[00:00] <pmatulis> nate_, ask and see
[00:01] <nate_> I've been trying to follow the instructions here: https://jujucharms.com/docs/stable/getting-started
[00:02] <nate_> But even juju bootstrap localhost lxd-test fails due to the proxy. I can't find any documentation for setting juju to use a proxy, thus I also am unable to deploy to any maas nodes.
[00:02] <vmorris> nate_: i have some info on this
[00:03] <nate_> Excellent! I would love to hear it
[00:03] <vmorris> nate_: when you bootstrap your juju controller, use --config=config.yaml where config.yaml contains something like...
[00:04] <vmorris> http://paste.ubuntu.com/23519637/
[00:04] <vmorris> then all of your juju created machines will pick up the proxy environment variables
[00:05] <pmatulis> '--config http-proxy' is prolly what you want
[00:06] <nate_> Awesome! Thank you for that. Yes, I need http-proxy, but I need everything listed in that yaml file. I will test this out and let you know how that goes. Is there a default place I can put that to make Juju automatically use those?
[00:08] <pmatulis> i don't think so
[00:08] <vmorris> nate_ you can set 'juju model-defaults' though you've gotta have the controller bootstrapped... so you can do it during bootstrap or after
[00:09] <vmorris> but i agree with pmatulis
[00:09] <nate_> So before I've done anything I just need to manually enter it. Got it
[00:11] <vmorris> nate_ actually i'm curious now, you might be able to set model-defaults to something and not have to set the config every time you bootstrap
[00:11] <vmorris> seems to be the purpose of that option
[00:11] <nate_> I tried that once before, but it wouldn't let me do it without a model created
[00:12] <vmorris> 'model-config' or 'model-defaults' ?
[00:12] <vmorris> because 'model-defaults' seems to apply to a cloud
[00:13] <vmorris> ah 2nd look, i think you're right about that..
[00:13] <nate_> I can only do that once I've registered a controller, but I can't register a controller without setting the proxy
[00:14] <vmorris> yep, you'll have to do what I do I suppose then.. pass it in with the config at bootstrap
[00:15] <nate_> vmorris unfortunately it didn't work. I still get the error "failed to bootstrap model: cannot start bootstrap instance: unable to get LXD image for ubuntu-xenial: Get https://cloud-images.ubuntu.../index.json: lookup cloud-images.ubuntu.com on... no such host
[00:16] <nate_> It appears it doesn't like my corporate DNS? I can't really tell if it's the DNS or the proxy that's the problem.
[00:16] <vmorris> nate_ was this on localhost lxd or the maas cluster?
[00:16] <nate_> This was localhost lxd
[00:16] <vmorris> do you have connectivity to cloud-images.ubuntu.com from there?
[00:17] <vmorris> or do you need to pass through the proxy as well?
[00:18] <vmorris> like, can you run "lxc launch ubuntu-daily:xenial" ?
[00:18] <nate_> Using curl I can access cloud-images.ubuntu.com from the maas server (which is localhost), and it's almost certainly using the corporate proxy from my environment variables to do that
[00:19] <nate_> But lxc launch ubuntu-daily:xenial failed
[00:19] <nate_> Same no such host error
[00:21] <vmorris> can you run 'lxc profile list'
[00:22] <pmatulis> nate_, maybe DNS is broken. try 'host cloud-images.ubuntu.com'
[00:22] <nate_> NXDOMAIN not found
[00:22] <pmatulis> there you go
[00:22] <nate_> But why would it work with curl?
[00:23] <pmatulis> maybe it's sucking in a variable somehow
[00:23] <vmorris> http vs https?
[00:23] <nate_> I tested curl with both http and https and they both work fine
[00:24] <vmorris> ah nah that wouldn't cause the dns issue
[00:24] <nate_> So DNS is broken in some cases, but working in others
[00:25] <vmorris> in my setup here, i had to create my own dns forwarder too :/
[00:26] <pmatulis> nate_, if host command gives you brokenness then you really need to debug that. try pointing to what's in /etc/resolv.conf
[00:26] <pmatulis> host <name> <resolv.conf-IP>
[00:26] <vmorris> but that was because maas-controller needed it, lxd localhost shouldn't be an issue
[00:27] <nate_> By the way, the maas cluster is fully provisioned, and maas itself isn't having any dns issues
[00:28] <nate_> Using the DNS in /etc/resolv.conf, host still returns not found
[00:28] <nate_> Yet it's the same IP that was used by curl for DNS
[00:29] <pmatulis> try another name? other than cloud-....?
[00:30] <vmorris> curl --dns-ipv4-addr <resolv.conf-IP> cloud-images.ubuntu.com
[00:30] <vmorris> just to confirm that
[00:30] <nate_> curl: (4) A requested feature, protocol or option was not found built-in in this libcurl due to a build-time decision.
[00:31] <nate_> It appears I don't have that feature
[00:31] <vmorris> bummer
[00:33] <vmorris> host google.com ?
[00:33] <nate_> Not found
[00:34] <nate_> But curl google.com works
[00:35] <nate_> maas also has no DNS issues running on this server, and I know it's using the same DNS IP that I am
[00:37] <vmorris> dig google.com ?
[00:38] <nate_> It made it
[00:38] <vmorris> i'll guess that 'dig cloud-images.ubuntu.com' also works --
[00:40] <nate_> ...actually, no, neither worked, sorry
[00:40] <vmorris> aha
[00:40] <nate_> ; <<>> DiG 9.10.3-P4-Ubuntu <<>> cloud-images.ubuntu.com ;; global options: +cmd ;; Got answer: ;; ->>HEADER<<- opcode: QUERY, status: NXDOMAIN, id: 10183 ;; flags: qr rd ra; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 1
[00:40] <nate_> ANSWER: 0 means no response, right?
[00:40] <vmorris> correct
[00:41] <nate_> The AUTHORITY section lists the closest corporate DNS to the proxy
[00:41] <nate_> And an exchange server
[00:42] <vmorris> at the bottom, you should see SERVER: -- does this match what you have in /etc/resolv.conf?
[00:44] <vmorris> i can't imagine why it wouldn't..
[00:44] <nate_> Yes it does.
[00:44] <nate_> I just tested this on another ubuntu machine and got the same results. Curl is successful, but host can't find anything
[00:45] <nate_> It appears this is the case for all Linux boxes on my intranet.
[00:46] <vmorris> cloud-images.ubuntu.com is 91.189.88.141... can you try, just for kicks, 'curl http://91.189.88.141' ?
[00:46] <nate_> Yes, that works, same results as curl cloud-images.ubuntu.com
[00:47] <vmorris> ah right
[00:47] <vmorris> host 91.189.88.141 ?
[00:48] <vmorris> it's possible that your corporate DNS is not setup for forwarding requests to ubuntu.com :/
[00:48] <nate_> Host 141.88.189.91.in-addr.arpa. not found: 3(NXDOMAIN)
[00:50] <vmorris> I should say, forwarding lookup requests to another server for "ubuntu.com"
[00:51] <nate_> But then why would curl work fine? I can also access it in the browser with no issues, and maas successfully downloaded images from there as well
[00:52] <vmorris> yeah, it's a strange problem
[00:55] <vmorris> does wget work?
[00:56] <pmatulis> nate_, try tcpdump on UDP 53 to see what nameserver curl is using. how does it compare to that in resolv.conf
[00:57] <nate_> Yes, wget works great.
[00:58] <nate_> tcpdump defaults to openstack0 now... That's odd
[00:59] <vmorris> yeah, dump udp 53
[01:00] <nate_> Alright. Lots of stuff coming through.
[01:00] <nate_> NXDomain does use the same DNS as resolv.conf
[01:01] <vmorris> see if it's different when you run curl vs host
[01:02] <nate_> host uses A
[01:02] <nate_> I don't actually see anything in there from curl
[01:03] <vmorris> ah yeah doesn't curl cache dns?
[01:04] <vmorris> http://stackoverflow.com/questions/25672005/curl-how-to-set-up-ttl-for-dns-cache-how-to-clear-the-curl-cache
[01:04] <vmorris> defaults to 60 seconds? i think you'd have noticed a problem by now
[01:04] <nate_> Yeah... Though it's odd that curl doesn't show up in my dump
[01:06] <nate_> The dns is always getting asked for ntp.ubuntu.com.maas though
[01:06] <nate_> And ntp.ubuntu.com as well. So I assume it doesn't have trouble with ubuntu.com
[01:07] <blahdeblah> I thought the resolver wasn't supposed to use the search domain for anything containing dots
[01:09] <vmorris> pretty strange nate_ ... best of luck sorting it out :/
[01:10] <nate_> Thank you!
[01:14] <pmatulis> nate_, try strace on curl to see what it's doing (reading files, etc)
[01:19] <nate_> pmatulis the output is enormous
[01:20] <nate_> I did a search for my DNS IP
[01:21] <nate_> I found sin_addr=inet_addr("my.dns.ip.addr")},16) = 0
[01:22] <nate_> Lots of other connections made to that IP in strace
[01:35] <pmatulis> nate_, yet tcpdump does not concur with that right? do you have multiple network interfaces? maybe you applied tcpdump on the wrong one
[01:35] <nate_> One network interface that actually goes to the internet, so the success of curl indicates it's using the right one
[01:36] <nate_> And the dump didn't have the domain name in it when I ran curl, so I'm not sure
[01:38] <pmatulis> try a dump with the failing host command. where does it want to go?
[01:38] <pmatulis> and what, exactly, is the full contents of resolv.conf ?
[01:42] <pmatulis> to assist with troubleshooting, i'm quite certain that the host command uses neither /etc/nsswitch.conf nor /etc/hosts
[02:44] <bala> hey
[02:44] <bala> can i use maas to provision ubuntu desktop version on bare metal ?
[02:50] <roaksoax> bala: you could install extra packages to install ubuntu destop
[02:50] <roaksoax> bala: or you can use lp:curtinator to create a desktop image
[02:51] <bala> thank you roaksoax
[02:52] <bala> can you share any link which has information installing additional packages for desktop
[02:56] <mup> Bug #1644071 opened: [SRU] MAAS 2.1.2 <maas (Ubuntu):New> <maas (Ubuntu Xenial):New> <maas (Ubuntu Yakkety):New> <https://launchpad.net/bugs/1644071>
[02:56] <pmatulis> 'sudo apt install ubuntu-desktop' ? seems that's what worked years ago anyway bala
[02:56] <roaksoax> bala: you could modify /etc/maas/preseeds/curtin_userdata to include something like: http://paste.ubuntu.com/23520222/
[12:58] <mup> Bug #1644219 opened: MAAS web ui does not display bond parameters <MAAS:New> <https://launchpad.net/bugs/1644219>
[13:40] <mup> Bug #1644229 opened: MAAS 2.1.1 - Failed to deploy CentOS7 <MAAS:New> <https://launchpad.net/bugs/1644229>
[14:10] <mup> Bug # changed: 1609975, 1613862, 1621635, 1629430
[14:40] <mup> Bug #1640298 changed: multiple subnets dhcp shared-network does not work with dhcp-relay <MAAS:New> <https://launchpad.net/bugs/1640298>
[17:45] <mup> Bug #1644316 opened: maas ui rejects tags containing "." w/o explanation <error-surface> <MAAS:Triaged> <https://launchpad.net/bugs/1644316>
[19:45] <mup> Bug #1644345 opened: The region should verify database migrations on start <MAAS:Triaged> <https://launchpad.net/bugs/1644345>
[20:51] <junaidali> i have a fresh maas (2.1.2) setup, getting a bunch of errors on commissiong "maasserver.preseed.TemplateNotFoundError: commissioning". Any idea, what might be the issue here?
[20:51] <junaidali> this error too "maasserver.preseed.TemplateNotFoundError: enlist"
[20:54] <junaidali> there aren't any files in /etc/preseeds directory. I think i should try re-installing maas
[20:57] <pmatulis> junaidali, interesting. in 2.1 commissioning templates were removed from their old location
[20:58] <pmatulis> roaksoax, ^^^ related to our previous convo?
[20:59] <junaidali> pmatulis: where can I find the template file?
[20:59] <junaidali> files*
[21:03] <pmatulis> junaidali, i think the best bet is to open a bug. i don't know enough to guide you
[21:03] <pmatulis> https://bugs.launchpad.net/maas/+filebug
[21:05] <pmatulis> remember to include logs (/var/log/maas/{rackd,regiond,maas}.log)
[21:07] <junaidali> thanks pmatulis , opening a bug..
[21:08] <pmatulis> welcome
[21:40] <mup> Bug #1644354 opened: Unable to enlist or commission new nodes with template related errors <MAAS:New> <https://launchpad.net/bugs/1644354>