| *** sgw has joined #starlingx | 03:38 | |
| *** Samiam1999 has joined #starlingx | 06:47 | |
| *** Samiam1999DTP has quit IRC | 06:48 | |
| *** jaewook_oh has joined #starlingx | 08:09 | |
| *** jaewook_oh has quit IRC | 08:09 | |
| *** ijolliffe has joined #starlingx | 13:52 | |
| *** sgw has quit IRC | 13:53 | |
| *** abailey has joined #starlingx | 13:56 | |
| slittle1 | good morning | 14:40 |
|---|---|---|
| *** sgw has joined #starlingx | 14:56 | |
| sgw | Morning all | 14:57 |
| slittle1 | I encourage everyone to review the Layered Build doc at https://review.opendev.org/#/c/705092. | 15:32 |
| sgw | I will try to get some additional input from the BLR team, | 15:33 |
| sgw | scott did you see my "quick start" notes, not sure if there is anything in there that might help with clarifications | 15:34 |
| slittle1 | I saw the thread, yes | 15:36 |
| slittle1 | I'll go through it | 15:37 |
| *** TheJulia has joined #starlingx | 15:38 | |
| sgw | scott did you see my "quick start" notes, not sure if there is anything in there that might help with clarifications manifest change now | 15:39 |
| TheJulia | Question, I noticed starlingx has references to ironic-python-agent images built on coreos in the documentation, specifically also pointing at Stein versions. Ironic only intentionally maintains a x-1,x,x+1 style support for images, so I'm a bit worried about the stein reference and I feel like the links should be pointing to more recent images since the CoreOS IPA ramdisks are no longer being built or maintained. | 15:40 |
| *** TxGirlGeek has joined #starlingx | 15:50 | |
| slittle1 | Is 'repo' giving folks an problems this morning ? | 16:20 |
| slittle1 | it looks like they pushed repo version 2 into their stable branch. You'll likely pick it up with your next 'repo init' or 'repo sync'. It demands git 2.10.2 or better, which isn't available on Centos 7 | 16:22 |
| slittle1 | my initial workaround... for folks that are hit by this ... | 16:23 |
| slittle1 | cd .repo/repo | 16:23 |
| slittle1 | git checkout repo-1 | 16:23 |
| slittle1 | cd - | 16:23 |
| slittle1 | then re-run the repo command that failed | 16:23 |
| slittle1 | Ok, there is a command line option | 16:50 |
| slittle1 | repo init --repo-branch=repo-1 ... | 16:50 |
| sgw | Maybe send email for this also, not everyone is here , maybe a docs note | 17:07 |
| *** ijolliffe has quit IRC | 22:14 | |
| * sgw wonders if anyone can tell me about pxeboot wait look and where that code is? | 22:17 | |
| dpenney_ | sgw: not sure what you're asking | 22:19 |
| sgw | dpenney_: When an unconfigured node boots fior the first time, it goes into a wait loop | 22:32 |
| sgw | and the node shows up under system host-list | 22:33 |
| dpenney_ | ah, gotcha. just a sec | 22:33 |
| sgw | How does sysinv find out about the unconfigured node? | 22:33 |
| sgw | PXE provides a mac address DHCP might be involved, is a modified grub shot over first? | 22:34 |
| dpenney_ | the node sends a DHCP request, and the response is the default boot, ie. https://opendev.org/starlingx/metal/src/branch/master/installer/pxe-network-installer/pxe-network-installer/default | 22:35 |
| dpenney_ | sysinv gets a notification from dnsmasq of the new node | 22:35 |
| dpenney_ | when the node is configured, a pxeboot config file is generated for that node, with appropriate parameters | 22:35 |
| sgw | that's after it's booted, how does the host-list get the info before the personality is set via host-modify? | 22:35 |
| dpenney_ | sysinv gets a notification from dnsmasq of the new node and adds the unconfigured host to its database | 22:36 |
| sgw | Oh dnsmasq of the host | 22:36 |
| sgw | host meaning controller-0 | 22:37 |
| dpenney_ | dnsmasq is the service running on the active controller | 22:37 |
| sgw | got it. | 22:37 |
| dpenney_ | that provides DHCP responses, as well as tftp access | 22:37 |
| sgw | right, yeah I understand that now, thanks that was the magic I was missing, also just to confirm, the tftp access is whitelisted to that MAC address also? | 22:39 |
| dpenney_ | dnsmasq listens only to the mgmt network for tftp/dhcp requests | 22:40 |
| sgw | I guess I am asking about the response to the pxeboot after the node has it's personality set, is it specified to that MAC? For example if another node comes up and the first one is turned off will the second node get the first node's pxeboot load? | 22:42 |
| dpenney_ | no, the generated pxeboot config files are specific to the MAC | 22:42 |
| sgw | I am guessing it won't just trying to understand what your doing to prevent that, | 22:42 |
| sgw | That was the MAC whilelist question above, perfect, that confirms part of what I thought. | 22:43 |
| dpenney_ | Files are in /pxeboot/pxelinux.cfg/ on the controller | 22:43 |
| sgw | Yes, got it all sorted now, was trying to explain this to someone else and this was not an area I had dived into in the past other than installing nodes | 22:47 |
| sgw | Thanks for your time. | 22:47 |
| dpenney_ | you can see the tftp logs in /var/log/daemon.log. It first requests a config file with its MAC as the name, then funnels down to "default" if it doesn't find the MAC-specific file | 22:48 |
| sgw | Beer is owed at some point in the future! | 22:48 |
| dpenney_ | :) | 22:49 |
| dpenney_ | I'm happy to get a question I can answer ;) | 22:49 |
| *** abailey has quit IRC | 23:16 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!