[00:08] hello everyone. I have a small problem that I need some guidance on. I'm using IPMI in shared mode and have the IP address set to 10.14.0.13/20. I have the subnet configured in MAAS. The node gets detected fine, but fails commissioning. If I configure IPMI for dedicated mode and give it it's own port on the switch, the node will commission fine. What am I missing? [00:09] This is the error I get: Failed to power on node - Power on for the node failed: Could not contact node's BMC: Connection timed out while performing power action. Check BMC configuration and connectivity and try again. [00:10] But the node does get powered on when trying to commission. === frankban|afk is now known as frankban [09:43] When I mark two interfaces on the node to create a bond, both of the interfaces disappear when I click "Save" then when I reload the page both of the interfaces show up again, but no bond is shown. [11:38] Okay, found the reason. The log says: [11:38] django.core.exceptions.ValidationError: {'vlan': ['A bond interface can only belong to an untagged VLAN.']} [11:38] Why can't the bond interface be in my VLAN 99 ? [11:39] Hi. Is thera anyone having storage problems with virtualized hardware? I'm using either VMware and VirtualBox and cannot commission disks. [11:41] miono: not sure exactly but is it possile that you need bond0.99 for tagged traffic? bond0 is always untagged in ubuntu I think [11:48] jacekn: So you're saying it's impossible to use bond0 as a trunked interface? [11:49] miono: it's possible but you end up with bond0 (untagged, native vlan) and bond0.XX (VLAN XX) [11:50] jacekn: hmm, okay [11:52] jacekn: I tried doing something similar to that. By setting the bond (made up of two interfaces) to untagged and not member of any subnet. And then I created a VLAN on that bond named bond0.99 (.1q-tag 99) [11:53] But then, when I did the deploy I got some strange in the post-install (I think), and couldn't ping the IP later. [11:53] miono: so that should work on the network level but I'm not sure about PXE/DHCP and the rest of it, maybe one of MAAS developers can help out with details [11:54] miono: for example as far as I know you can't PXE boot using tagged interface [11:55] jacekn: no, the PXE-radiobutton was on bond0 and not on bond0.999 [11:55] -9 [11:56] miono: ok so what IP address and in which subnet will MAAS assign to bond0? [11:57] jacekn: this is how the network configuration looks: http://lagga.se/network_config.png [11:57] jacekn: to my understanding it will assign 10.0.99.10 to bond0 [11:58] miono: nope, 10.0.99.10 is in VLAN99 and your server will PXE boot using native vlan for bond0 children, not sure what that VLAN ID is [11:58] jacekn: well, it does boot over PXE... [11:59] and the installation starts and it's working pretty good up until I get an error, I will have that error pretty soon. [11:59] And when it's booted over PXE and is installing it's using that IP-address (10.0.99.10) [12:00] miono: hmmm that's strange, my best bet is that something strange happens because MAAS can't make sense of your config [12:00] yeah, could be. [12:01] miono: so it boots from bond0 with 10.0.99.10 and then you tell it to use 10.0.99.10 with bond0.99 [12:01] well yeah, I guess. [12:01] I mean, I'm sure that it boots from "some interface" with 10.0.99.10 [12:03] and after the install, when it reboots I can't reach that interface anymore. [12:39] Very strange, when I created the bond through the maas-interface I chose active-backup as the bonding-mode. [12:40] But after deploying Ubuntu to the machine the interfaces-file has some options that seem related to LACP, which I find strange since that's not the bonding-mode I opted for. [17:44] Bug #1618543 changed: freeipmi lacks IPv6 support === frankban is now known as frankban|afk [18:50] Bug #1639930 changed: initramfs network configuration ignored if only ip6= on kernel command line [18:50] [19:20] Bug #1628313 changed: [2.1 ipv6] comissioning should ignore SLAAC addresses [20:20] Bug #1654063 opened: CentOS images have overly specific (and wrong) version numbers [22:25] Ok, I have a maas server (both region and rack) that is not acting as a router. I have a separate router. the maas server has multiple interfaces on it, one for the lab network, and one for the rack network. when a machine boots on the rack network, cloud-init tries to talk to the lab network interface and can't. the machine can talk to the rest of the world as a whole. I tried turning on net.ipv4.ip_forward via sysctl but that doesn't seem to [22:25] help. Any ideas? [22:26] Is it possible to customize the url cloud-init tries to talk to on a per subnet basis? [23:51] does anyone know why cloud-init running on a maas deployed node would be failing to post events?