[00:03] I think it is you and I only bud [00:03] no one else in the world [00:04] Prolly. I'm trying to figure out where the setting is to allow nodes to talk to the outside world [00:05] What do you mean allow them to talk to the outside world? [00:06] Basically, I have the region (and rack) controller on the office network, and all the nodes on a separate wired network, with the only line out through the region controller. It should act like a proxy, but it only does it for apt out of the box, I think. [00:08] Oooh [00:10] You'll need a VLAN no? [00:11] I've got a lot of those, too. I'm vaguely following a tutorial by a Juju dev ( http://blog.naydenov.net/2015/11/deploying-openstack-on-maas-1-9-with-juju-network-setup/) [00:11] Mainly, I'm still learning how the various configurations are set and overridden. [00:11] You'll have one VLAN to do the management and another VLAN on the same port that allows outside access [00:13] So maas uses squid as a proxy at /var/lib/maas/maas-proxy.conf, and I'm trying to figure out if it's allowing my node to talk to the outside world or not. Hilariously, it's handling outside DNS great, which means I have the opposite problem I had a long time ago [00:14] Fun! [00:14] Do you know the difference between fabric and space? [00:15] I think a space is an arbitrary designation for grouping things together [00:16] But a fabric should (at least semantically) be on its own physical network [00:17] In the tutorial linked there, I set up two fabrics: `maas-external` which is just on the one NIC on the region controller going to the office, and `maas-management` which is all things connected to the maas switches [00:19] So lets say I have two subnets: 1 management and another that is my block of IPs from the isp. [00:19] And when I go to provision new nodes I'd like for MAAS to use the subnet from my ISP and hand out IPs to the nodes [00:20] Is maas capable of doing so while keeping track? [00:20] I think so [00:22] But honestly I am not sure exactly how to go about that - I'd guess that after the node is commissioned, you would set the NIC to have its IP autoassigned [00:44] Hmm in the rack controller, there isn't a green check next to the dhcpd [00:46] That's a bad day - I had that happen too. There can be a few reasons for that from what little I've seen: make sure you're actually serving dhcp on a subnet (make sure there isn't a conflict), make sure there isn't a bogus subnet (like a blank IPv6 one) that is confounding it [00:50] Hmm [00:51] I have the interface set with a CIDR 192.168.1.0/24 [00:51] with a gateway at 192.168.1.1 [00:52] I assume that is also set as the region controller's IP address? [00:52] err, and/or racks? (sometimes that needs a dpkg-reconfigure to get set after install) [00:55] I only have one management node which is a region and a rack? [00:55] 1 controller that is a Region and a rack controller [00:55] Makes sense. [00:56] And on the subnet that I have set for fabric 3 on interface em3 it says it is serving dhcp [00:56] according to https://docs.ubuntu.com/maas/2.1/en/installconfig-subnets-dhcp#enabling-dhcp [00:56] but ifconfig doesn't show anything on that port [00:57] I don't think ifconfig will show that, though I don't know enough to say why [00:58] But if the subnet says that it is serving dhcp, and if rackd shows dhcpd is running, then I'd bet it's working [01:00] So rackd doesn't have a check next to dhcp [01:01] ah, I was hoping that got fixed for you; hmm, I don't really know too well how to troubleshoot that [01:01] Hmp [01:01] I'd guess there's a conflict preventing it from starting [01:01] The thing is there is nothing else on that machine [01:01] perhaps check /var/log/maas/? [01:01] it has a static route on a different port [01:02] 2017-01-25 23:42:43 maasserver.dhcp: [info] Successfully configured DHCPv4 on rack controller 'ncgqxy'. [01:03] I guess I'm out of my depth on that [01:06] shame [01:07] Much so. I really wish I understood better how that particular bit works [01:07] me to [01:32] hmm [01:32] http://askubuntu.com/questions/683733/maas-lan-cable-wiring-topology-for-dhcp?rq=1 [01:32] did you set your gateway in the subnet to the upstream router perhaps? [01:33] I don't think so... If I do a wget on my earlier pastebin ( `wget http://pastebin.com/raw/FkMnpY8f` ) it will correctly resolve the ip of it, but it will never connect [01:33] from a deployed node, that is [01:34] hmp [01:34] The maas controller is gold [01:35] If I set the proxy to my maas controller via `export http_proxy=http://10.222.222.10:3128/`, then I get a 403 forbidden. So it's like the built-in maas proxy could work, but it rejects everything [01:36] http://www.openstackbasement.com/maas-network-hardware [01:40] I think that may be for an ealier version of maas (maybe pre-1.8?) [01:40] Though that's close to my setup, where I just don't have that red line [01:43] Part of the problem is the docs and a lot on the net are for pre-maas2.0, and a lot has been added and managed in maas 2, so it's hard to tell what's still meaningful [01:45] Well, there are a lot of 2.x docs, too. But it's not quite as expansive yet [01:50] yeah [01:50] So I have teh checkmark next to the dhcp now [01:50] awesome [01:51] but there is no activity on that port [01:51] no activity light* blinking on that port [01:52] Might not do much until it commision/deploy? [01:59] hmp [01:59] on your maas server [01:59] are you doing dhcp? [02:00] if so do you get anything in ifconfig? [02:00] Quite a bit of it [02:00] ? [02:01] The maas controller is doing dhcp on all subnets that are not on the wan or office (basically all the 10.xxx stuff) [02:01] but on that interface port? [02:01] like em# [02:01] ifconfig shows that all my adapters and vlans are live [02:01] hmp [02:01] interesting [02:01] mine doesn't show anything but the main network [02:01] There's almost no traffic, since only one node is live [02:02] I have em1 which is the network and I'm trying to set em4 to the private network [02:02] Ah, mine's enp3s0f0 and enp3s0f1 [02:03] yeah I don't expect the names to match [02:03] So the first is the external, and the second is the internal. Everything internal seems to be sane, though I haven't bootstrapped juju or fully configured it all yet [02:03] Externally, I serve no dhcp [02:03] are you serving dhcp from the region /rack controler? [02:03] The little office router will handle that (well, the Windows domain controller, I guess) [02:04] I see [02:04] http://pasteboard.co/2EyTOQSDf.png [02:04] The rack/region controller handles dhcp for everything on the maas management network [02:04] Yeah [02:05] One sec, snapshotting mine [02:06] also if you could pastebin your cat /var/lib/maas/dhcpd.conf ? [02:07] http://pasteboard.co/qxGFsGdut.png [02:08] http://pasteboard.co/qxHrk6m9K.png [02:09] http://pastebin.com/raw/qYDuH0c1 < cat /var/lib/maas/dhcpd.conf [02:12] Note that I only configured names. The maas setup automatically filled in all the stuff from my /etc/network/interfaces file [02:12] yeah [02:12] it basically matches mine [02:12] with more vlans [02:13] I also have to ifup the vlans after startup, because the interfaces file apparently trips overitself if you do it all there at once [02:25] what does your /etc/network/interfaces look like? [02:26] did you have to configure the interface there? [02:26] I got the activity light to come on by setting it in /etc/network/interfaces [02:26] but if I set it to dhcp it doesn't come up [02:26] I've configured them all there [02:26] I seee [02:27] http://pastebin.com/raw/BVCfWxan [02:28] I'm just noticing that there's some iptables stuff there. I dunno if that's not needed anymore; the EC2 redirect was a response to what I think was a bug in the setup, but again, I may need to try it without the rules [02:29] Note that I commented out the vlan auto enp... lines, since it breaks the setup. I have to ifup the vlans after the network is up [02:31] OH NICE [02:31] *does a dance* [02:31] Make sense? [02:31] Though I have two of the rackcontrollers popping up [02:31] on 192.168.1.1 and 192.168.1.2 [02:31] but the other machine plugged in is popping up on the dhcp [02:32] Oh, good! [02:32] I don't understand the two, but hey [02:32] Now I see the other machine in the subnet [02:33] Seems like something that could at least be troubleshot, right? [02:34] how do I now commission? [02:34] hmm [02:37] It isn't showing up in device discovery [02:38] I've added machines manually; I have a text file with all the machine's MAC addresses and the static IPs I set for the IPMI setup, and then just commision manually [02:38] I think it will eventually discover the machine, but that's a new feature that I just haven't gotten to trying yet [02:42] Alright, giving up for the day. Good luck! [02:43] thanks buddy [02:50] now it is stuck on commisioning [02:50] It can reboot [03:12] Hmm now when I went to commision it did it's PXE thing. THen a new machine popped up [03:12] and then I was able to commission that one? [03:12] acquire [03:16] g3, what are you trying to do? the backscroll is too long [03:21] No worries [03:21] I finally commisioned and acquired [03:22] when I went to deploy [03:22] Stdout: "Error: /dev/sda: unrecognised disk label [03:22] Installation failed with exception: Unexpected error while running command. [03:22] Command: ['curtin', 'block-meta', 'custom'] [03:22] Exit code: 3 [03:22] Reason: - [03:22] Stdout: "Error: /dev/sda: unrecognised disk label\nError: /dev/sdb: unrecognised disk label\nno disk with serial 'PHFT6401010P800CGN' found\n" [03:22] Most of the backscroll is myself and andrew-ii trying to get my dhcp to work [03:23] I think I sorted that out. [03:23] Though the machine wasn't showing up in the device discovery, so I manually added a node. [03:23] using the IPMI 2.0 I was able to reboot / commission [03:24] but the node that I created hung on commisioning, while a new node popped up with a random name (warm-spider.maas) [03:24] I was able to then commission that node (which was the one that I created...as that is the only machine plugged in) [03:24] and it had the same DHCP ip of x.x.x.101 [03:25] then after it commisioned I acquired it, setup a RAID1 on sda&sdb [03:25] went to deploy [03:25] and it failed [03:27] So now I released it and am currently erasing the disks [03:29] you don't need to acquire. that's really just used if a user wants to reserve a node. maas will never automatically use a node that is acquired [03:33] ok so correct path is to commision then deploy? [03:34] g3, yes. curious. did you start by reading the maas documentation? [03:35] I did [03:35] but frankly, it seems a little spotty [03:35] this gives a runthrough: https://docs.ubuntu.com/maas/2.1/en/installconfig-checklist [03:35] subheadings: 'Commission a node', then 'Deploy a node' [03:36] Okay [03:36] so what does acquiring actually do? [03:37] And under DHCP I had to setup the MAAS interface (em4 for instance) under /etc/network/interfaces as a static route [03:37] then in MAAS give that interface a static route as well [03:39] g3, re acquire, i pretty much told you [03:39] but you can read here: https://docs.ubuntu.com/maas/2.1/en/intro-concepts [03:39] (see 'acquire' action and 'allocated' status) [03:40] Okay [03:40] thanks [03:40] Where did the random name come from? [03:40] warm-spider [03:41] when used with a service orchestration tool, such as Juju, acquire happens under the hood [03:41] g3, from some python library probably [03:41] (random names) [03:41] fair enough [03:42] g3, i have never had to set up a static route in MAAS yet. why did you do that? [03:43] Because nothing was coming up? [03:43] I went to the VLAN and setup DHCP [03:43] went to the interface on the region / rack controller and set it to DHCP [03:44] but nothing was showing up in mass, with ifconfig and the activity light wasn't blinking on the port [03:44] Only once I setup static routes for MAAS did the green check next to DHCPD on the region/rack controller page appear [03:45] And only then did the other machine I am attempting to deploy show up in the node list [03:45] you set the interface on the region/rack controller to 'DHCP'? [03:45] I did, but nothing was happening [03:46] I can remove the static from /etc/network/interfaces and give it a go again [03:46] that's not how you enable DHCP [03:47] i recommend you read the docs: https://docs.ubuntu.com/maas/2.1/en/installconfig-subnets-dhcp [03:49] anyway, i need to go to bed. please read the docs. if you find them lacking kindly open a bug. there is a link at the bottom of the menu (left pane) [03:49] I have [03:49] and I've done that [03:49] I've now removed the route from /etc/network/interfaces [03:49] in the vlan on the 192.168.1.0/24 subnet I have enabled the dhcp [03:49] x.x.x.101-x.x.x.254 [03:50] in the region/rack controller I have the em4 interface set to dhcp [03:50] but now the check mark next to the dhcpd is gone and there is no activity on the port again [03:53] http://pasteboard.co/qzuzQs9d1.png [03:53] hi [03:54] http://pasteboard.co/qzv9izzdK.png [03:54] help configuring maas? [03:54] http://pasteboard.co/qzvF1IMh5.png [03:54] help me? [03:55] how? [03:55] Well first I can't get the dhcp to work unless I set a static in MAAS and in /etc/network/interfaces [03:56] is there another dhcp server on the same network? [03:57] No [03:57] This side of the network is all static routes as it is our public IP block [03:57] and I am setting up MAAS on a private OOB network [03:59] cool [03:59] its supposed to be static [04:00] Hmm okay [04:00] pmatulis said otherwise, but that is the only way I've been able to get it to work [04:01] okay lets say it is working [04:01] It wasn't discoverying the device [04:01] So I added it manually and commisioned the nod [04:01] e [04:01] it then hung at commissioning forever, while a new node of random name popped up [04:01] with the correct cpu, mem, drives, ect.. [04:02] I commissioned that one. then acquired. then deployed with a raid1 [04:02] yup that happens [04:02] which part? [04:02] random name? [04:02] the entire thing [04:02] okay [04:02] then it failed though [04:03] commissioning failed? [04:03] deploy [04:03] oh [04:03] Stdout: "Error: /dev/sda: unrecognised disk label\nError: /dev/sdb: unrecognised disk label\nno disk with [04:03] serial 'PHFT6401010P800CGN' found\n [04:04] well, I tried to configure it for openstack [04:04] and juju bootstrap failed [04:05] And how hmp [04:08] im guessing youre stuck where i am [04:08] nodes arent properly configured [04:08] how have you networked the nodes? [04:09] I currently only have the node connected to the OOB network [04:10] I did see lots of 401's popping up [04:12] is your region controller node configured to access the internet [04:13] It is, but I don't have that subnet forwarded to the nodes [04:14] is your network configured to something like this? - www.openstackbasement.com/maas-network-hardware/openstack%20network.png?attredirects=0 [04:15] minus the 5port dumb switch yes [04:16] have you updated the boot images? [04:16] I only pulled them down maybe 4 hours ago [04:17] have you tried recommissioning the same node again? [04:17] I am doing it right now [04:17] without the raid1 [04:19] okay [04:21] THough now it appears that the machine is off and it is stuck at deploying [04:22] Give it some time [04:22] say 10-15 min [04:22] which maas version are you using? [04:23] 2.1 [04:23] Okay well I'll head home then check on it in 30 min or so [04:24] 2.1 out? updating maas only gets me 1.9 [04:24] are you on 14.4? [04:24] 14.04 [04:25] https://docs.ubuntu.com/maas/2.1/en/installconfig-upgrade-to-2 [04:25] yep [04:25] states you need xenial for 2.0+ [04:25] right [04:25] for 2.0 and above** [04:27] is your node configured to wake on lan? [04:57] ' [05:40] hello [05:45] so I uploaded a sshkey to the MAAS admin account. Deployed a server, but I can't access it? [05:45] ssh ubuntu@ asks for a pass [05:54] Oh it gave it a ip from the reserved block. got it. === frankban|afk is now known as frankban [08:43] Bug #1656717 opened: Juju -> MAAS [2.2+] API integration needs to account for null spaces [08:43] Bug #1643001 changed: Moonshot iLO4 'Power HW address' prevent ipmitool from working [12:30] Bug #1659538 opened: juju deploy fails, unable to contact charm store === CyberJacob is now known as zz_CyberJacob [17:09] Bug #1657491 changed: Several IPs assigned to the same iface in the DB due to 'free' leases [17:09] Bug #1659607 opened: [2.1.3] Can't link subnet with link_up from maas API as in WebUI- Cannot configure interface to link up (with no IP address) while other links are already configured. [17:42] Bug #1659613 opened: [Map subnet] When the user takes action Map subnet, (s)he gets a confirmation message including the link to the dashboard, while the mapping has not started yet and they have to click the "Go" button === frankban is now known as frankban|afk [19:58] Bug #1659642 opened: maas manpage? [20:10] Bug #1659642 changed: maas manpage? [20:16] Bug #1659642 opened: maas manpage? [22:31] Bug #1659672 opened: [2.1, rev5670] Can't release machines previously deployed/failed deployment/or newly deployed after upgrade to latest trunk

[23:10] Question where do I report a BUG ??? [23:53] Teranet, https://bugs.launchpad.net/maas/+filebug [23:55] Thank you