14:18:10 #startmeeting Fedora Base Design Working Group (2015-06-29) 14:18:10 Meeting started Mon Jun 29 14:18:10 2015 UTC. The chair is haraldh. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:18:10 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:18:14 #meetingname Fedora Base Design Working Group 14:18:14 The meeting name has been set to 'fedora_base_design_working_group' 14:18:18 #chair haraldh msekleta jreznik dgilmore vpavlin masta dwalsh 14:18:18 Current chairs: dgilmore dwalsh haraldh jreznik masta msekleta vpavlin 14:18:21 hi 14:18:34 hi 14:18:41 hi 14:21:33 anybody else? 14:21:46 ping dgilmore dwalsh haraldh masta msekleta vpavlin 14:22:06 ok, msekleta is travelling 14:22:57 ping dgilmore dwalsh masta vpavlin 14:25:54 #topic Summer Time .. it's hot 14:26:08 everybody on the beach, it seems 14:27:59 still no excuse, you can have your laptop there. :) 14:32:50 * masta looks in 14:32:56 hello folks =) 14:34:32 uuuhh.. hi masta 14:35:31 #topic Accept new members 14:35:53 so, jreznik and masta , do you think we can add lnykryn to our team? 14:35:59 Do you know him? 14:36:12 Want to ask him some questions? 14:36:22 hi lnykryn 14:36:26 hi 14:36:27 team == Base WG 14:36:49 haraldh: no, have not had the pleasure to know him yet. =) 14:37:15 lnykryn, maintains systemd and initrscripts 14:37:18 maybe small introdution, I maintain some init related stuff and want to get rid of some legacy parts of the base system 14:37:24 so he is as base as it can get :) 14:37:52 hehe... seems that way. 14:39:07 ofc, additional non-RH members would be preferred... but ... well seems like they don't grow on trees 14:39:22 hehe 14:39:43 so I *was* to be the token no-RH person, but then RH hired me. =( 14:40:04 jreznik, still there? 14:40:07 so anyways... 14:40:35 So, it seems we have a +1 from msekleta, my +1 14:40:45 lnykryn: after doing some quick google stalking on you, I've decided you would be a good fit for the team. 14:40:55 :) thanks 14:40:58 and if masta and jreznik have a +1, too, we would be good to go :) 14:40:58 +1 lnykryn 14:41:17 vpavlin has also voted by mail a couple of weeks ago 14:41:22 ah, true 14:41:47 well, then congrats and welcome lnykryn :) 14:41:56 thank you for accepting me 14:41:57 haraldh: disrupted by phone call :( 14:42:05 give a second 14:42:09 Yeah! welcome. 14:42:15 lnykryn: the systemd guy? 14:42:41 +1 lnykryn 14:42:52 #topic readonly root - plan to get rid of fedora-readonly? drop in favor of ? 14:43:10 ok so that was my suggestion for a topic 14:43:20 So, one of the first suggestions by lnykryn were to deprecate fedora-readonly 14:43:33 in favor of the systemd mechanisms 14:43:56 currently we have nastay script which was part of rc.sysinit back then 14:44:15 has anybody tested fedora-readonly recently? does that even work anymore? 14:44:28 masta: it does sort of 14:44:30 masta, that is a really good question 14:44:39 we have tested it with rhel7 release 14:45:13 so at least back then it was working 14:45:40 only downside with systemd solution is, that we will break current /etc/rwtab.d/* "interface" 14:46:05 could we provide a generator for that? 14:46:16 I think so 14:46:33 on the other hand, we would have to deprecate it first for one RHEL release... right? 14:46:59 right 14:47:31 Well, then my proposal would be to try and write a generator 14:47:38 ok, will do 14:47:53 and add a big fat "deprecated" warning to the config files 14:47:59 and maybe documentation 14:48:21 or just leave it as it is with the deprecation warning 14:48:26 IDK 14:48:35 so rwtab.d is just a list of stateful mount, right? 14:48:43 yes/no 14:48:46 yes 14:48:49 it's also "copy" and mkdir 14:49:07 we can do the same things with tmpfiles 14:49:11 true 14:49:39 RW_MOUNT=/var/lib/stateless/writable 14:49:44 STATE_MOUNT=/var/lib/stateless/state 14:49:58 the statetab will be more interesting 14:50:23 because it needs some network volume, but it should be fine 14:50:29 in most cases 14:50:50 /etc/statetab /etc/statetab.d/* 14:51:16 ... ouch: 14:51:17 if mount | grep -q /var/lib/nfs/rpc_pipefs ; then 14:51:17 mount -t rpc_pipefs sunrpc /var/lib/nfs/rpc_pipefs 14:51:17 fi 14:51:28 eww 14:51:53 anyway, can be removed, with systemd dependencies, I guess 14:52:24 yeah, probably 14:52:29 I think so, I will look into it 14:53:25 maybe a documentation for the transition from the old config files to tmpfiles would be handy also 14:53:55 that would be nice to have 14:54:26 reminds me of /etc/fstab and systemd mount units 14:54:33 I am not sure, I would rather leave this for backward compability and create a proper stateless solution 14:54:38 fstab is much easier to write :) 14:54:45 I have to admit, it's been a while since I've played with these kinds of configuration. I'd like to refresh myself on these before deciding anything. 14:55:23 you mean s.th. like http://0pointer.net/blog/projects/stateless.html 14:55:32 yep 14:55:48 where the rpm packages themselves take care of stateless systems 14:56:12 to get there we would have to add some rpm spec magic first, I think 14:56:21 to populate /usr/share/factory 14:56:45 and create the correct tmpfiles config file 14:57:06 that sounds like a huge task 14:57:17 and get that to the packaging guidelines 14:57:37 well, we want that in the end anyway, don't we? 14:57:58 "Factory Reset" and stateless systems are just tooo nice to have 14:58:20 sure, but wrangling the package maintainers to that... 14:58:33 well, one package at a time 14:58:41 we don't have to force anybody 14:58:52 just have the most basic components ready 14:59:01 yep 14:59:10 agree. Having that in minimal system would be nice 14:59:26 so, the core of base should be ready 14:59:36 which would be systemd 14:59:44 and probably iSCSI, NFS 14:59:49 and network 15:00:11 and LVM 15:00:16 if we want iSCSI 15:00:58 network config via dhcp in the initramfs 15:01:12 hrm... 15:01:31 lvm config on the kernel cmdline.. PXE 15:01:35 or dhcp 15:01:40 if needed 15:01:53 dhcpv6 doesn't really have the capabilities for those nice things 15:02:33 any other autoconf daemons, which download config from the network? 15:03:42 hum... yeah, I guess some kind of network daemon would be needed 15:04:54 in minimal setup you can stick with networkd 15:08:32 ok, I'll make a list of components 15:09:13 #topic Open Floor 15:12:20 I've got nothing 15:12:40 ok, then ... have a nice week then! 15:12:40 me neither 15:12:45 thanks for hosting the meeting haraldh, I'm going to wonder away now. 15:12:59 #endmeeting