:19:40 http://gpl-violations.org/news/ is an interesting read... Jul 10 18:26:05 sadly, the mailing lists linked from http://gpl-violations.org/mailinglists/ near the bottom aren't around Jul 10 18:26:52 (at least the https links for browsing don't work) Jul 10 19:44:22 Hi! I'm new here. Can someone help me to run PyOBD on my n900 pls? Jul 10 19:47:52 if it's not packaged in maemo, try a chroot Jul 10 19:49:52 it is packaged, but i have dependencies issues Jul 10 19:51:08 and nokia droped down repos Jul 10 19:57:32 is there mirrors of repos? Jul 10 19:57:59 yes, but idk the shortcut for the help bot. Jul 10 19:58:21 maemo-repos Jul 10 19:58:25 ~maemo-repos Jul 10 19:58:26 it has been said that maemo-repos is http://wiki.maemo.org/Repository#List_of_Maemo_repositories Jul 10 20:04:53 DocScrutinizer05: hey, do you think it's possible to create a bridge over a loopback? :) Jul 10 20:07:44 kalin: i did use it for some time .. worked reasonably ok. would hang sometimes (although i think problem was largely my clone, cheap reader, as well as the car itself (early merc)) Jul 10 20:07:56 but yes, fix your repos first Jul 10 20:11:45 how Jul 10 20:12:33 you have been given links in KotCzarny's last comment .. Jul 10 20:14:11 there are also one click repos there for slower learners Jul 10 20:19:04 i will try it Jul 10 20:19:12 DocScrutinizer05: so... i think my problem is that my containers are systemd Jul 10 20:19:16 i wonder if they "upgraded" Jul 10 20:19:29 systemd in containers is really not friendly to linux kernel 2.6 Jul 10 20:20:21 then deinstall systemd Jul 10 20:20:25 yep Jul 10 20:20:29 lxc doesn't require it Jul 10 20:20:32 * timeless considers moving from debian to devuan Jul 10 20:20:36 neither does a chroot Jul 10 20:23:29 * timeless nods Jul 10 20:23:35 * timeless starts the painful process Jul 10 20:24:33 joy https://www.irccloud.com/pastebin/Wrwp6NMa/ Jul 10 20:27:52 * timeless sighs https://www.irccloud.com/pastebin/kBSn0YIO/ Jul 10 20:44:59 I keep getting the same text every 30 mins or so Jul 10 20:45:11 anyone know wtf is up? Jul 10 20:45:26 is this a maemo bug, operator bug...? Jul 10 20:45:36 I've gotten it like 14 times Jul 10 20:45:55 text? sms? o r email? Jul 10 20:45:58 sms Jul 10 20:46:26 it's keeping me awake at this point Jul 10 20:46:41 I guess I can turn off vibration in silent mode... Jul 10 20:47:39 no idea. never had that, although i vaguely remeber someone else once had .. maybe on tmo. Jul 10 20:48:30 I'll turn it into offline mode if it persists. Jul 10 20:54:16 Wizzup: u can disable vibration in same place where u set device silent... Jul 10 20:54:38 I know. Jul 10 20:54:44 But I don't like doing that. Jul 10 20:55:16 then i don't get ur problem :) Jul 10 20:55:44 ... Jul 10 20:55:51 I keep getting the same sms Jul 10 20:55:55 like it's a new one Jul 10 20:55:57 yes Jul 10 20:55:59 every 30 mins Jul 10 20:56:08 that's a *bug* somewhere Jul 10 20:56:31 hmm maybe network fails to get "ok got it" reply from ur phine Jul 10 20:56:35 I don't want to make my phone unusable by removing vibrate from silent Jul 10 20:56:38 phone* Jul 10 20:56:45 could. other texts worked fine Jul 10 20:58:42 somewhat strange.. Jul 10 20:58:54 rebooted also Jul 10 20:58:59 still got it Jul 10 20:59:00 ... Jul 10 20:59:21 if you have another phone, i'd say try moving the sim across for a while Jul 10 20:59:26 i think max dbls was like 3 of same, but that was when i cross country via boat Jul 10 20:59:54 and they advertised roaming prices Jul 10 20:59:55 I am not in my home country, but still Jul 10 21:00:38 hmm im smelling vague coincidence Jul 10 21:03:03 well, time to sleep Jul 10 21:23:38 DocScrutinizer05: hurray! switching debian to devuan fixed my lxc hell Jul 10 21:23:48 (yes, i know, rhel6 is old) Jul 10 21:43:57 Wizzup: could be an issue on the other side (sender) Jul 10 21:44:12 well, actually, it only fixed 1/2 systems in lxc hell Jul 10 21:50:56 DocScrutinizer05: http://www.iomem.com/index.php?archives/4-Ethernet-Bridges-under-Linux.html&serendipity[entrypage]=all Jul 10 21:56:44 Wizzup: there's a bug with some SMS providers that reply an error code that makes a lot of phones resend the SMS. WATCH OUT!! it caused immense bill to the users that suffered it Jul 10 22:02:25 prolly the same works in inverse direction: the SMS delivery doesn't succeed 100% due to a bug in SMS provider not getting the "OK, received" feedback from phone, and thus trying again to deliver it. Luckily no provider will keep any SMS for delivery longer than one week (iirc, max 2 weeks when I'm wrong with 1), So even when this would go on, at least after a week the SMS gets discarded as undeliverable Jul 10 22:05:15 DocScrutinizer05: so... wanna see if we can get my lxc hell sorted? Jul 10 22:05:27 RHEL6 host; devuan "jessie" guest Jul 10 22:06:26 sounds good, I'm happy for every "devuan fixed it for me" report :-) Jul 10 22:07:15 I already considered crossposting your "hooray" to #devuan, but seems you reported there as well :-) Jul 10 22:07:32 it's based on which debian release by the way? Jul 10 22:12:29 hm? Jul 10 22:12:44 devuan? jessie Jul 10 22:13:59 ok Jul 10 22:14:58 DocScrutinizer05: the repos still claim to be jessie about half the time Jul 10 22:15:35 so, here's a working system: https://www.irccloud.com/pastebin/2XxkRl74/ Jul 10 22:15:45 and here's the broken system https://www.irccloud.com/pastebin/C0QsIB2J/ Jul 10 22:16:00 timeless: hmm? Jul 10 22:17:04 * timeless tries to figure out where virbr0-nic came frome Jul 10 22:17:59 -e Jul 10 22:18:34 https://devuan.org/os/filenaming Jul 10 22:19:08 eureka https://www.irccloud.com/pastebin/JJHJ7zue/ Jul 10 22:19:15 working https://www.irccloud.com/pastebin/zWcQVElg/ Jul 10 22:19:16 err https://devuan.org/os/releases Jul 10 22:19:33 please post that in #devuan :-) Jul 10 22:19:41 once i get it to do something useful Jul 10 22:20:27 comparing two fairly large systems to figure out how they differ is painful Jul 10 22:21:43 O_o semantic versioning for an OS distribution? How does that even work? Jul 10 22:42:22 Maxdamantus: whichi part? Jul 10 22:42:44 0.10.2 / 62 / el6 / x86_64 Jul 10 22:42:50 the arch is usually irrelevant Jul 10 22:43:13 but in theory if you had a choice between an el6 and an el7 and your system wasn't rejecting one/the other, you'd probably get the newer one Jul 10 22:43:27 the 62 is roughly a "packager" patch number Jul 10 22:43:45 0.10.2 is the upstream version Jul 10 22:44:48 anyway... looks like i get to migrate pydio from sqlite to mysql Jul 10 22:44:53 which should be loads of fun Jul 10 22:45:01 pydio didn't provide a migration script Jul 10 22:45:15 so i get to perform a "dump" and an "import" and then massage all the config Jul 10 22:45:19 * timeless grumbles Jul 10 22:47:57 Well, semantic versioning has to do with constraints on version WRT API Jul 10 22:49:35 I guess it could kind of make sene for an OS if you know all software that has changed versions has similar constraints on their versions, but that seems like it'd be really hard to reason about. Jul 10 22:50:23 s/\/sense/ Jul 10 22:52:03 also, haven't really looked into it, but it seems weird to have such precise versioning for a Debian derivitive, since Debian itself is usually considered a "rolling release" since it doesn't seem to have those. Jul 10 22:53:01 Hm. Are you saying that the semantic version only applies to the package manager and not all the packages included? Jul 10 22:53:38 So `0.10.62` might have one of various versions of libpng? Jul 10 23:05:56 ouch **** ENDING LOGGING AT Tue Jul 11 03:00:03 2017