=== frankban|afk is now known as frankban [12:14] Hi, it seems that MAAS generates network config during deployment even for devices that are in Unconfigured (interfaces are not used or connected) state after commisioning. I have version 2.2.1 (6078-g2a6d96e-0ubuntu1~16.04.1) [12:14] do you have same experience ? [12:14] seems like bug to me [12:14] get-curin-config shows [12:15] - id: eno3 [12:15] mac_address: 80:18:44:e0:d4:0a [12:15] mtu: 1500 [12:15] name: eno3 [12:15] subnets: [12:15] - type: dhcp [12:15] type: physical [12:15] - id: eno4 [12:15] mac_address: 80:18:44:e0:d4:0b [12:15] mtu: 1500 [12:15] name: eno4 [12:15] subnets: [12:15] - type: dhcp [12:15] type: physical [12:16] they should be ignored [12:16] v92: i tested something with network configuration yesterday and it did not configure interfaces with DHCP that are "unconfigured" [12:17] what version do you use ? [12:17] v92: 2.2.2, but between 2.2.1 and 2.2.2 there hasn't been changes in network config [12:18] let me try to re-test [12:18] I'm using dell for this test and I'm using only first two interfaces [12:19] it prolongs boot time of server because it tries to get address with DHCP which wont work but it takes about two minutes [12:21] yup, i'll run a quick test [12:21] in the meantime, do you mind getting the output of [12:22] output of ? [12:22] maas admin interfaces read [12:23] https://pastebin.com/JAgFdrdx [12:26] v92: so eno3/eno4 are unconfigured [12:26] v92: but do show this " "links": [ [12:26] { [12:26] "id": 939739, [12:26] "mode": "dhcp" [12:26] " [12:29] * roaksoax brb, reboot [12:30] yep, that's what I'm saying [12:30] for some reason they're configured for dhcp even they're not configured to be [12:31] v92: do you have a external dhcp server connected to those interfaces ? [12:39] to eno3 and eno4 ? they aren't connected [12:41] v92: I'm testing now to see if I can reproduce [12:46] gr8 [12:48] v92: http://paste.ubuntu.com/25382734/ [12:48] v92: http://pastebin.ubuntu.com/25382746/ [12:48] v92: jus commissioned and deployed a machine and it works as expected [12:49] v92: has this maas been upgraded or did you use to manage those interfaces with DHCP or anything of the sort ? [12:50] it was upgraded but now I realize why does it behave this way [12:52] I have oneline patch that resets devices to DHCP because I needed them to be in that configuration after release.I didn't trust users that they will set it every time before deployment as it was needed. [12:53] gotcha :) [12:53] I was thinking it it does only on enabled devices but device can be enabled regardless of its connected state [12:53] # Set LINK_UP mode on all the other enabled interfaces. [12:53] for interface in self.interface_set.all(): [12:53] if interface == boot_interface: [12:53] # Skip the boot interface as it has already been configured. [12:53] continue [12:53] if interface.enabled: [12:53] interface.ensure_link_up() [12:53] interface.link_subnet(INTERFACE_LINK_TYPE.DHCP, None) [12:53] in maasserver/models/node.py [12:54] that last line is mine [12:54] v92: yeah "link_subnet" will basically tell it to bring it up [12:56] I will probably just delete those sections in interfaces file during deployment and it would behave as expected [12:58] roaksoax: anyway, thank you for your time ! [13:18] v92: glad to help [15:11] roaksoax: did you hear anything from the bug with the resolv.conf? [15:13] Bug #1441624 changed: MAAS squid proxy doesn't support using an upstream proxy [15:19] Bug #1441624 opened: MAAS squid proxy doesn't support using an upstream proxy [15:25] Bug #1441624 changed: MAAS squid proxy doesn't support using an upstream proxy [15:35] ybaumy: nothing really, one thing I have to test is whether the dns resolution while cloud-init is working [15:35] ybaumy: cause it could be that the interface re-requested dhcp and something cased it not to obtain dns === frankban is now known as frankban|afk [18:25] Bug #1597801 changed: cannot create node VLAN interfaces on centos [18:25] Bug #1697931 changed: [2.2,snap] Proxy isn't running after installing the snap [19:44] Hi. I'm searching for a solution for the following problem: I have a rack controller with 5 interfaces which are automatically picked up by maas, i.e. creating fabrics and subnets etc. I want MAAS to handle only two of these subnets so I set all the other subnets to "unmanaged". However it seems that this does not affect DNS behavior. When I try to ping the rack controller from a node in my network it retrieves to a local ip of an unman [19:44] Is there a way to specify which subnet to prefer? Ultimately I don't want MAAS to care about any of the other subnets. [19:57] Pepper: i dont understand this "itr retrieves to a local ip of an unman" ? [19:58] To a ip on the rack controller which is not connected to the rest of the network [19:59] there are 5 subnets on the rack controller but only 2 are accessible from the deployed nodes [19:59] Pepper: so you are pinging to the hostname of the rack instead of the ip of the rack [19:59] Pepper: try .. [19:59] Pepper: try .. [19:59] sure that works [19:59] but I want it to resolve to a different default ip [20:00] I don't want to specify that specific network interface [20:01] i can't remember fi we did that or not [20:01] mpontillo: ^^ can you ? [20:01] i /thought/ you could decide to change which interface gets the hostname ... remember ? [20:01] Previously I just deleted the unused subnets but every time I restart maas they are picked up again [20:02] roaksoax: Pepper: I think if you set the default gateway interface using the API, that's what we'll select [20:03] Pepper: someone on the mailing list just mentioned this problem; the solution they came around to was to use lxd/lxc to create a container to run MAAS inside, so that only the relevant interfaces would appear to be on the controller in the first place [20:05] huh. 2017-08-24 20:05:24 maasserver.preseed: [warn] WARNING: '/etc/maas/preseeds/curtin_userdata' contains deprecated preseed variables. Please remove: main_archive_directory, ports_archive_directory, http_proxy [20:05] Ok thx I'll check if the default gateway works otherwise I will try to setup the container solution for now. It would be nice to permanently delete subnets from maas ;) [20:06] (oops, didn't mean to paste that log here, sorry for the spam) [20:06] Pepper: the other option would be the add a dns resource record [20:06] Pepper: either a CNAME or a A record [20:06] This overrides the other A record? [20:20] Pepper: this is what I did: [20:20] ubuntu@maas00:~$ ping maas00.maas [20:20] PING maas00.maas (192.168.1.13) 56(84) bytes of data. [20:20] maas admin dnsresources create fqdn=maas00.maas ip_addresses=10.90.90.1 [20:21] ping maas00.maas [20:21] ubuntu@maas00:~$ ping maas00.maas [20:21] PING maas00.maas (10.90.90.1) 56(84) bytes of data. [20:21] Ok that looks like what I want :-) [20:21] thx [20:27] Just tested it and it works! Thanks again for your help. [21:11] Bug #1712917 opened: [2.3] Cannot use fqdn parameter when creating dnsresource-records === sfeole is now known as sfeole-away