| *** tpb has joined #timvideos | 00:00 | |
| *** tpb has joined #timvideos | 00:54 | |
| *** puck``` has joined #timvideos | 01:00 | |
| *** tumbleweed has quit IRC | 01:01 | |
| *** puck`` has quit IRC | 01:01 | |
| *** tumbleweed has joined #timvideos | 01:01 | |
| *** tumbleweed has quit IRC | 01:01 | |
| *** tumbleweed has joined #timvideos | 01:01 | |
| *** rohitksingh_work has joined #timvideos | 03:19 | |
| CarlFK | tumbleweed: not sure I have a PR for the usb stick stuff | 03:27 |
|---|---|---|
| CarlFK | https://github.com/CarlFK/video-stack-deploy/blob/usbstick/scripts/mk_usb_installer.sh | 03:27 |
| tpb | Title: video-stack-deploy/mk_usb_installer.sh at usbstick · CarlFK/video-stack-deploy · GitHub (at github.com) | 03:27 |
| tumbleweed | not sure what I'm looking at | 03:28 |
| CarlFK | I scripted the instructions from the dc vid readme on how to make a usb stick installer | 03:30 |
| CarlFK | and tried to roll in lots of options: debian/ubuntu , and where the files would be hosted (preseed, late_command) | 03:31 |
| CarlFK | one thing I concluded - everyone expects there to be a connection to ... something internet like | 03:31 |
| CarlFK | for the deb repos | 03:32 |
| tumbleweed | it's to be able to avoid that, but more work | 03:32 |
| CarlFK | I don't think anyone is expecting a stand alone install, including the ansible stuff | 03:32 |
| tumbleweed | yeah, I mean, that script looks reasonable enough | 03:32 |
| tumbleweed | but what do you want me to review? | 03:32 |
| CarlFK | start with high level - is what I am doing too "noisy" | 03:33 |
| CarlFK | script looks reasonable - is what I was hoping for. | 03:33 |
| CarlFK | script scares me - is what I was expecting | 03:33 |
| tumbleweed | I mean, bits are amusing, but nothing scary | 03:34 |
| CarlFK | there are a bunch of ways of boot strapping the 'thing' - like 1. install os from cd, log in, apt install and setup keys.. then ansible either --local or remote | 03:39 |
| tumbleweed | yeah | 03:40 |
| CarlFK | what is here: clone the repo to your personal box, mk_usb.. put usb stick in new, boot, enter host name... | 03:40 |
| CarlFK | hostname is in ansible hosts somewhere.. and it can become a pxe server | 03:40 |
| tumbleweed | that's what we want for debconf | 03:40 |
| CarlFK | or a vocto box, or veyepar... | 03:40 |
| tumbleweed | (err before you mentioned pxe server) | 03:41 |
| tumbleweed | right | 03:41 |
| CarlFK | right right | 03:41 |
| CarlFK | do you see how I am hosting the preseed file? | 03:42 |
| CarlFK | thats ... goofy ;) | 03:42 |
| CarlFK | https://github.com/CarlFK/video-stack-deploy/blob/usbstick/scripts/mk_usb_installer.sh#L106 ./http_server.sh | 03:42 |
| tpb | Title: video-stack-deploy/mk_usb_installer.sh at usbstick · CarlFK/video-stack-deploy · GitHub (at github.com) | 03:42 |
| CarlFK | https://github.com/CarlFK/video-stack-deploy/blob/usbstick/scripts/http_server.sh python -m SimpleHTTPServer | 03:43 |
| tpb | Title: video-stack-deploy/http_server.sh at usbstick · CarlFK/video-stack-deploy · GitHub (at github.com) | 03:43 |
| tumbleweed | not that goofy, we did something similar at lca | 03:43 |
| CarlFK | good. another one that I wasn't sure it it was too kludgy | 03:44 |
| tumbleweed | although you're copying it to the flash drive too? | 03:44 |
| CarlFK | yeah - that was left over from the readme | 03:44 |
| CarlFK | and thinking I could put all the things on the usb stick | 03:44 |
| tumbleweed | that's probably what one would do once the preseed file is stable | 03:44 |
| CarlFK | and then realized that is too much work | 03:44 |
| CarlFK | im not sure it matters given you need to host something somewhere... like ansible's inventory files, and the whole playbook | 03:45 |
| tumbleweed | yep | 03:46 |
| CarlFK | so as long as that is provided for, may as well host the preseed.cfg there too | 03:46 |
| tumbleweed | from the point of view of the ansible stuff, this is all implementation details | 03:46 |
| tumbleweed | there are may ways of skinning the cat, we suggest picking something that works well for you | 03:47 |
| CarlFK | but... copy it onto the stick, and now you have a copy you can mess with for a 'odd box' so seems like an ok thing to have floating around | 03:47 |
| tumbleweed | screws you when you want to customize it, of course | 03:47 |
| CarlFK | depens... | 03:48 |
| CarlFK | um.. you know what the sharp edges are ;) | 03:48 |
| CarlFK | I am still fuzzy on where I think "production" files should live. a git repo for inventory/ makes me fidgit for some reason. not sure why. | 03:50 |
| tumbleweed | version-controlling your config is usually a good idea | 03:50 |
| tumbleweed | and makes it friendlier for multiple people to hack on | 03:51 |
| CarlFK | makes it yucky for testing | 03:51 |
| CarlFK | ive got like 30 commits as I try this and try that | 03:51 |
| CarlFK | but anyway.. I can live with hit | 03:51 |
| CarlFK | https://github.com/CarlFK/video-stack-deploy/blob/usbstick/scripts/ec/early_command.sh is empty. | 03:52 |
| tpb | Title: video-stack-deploy/early_command.sh at usbstick · CarlFK/video-stack-deploy · GitHub (at github.com) | 03:52 |
| CarlFK | on my box, I have a few scritps that help debug failed installs | 03:52 |
| tumbleweed | I tend to hack locally, and when it's looking good then I commit. That helps reduce the noise | 03:53 |
| CarlFK | like bring up sshd in the isntaller and scp syslog and stuff off to another box | 03:53 |
| tumbleweed | should I say, when it's looking good, then I try to break down what I've done into explainable commits | 03:53 |
| CarlFK | comit noise - I am talking about: test box foo as dhcp, test foo as vocto, foo as opsis... | 03:54 |
| tumbleweed | right, but that's a mission :) | 03:54 |
| CarlFK | those are ansible inventory changes - that currently live in xfxf's git repo. | 03:55 |
| tumbleweed | why don't you just have vocto, dhp, and opsis as hostnames? | 03:56 |
| tumbleweed | then you don't need to change it | 03:56 |
| CarlFK | anyway, as long as it worked for me and you are ok with it. good. | 03:56 |
| CarlFK | hmm | 03:56 |
| CarlFK | because I have always pinned hostnames to hardware | 03:56 |
| CarlFK | hardware has a sticker on it with the hostname writtin with a shapie. | 03:56 |
| CarlFK | but I see your point | 03:57 |
| tumbleweed | :) | 03:57 |
| CarlFK | maybe. I'll file that for next time | 03:57 |
| CarlFK | anway, early_command.sh gets loaded/run from preseed.cfg but as is does nothing | 03:58 |
| tumbleweed | yeah, I see | 03:59 |
| tumbleweed | what sort of things do you find early_command useful for? | 03:59 |
| CarlFK | the stuff I do .. is yucky and may have keys and stuff. and if you have never wanted it before, for now leave it out - less for us to review | 03:59 |
| CarlFK | sec.. it is in my old veyepar pxe repo... | 03:59 |
| CarlFK | https://github.com/CarlFK/veyepar/blob/master/setup/nodes/pxe/shaz/var/www/ec/early_command.sh | 04:01 |
| tpb | Title: veyepar/early_command.sh at master · CarlFK/veyepar · GitHub (at github.com) | 04:01 |
| CarlFK | I thought there was a 3rd thing - maybe 3. install scritps to bring up 1. sshd and 2. gather stuff and copy to some other hardcoded box | 04:02 |
| tumbleweed | an sshd, by the look of it | 04:02 |
| tumbleweed | for debugging? | 04:02 |
| CarlFK | yes | 04:02 |
| CarlFK | when the isntaller errors, alt-f2, tail syslog.. and 90% of the time "I can't see what I need... /tmp/misc/cpl.sh a" | 04:03 |
| CarlFK | and now "everything" is on a real box with real text tools | 04:03 |
| tumbleweed | :) | 04:03 |
| CarlFK | er | 04:04 |
| CarlFK | thats the cpl.sh | 04:04 |
| CarlFK | but sshd - same thing. box is often not easy to get to, type on, see monitor ... ssh from my laptop is much nicer | 04:05 |
| CarlFK | oh yeah, I can cut/paste stuff at cjwatson in #u-installer :p | 04:05 |
| tumbleweed | lol | 04:05 |
| CarlFK | https://github.com/CarlFK/video-stack-deploy/blob/usbstick/scripts/lc/late_command.sh#L37 echo "none /dev/shm tmpfs rw,nosuid,nodev,noexec,noauto 0 0" >> /etc/fstab | 04:06 |
| tpb | Title: video-stack-deploy/late_command.sh at usbstick · CarlFK/video-stack-deploy · GitHub (at github.com) | 04:06 |
| tumbleweed | I noticed that, what's that all about? | 04:07 |
| CarlFK | read the 4 #comments above | 04:07 |
| tumbleweed | it doesn't really answer that :P | 04:07 |
| tumbleweed | does d-i not have a /dev/shm? | 04:07 |
| CarlFK | I guess not | 04:08 |
| tumbleweed | we didn't have this problem at lca | 04:08 |
| tumbleweed | but that would have been an older d-i, and an ubuntuised one | 04:08 |
| CarlFK | there were a few packages that seemed to trip it | 04:09 |
| tumbleweed | if you're worried about the fstab entry, just mount it, rather than adding an fstab line | 04:09 |
| CarlFK | I couldn't figure out the mount command | 04:10 |
| tumbleweed | but what unmounts it? | 04:10 |
| CarlFK | it reboots ;) | 04:10 |
| tumbleweed | happily? | 04:10 |
| CarlFK | i think so.. what do you mean? | 04:10 |
| tumbleweed | I would have expected it to want to unmount /target before rebooting | 04:11 |
| tumbleweed | but it can't do that if /target/dev/shm is still mounted | 04:11 |
| CarlFK | hmm | 04:11 |
| tumbleweed | btw mount -t tmpfs none /dev/shm should work | 04:11 |
| CarlFK | maybe the shutdown unmounts it? | 04:12 |
| tumbleweed | I thought in-target unmounted several bind mounts when leaving the chroot | 04:12 |
| tumbleweed | but that's just a vague memory | 04:12 |
| tumbleweed | I can play with this if you want | 04:12 |
| tumbleweed | but it's probably more useful to talk about ansible, I mean, that's what we're here for right? | 04:13 |
| CarlFK | 39 # mount -t tmpfs none /dev/shm | 04:13 |
| CarlFK | added for later | 04:13 |
| CarlFK | yeah | 04:13 |
| CarlFK | well, I was more worried about the scaffolding? to have something to ansible... but sounds like you are OK with it | 04:14 |
| tumbleweed | well, this stuff matters when you're trying to image 100 machines | 04:15 |
| tumbleweed | but it's not that important for development of the ansible repo | 04:15 |
| tumbleweed | because you just directly run ansible against something | 04:15 |
| CarlFK | k | 04:15 |
| CarlFK | ok, so PRs - only 2? I guess we merged the 3rd | 04:16 |
| CarlFK | https://github.com/CarlFK/video-stack-deploy/pull/3 | 04:16 |
| tpb | Title: Pxe toucheup by CarlFK · Pull Request #3 · CarlFK/video-stack-deploy · GitHub (at github.com) | 04:16 |
| CarlFK | this is about the same issues as the usb stick stuff | 04:17 |
| tumbleweed | I don't know what that means | 04:20 |
| tumbleweed | there's a lot in here, I don't really know how to pull this apart | 04:20 |
| tumbleweed | what's the "distro" variable? | 04:20 |
| CarlFK | debian/ubuntu | 04:20 |
| tumbleweed | but ansible_distribution gives you that | 04:21 |
| tumbleweed | ah, I guess not for the pxe config | 04:22 |
| tumbleweed | but you're using it everywhere | 04:22 |
| CarlFK | im trying to remember... | 04:23 |
| tumbleweed | lol, yet another serial console implementation | 04:24 |
| tumbleweed | we support 4 now :P | 04:24 |
| CarlFK | lol | 04:25 |
| CarlFK | what comes to mind is wanting to be able to have 1 pxe server config that I can use to create both debian and ubuntu boxes | 04:25 |
| tumbleweed | that's going to be a very rare use-case | 04:25 |
| CarlFK | but that doesn't really justify using distro | 04:25 |
| CarlFK | except I account for 1/2 the use cases ;) | 04:25 |
| tumbleweed | true, but it adds some complexity | 04:26 |
| tumbleweed | anyway, that explains it in the ansible template, but not anywhere else | 04:26 |
| tumbleweed | and I don't see a default / anything that documents what itis | 04:26 |
| tumbleweed | err in the pxe template | 04:26 |
| CarlFK | oh yeah.. I realized at some point I wasn't doing 'that' | 04:27 |
| *** tpb has joined #timvideos | 05:22 | |
| CarlFK | thats why I want all the late stuff in its own dir, seperate from the early stuff | 05:22 |
| CarlFK | I think late is where I put the bm blobs | 05:23 |
| CarlFK | hmm, but now ansible... | 05:24 |
| CarlFK | I think I remember this conversation now | 05:24 |
| tumbleweed | that makes sense for *your* lc directory, it doesn't for this, if this is going into debconf's one | 05:27 |
| tumbleweed | because then you'll have two lc directories, that come from different places | 05:27 |
| CarlFK | two - one for usb, 2nd for pxe? | 05:29 |
| tumbleweed | the one in this repo, with one script in it | 05:29 |
| tumbleweed | and the one in your veyepar repo | 05:29 |
| CarlFK | ah - my veyepar one is 'dead' | 05:29 |
| CarlFK | I was thinking of it like the early - where I have stuff I may toss in there on the fly, but don't want to clutter the repo | 05:30 |
| CarlFK | but Im not sure thats needed for late given all that stuff is now handled by ansible | 05:31 |
| tumbleweed | yeah | 05:31 |
| tumbleweed | the preparing for clutter is clutter in its own right | 05:31 |
| CarlFK | lol - true | 05:31 |
| *** tpb has joined #timvideos | 05:32 | |
| tumbleweed | there are some good changes in there, some things that I don't think you're using, and some things I think are blatently wrong | 05:32 |
| tumbleweed | (and lots to nitpick about, but meh)( | 05:33 |
| CarlFK | lol | 05:33 |
| CarlFK | I think you are mis understanding the dns hosts thing. but I miht have gotten it wrong too | 05:33 |
| tumbleweed | no-hosts means it won't read /etc/hosts | 05:34 |
| CarlFK | the intent is to provide the stuff you want | 05:34 |
| tumbleweed | so the DNS server won't know the IP:hostnames of any of the machines | 05:34 |
| CarlFK | I think I see where I am confused | 05:43 |
| CarlFK | https://github.com/CarlFK/video-stack-deploy/blob/pxe-toucheup/roles/staticips/templates/hosts.j2 {{ host.ip }} {{ host.hostname }}.{{ domain }} {{ host.hostname }} | 05:43 |
| tpb | Title: video-stack-deploy/hosts.j2 at pxe-toucheup · CarlFK/video-stack-deploy · GitHub (at github.com) | 05:43 |
| CarlFK | https://github.com/CarlFK/video-stack-deploy/blob/pxe-toucheup/roles/dhcp-server/templates/machines.conf.j2 dhcp-host={{ host.mac }},{{ host.ip }},{{ host.hostname }} | 05:44 |
| tpb | Title: video-stack-deploy/machines.conf.j2 at pxe-toucheup · CarlFK/video-stack-deploy · GitHub (at github.com) | 05:44 |
| tumbleweed | dhcp-host is not host-record | 05:44 |
| CarlFK | right. | 05:45 |
| CarlFK | I think I didn't notice those were two different things | 05:46 |
| *** puck``` is now known as puck`` | 05:47 | |
| CarlFK | huh - running fatresize on the ubuntu boot.img .. breaks it. it won't boot now | 05:55 |
| CarlFK | 1am - I need to get to bed. | 05:57 |
| CarlFK | I'll take a shot at addressing the issues tomorrow | 05:58 |
| *** tpb has joined #timvideos | 06:39 | |
| *** cr1901_modern has quit IRC | 07:44 | |
| *** sb0 has quit IRC | 07:49 | |
| *** sb0 has joined #timvideos | 08:30 | |
| *** rohitksingh_work has quit IRC | 09:06 | |
| *** rohitksingh_work has joined #timvideos | 09:07 | |
| *** rohitksingh_wor1 has joined #timvideos | 10:28 | |
| *** rohitksingh_work has quit IRC | 10:28 | |
| *** Ishan_Bansal has joined #timvideos | 11:58 | |
| *** rohitksingh_wor1 has quit IRC | 12:36 | |
| *** Ishan_Bansal has quit IRC | 13:21 | |
| *** sb0 has quit IRC | 13:58 | |
| *** sb0 has joined #timvideos | 16:18 | |
| *** sb0 has quit IRC | 16:32 | |
| *** sb0 has joined #timvideos | 16:36 | |
| *** sb0 has quit IRC | 16:44 | |
| *** sb0 has joined #timvideos | 16:49 | |
| *** rohitksingh has joined #timvideos | 16:56 | |
| *** sb0 has quit IRC | 17:00 | |
| *** sb0 has joined #timvideos | 17:05 | |
| *** rohitksingh has quit IRC | 18:13 | |
| *** cr1901_modern has joined #timvideos | 20:14 | |
| *** puck`` is now known as puck | 23:00 | |
| *** CarlFK has quit IRC | 23:53 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!