[00:59] tych0: i'd file a bug and provide the version you are using. But otherwise i woudln't really know. I have not touched a UCS in a long time [00:59] roaksoax: ok, will do. we might send a patch in the future, but right now we have some other stuff to deal with [00:59] thanks for your help [01:41] Bug #1739262 changed: [2.4, snap] Snap fails to build in bionic due to "file paths in common" between parts [13:46] roaksoax: so to use manual provisioning, i hit 'commission' and then what, immediately reboot the node? [13:57] hallyn: yes, you need to manually turn on/reboot the machine [17:09] roaksoax: if i restore prod database to a bare metal system to test, do i need to change anything in the DB? [17:15] xygnal: you shouldn't need to. https://docs.maas.io/2.3/en/manage-backup [17:16] xygnal: but maas-regiond.conf should be able to log into the db [17:34] Bug #1749757 opened: [2.4] Spurious errors on machine configuration page for libvirt machines [18:06] roaksoax: yes, my question is really how do i know when i should reboot it. it doesn't tell me to in 'even'ts after i say 'commision', [18:06] and when i did it last night after hitting 'comissions' i didn't get any further events. [18:06] do i need to time it somehow? [18:19] hallyn: so manual power type just basically means that you will turn on the machines. so when you hit "commission" you need to turn on your machine and make sure it can/will pxe boot off MAAS [18:19] hallyn: only then is when you will see events of the machine pxe booting [18:23] ok - so while it may seem obvious to you, it would be helpful i think, when someone hits commision on a manual node, to pop up an event saying "pelase reboot the system" [18:24] so after that first reboot, it should ask me if i need to do anything? [18:25] hallyn: sure, could you please file a bug for that ? [18:26] will do thx [18:50] well no i won't do until i have a mouse or a real browser where i don't have to tab through hundreds of links to get to 'report a bug' i guess [19:25] Bug #1674959 changed: [UI] Close Add Pod button is redundant on pods page [19:34] Bug #1749782 opened: [2.4] Machine incorrectly shows "Locked" after failed commissioning [20:16] roaksoax: https://bugs.launchpad.net/maas/+bug/1749794 [20:16] trying it out now [20:17] Bug #1749794 opened: help a manual user out [20:21] Hi I followed the MAAS KVM pod tutorial to create the virsh pod on maas itself. I composed three machines, they don't show up on maas web ui, but they are listed as shut off in the virsh. I am on MAAS 2.3. is this a known issue? [20:23] Node commissioning failure - 'cloudinit' running modules for config :( [20:25] catbus, I connected MAAS Pod to KVM, setup a bridge interface for virsh, and added VMs one at a time, set them to PXE boot first, and booted them. MAAS was able to commission them like any other machine [20:25] this was in version 2.2 [20:26] maas was able to connect to my virsh and detect the status of all machines [20:27] i think if the virsh is on the same node as MAAS, you need to set the address to qemu:///system [20:28] yosefrow__: I thought when I compose them via maas web ui, they will be automatically commissioned, so I expect maas to take care of the part of creating vm, setting it to pxe boot. [20:29] catbus, I didnt use this feature. But MAAS had no problem automatically creating machines for me when juju requested them [20:29] catbus, try this command: sudo -u maas virsh -c qemu:///system list --all [20:29] what do you get? [20:30] it lists three machines in shutoff state. [20:30] hmmm [20:30] and they dont show up in the ui >> [20:30] they are set to pxe boot. [20:30] did you click the pod section ? [20:30] no they don't. [20:31] or are you looking in the main node section [20:31] yes I look at the pod section. it says 0 composed machines. [20:32] did you try restarting the maas service? [20:34] ah, the image is still being imported. nm. let me wait till it's done and I will try again. [20:35] yosefrow__: all services are running fine, so unless there is any information that suggests something wrong with any of the services, I'd have them not touched. [20:36] catbus, fair enough [20:36] catbus, I was just curious if maas had been restarted since the time you added the pod [20:37] No, it's just freshly built. [20:37] ok, well I'm out of idea for now then [20:37] I hope you figure it out [20:38] I'll be watching to see what the solution is [20:38] there are probably many people here with far more experience than me in MAAS [20:45] roaksoax: so a bunch of commisioning steps go fine, but then it says 'failed commisioning' with the only failed event being 'Node commissioning failure - 'cloudinit' running modules for config' [20:45] If that doesn't ring a bell for you then i'll just say f it and wipe maas and setup a *)$&%)($*%)( pxe server [20:56] hallyn: all commissioning & testing scripts pass ? [20:56] hallyn: that would normally mean that something failed at the end of cloud-init processing and told maas about it [20:57] hallyn: check /var/log/maas/rsyslog///messages [20:57] and see if you can see the issue there? [20:58] checking, [20:59] Feb 15 12:45:40 atom-lab-10 cloud-init[2067]: FileNotFoundError: [Errno 2] No such file or directory: 'ipmitool' [20:59] hallyn: so that probably means that it cannot access the ubuntu archives to download software [20:59] or whatever archive its configured [20:59] hm, [21:00] yeah it looks like it's having a bitch of a time getting out to the real world. [21:02] it seems to know about the proxy to use, [21:03] i'd like to set a specific ip address (that may be needed) but apparently it has to be commissioned first. [21:03] that's idiotic [21:04] hallyn: not at all, commissioning is hardware discovery/network/interface discovery/etc [21:04] once the machine has been commissioned [21:04] it can be used for other purposes, like deployment [21:04] if the machine commissions, you are saying its ready for prime time [21:05] if it doesn't then there's a problem [21:06] hm? no i'm saying i think the switch may enforce a specific ip address for the host [21:06] but if that were the case it couldn't be reporting back the hardware list, so no [21:06] i'm at a loss [21:07] when the os comes up for commisioning, can i log in? [21:07] user ubuntu password something? [21:07] just wanna browse the network config [21:11] roaksoax: ignorant question: are the power mgmt api calls done by scripts? can i edit them live? [21:25] hallyn: yes you can, there's an option where you can enable SSH and will import your ssh key and prevent it from powering off [21:25] hallyn: and no, they are python code [21:32] roaksoax: good enough :) [21:32] where are they? [21:32] i should have said 'interpreted', not 'shell' :) [21:35] hallyn: /usr/lib/python3/dist-packages/provisioningserver/drivers/power/ [21:37] roaksoax: thanks [21:37] so, [21:37] i changed the reserved dhcp range (made it smaller), now things aren't getting dhcp offers [21:37] known bug or did i do it wrong? [21:41] Bug #1749812 opened: [2.4, UI] Device Discovery page needs fixing [21:59] hallyn: is dhcpd running ? [22:05] yeah - the table was full :( [22:05] roaksoax: thanks, it looks like it needs to straighten a few things out, i'll try again once that's done.