[00:16] Bug #1659692 opened: MAAS Nodes don't rescan RAM when rebooted [00:19] Bug #1659692 changed: MAAS Nodes don't rescan RAM when rebooted [00:25] Bug #1659692 opened: MAAS Nodes don't rescan RAM when rebooted [00:25] Bug #1659694 opened: Cannot set min kernel to hwe-x [03:00] Does the maas controller only proxy apt and ubuntu images? Or, does it need to be reconfigured to allow nodes access to the internet? [03:10] I'm not sure, but I'd guess that it would have to do with the rules in the maas squid configuration? The file says not to directly modify it, so ... is there another way? [03:24] Or is it only reasonable to dedicate one NIC on each node to the public network and (at least) one to the internal maas network? [03:39] andrew-ii_, maas has a general HTTP proxy that nodes can partake in [03:39] Is it configured by default? [03:39] yes [03:40] Huh. That's both exactly what I wanted to hear and really disheartening [03:40] andrew-ii_, what do you mean? [03:41] Basically, a deployed node can resolve DNS fine, but it never seems to connect to outside sources [03:41] As in, a wget will properly be sent, but fails to ever retrieve [03:41] can the maas server contact outside sources? [03:41] Yes [03:42] is the maas server also set as the default gateway (for the nodes)? [03:42] Yes? It is for every subnet [03:42] is the maas server able to route node traffic to the internet? [03:43] I feel like this is where I'm dropping the ball on the configuation: the maas controller does not seem to route traffic to the internet [03:43] so you probably need to set up masquerading on the maas server [03:44] Ah, so that really is the solution. I had seen reference to it, but I hadn't really understood if that was a workaround or a special case [03:44] Do you have a reference implementation off the top of your head? [03:45] I'm afraid I am not familiar with proper iptables configuration details yet, but I'll learn if that gets the nodes online [03:46] i don't have anything specific on hand [03:46] (I'm currently at the "chinese room" stage of network configuration understanding [03:47] No worries. That basically sets me on a course [03:47] but you should prolly capture packets on the maas server when a node tries to access an internet resource [03:47] to see what's going on [03:48] Yep. That would probably be a very sensible thing to do [03:48] confirm that the requests are at least hitting the maas server but not going out the other interface [03:48] Really wish that occurred to me sooner :P [03:49] Do I need to do anything to the nodes? Like set an http_proxy environment variable? [03:49] Or should the maas controller take the request and route it transparently? [03:49] that should already be set [03:49] Cool [03:49] also, confirm that the maas proxy config file has an ACL for your particular subnet. it is supposed to include one automatically but i have seen it happen that this is broken [03:50] Last I checked it did, but I'm double checking now [03:50] /var/lib/maas/maas-proxy.conf [03:51] you can also set a custom/upstream proxy [03:51] read here: https://docs.ubuntu.com/maas/2.1/en/installconfig-proxy [03:52] Nothing upstream yet, and the proxy has the acl localnet src 10.x.x.x subnets [03:54] and this is the maas-proxy log, to see HIT/MISS or DENIED: /var/log/maas/proxy/access.log [03:55] Alrighty, releasing, redeploying, and checking this out fresh [03:58] Is there any reason to put the maas proxy in the proxy field? That would be redundant, right? [03:58] right, redundant [03:59] Like if my controller was 10.0.0.1, putting http://10.0.0.1:3128 in the field is silly [03:59] right [03:59] ok, that's a whole set of things I need not try again [04:01] Oh, and thanks a mil - just getting confirmation of proper/expected behavior is huge! It will save me huge amounts of second guessing [04:02] i guess the docs are not clear then. please open a bug if you have suggestions on how to improve them: https://github.com/CanonicalLtd/maas-docs/issues/new [04:03] My money is that it's just a hidden learning curve for newbs [04:04] I know just enough to sorta make sense of it, but not enough to be wise to the actual cause of the trouble [04:08] alright, i need to retire. do leave a msg on how you make out and i'll see it tomorrow [04:08] I appreciate the help immensely; Sleep well! [05:06] pmatulis: Turns out your first note was spot on: I needed to properly set up the iptables masquerade. I followed the setup at the end of http://askubuntu.com/a/594723 - I think my settings were just out of place *and* I missed the net.ipv4.ip_forward=1 setting. Way back I had a configuration that *did* have it set, but then the masquerade was wrong. [05:09] The answer links to a tutorial at http://wiki.cloudbase.it/maas that notes the settings that set the NAT from the private network to the public. === frankban|afk is now known as frankban [08:22] Hi roaksoax Do you have time for a short query? [09:20] jlec__, roaksoax isn't around at this time, what's your query? [10:56] brendand: lou peers referenced me to roaksoax regarding some questions I have around Maas. [10:57] jlec__, you can ask them here and he'll see them, or maybe someone else can help [10:57] I am looking for documentation for creating custom images. [10:57] Secondly the CentOS 7 images from Canonical do not work for me. Maas will not assign IP adresses to the nodes [11:14] jlec__, are you using MAAS 2.x? [11:14] brendand: yes, latest on the LTS [12:49] Hey, is there a recommended boot device order for nodes managed by MAAS? |PXE,Disk|, |Disk,PXE| or just |PXE|? [13:02] I believe pxe as to be first but for the rest I'm don't know [13:33] zeestrat, always pxe first. i believe it *should* work with just pxe, but not positive [14:43] has anyone had success deploying centos with LVM? All my nodes are deployed with flat storage. [15:31] jlec: maas doesn't configure networking manuallyt for centos. we only assign a DHCP address for the PXE interface in CentOS [15:32] zeestrat: if your machine supports being told to PXE every time MAAS power manages it, it doens't matter [15:33] zeestrat: maas will tell the machine to PXE every time when you do an action, but if that is not supporte din the BMC, then you will need to make sure to set the boot order [16:24] Trying to install a custom storage layout with LVM. all the partitions are created but curtin fails because it tries to install grub on /dev/dm-0. Is there a way to direct curtin not to try /dev/dm-0? [16:41] Bug #1659891 opened: Trusty deploy fails, with unrecognised disk label, on nodes commissioned with Xenial [16:59] Bug #1659891 changed: Trusty deploy fails, with unrecognised disk label, on nodes commissioned with Xenial [17:04] Hi, I'm having the following problem with maas-region-controller on ubuntu 16.04 when trying to reinstall the package I'm getting the errors below. [17:04] https://pastebin.ubuntu.com/23875672/ [17:04] I believe it is because it verifies that there is already a database and an admin user. [17:07] thiagolib, ah i see you're here [17:12] stokachu: I came in a few minutes ago. [17:12] thiagolib, cool, these be the people that can help with your problem === frankban is now known as frankban|afk === wolverin_ is now known as wolverineav [21:01] Bug #1659948 opened: stuck tgt and tgt-admin, super slow boot [21:27] Anyone do freelance for MaaS Deployment? [21:36] just wondering ... is there a kub power driver floating around anywhere? [21:46] Bug #1659959 opened: [2.2b1] Cannot add a device from the dashboard [21:52] Bug #1659960 opened: LVM support in non Ubuntu OS [21:57] Hello there [21:57] When using device discovery, can one add that as a machine? [22:18] Hmmm so I went to commision four new servers [22:18] and in MAAS they are showing up with no drives [22:19] But in the full output of 00-maas-07-block-devices.out from the commissioning script [22:19] it found all 15 drives [22:26] buut it found them on the second try at commisioning? [22:31] andrew-ii where you at! [23:17] Hi! [23:18] g3 - I dunno about it [23:18] I would actually be curious to know if the system can use any previous knowledge when commissioning a second time [23:19] If so, that may explain why it figured out your drive array the second time (but I wouldn't know :( ) [23:35] Bug #1659982 opened: [2.2b1, UI] When in the node listing, the top menu disappears when scrolling down (when too many machines)