[00:01] Bug #1548560 changed: twisted 16 removes deprecated attributes in Request and DummyRequest [00:04] Bug #1548560 opened: twisted 16 removes deprecated attributes in Request and DummyRequest [04:02] Bug #1548601 opened: Machine attempted to boot on a non-existant image, but it didn't mark as failed [04:05] Bug #1548601 changed: Machine attempted to boot on a non-existant image, but it didn't mark as failed [04:11] Bug #1376483 opened: AttributeError: 'Port' object has no attribute 'socket' [04:11] Bug #1548601 opened: Machine attempted to boot on a non-existant image, but it didn't mark as failed [04:14] Bug #1376483 changed: AttributeError: 'Port' object has no attribute 'socket' [04:29] Bug #1376483 opened: AttributeError: 'Port' object has no attribute 'socket' [12:13] Can anybody help me? I have a sever with 2 interfaces with MAAS running, eth0 is 192.168.202.145/24 and is our main network, eth1 is 192.168.140.1/24. On MAAS i have eth0 set to unmanaged and it just gets dhcp in /etc/network/interfaces, eth1 is statically assigned in that file and under MAAS it does dhcp and dns. I have 5 servers plugged into the switch that eth1 is also plugged into. I can PXE boot, the server will start doing it's t [12:14] the network info it starts timing out grabbing "'http://169.254.169.254/2009-04-04/meta-data/instance-id'" [12:16] I've imported the 14.04 LTS amd64 images in maas [12:17] google leads me to https://bugs.launchpad.net/maas/+bug/1402861 but that doesn't really help [12:21] Am I better off asking this question elsewhere? [12:36] hi folks! in MAAS, if I go to Subnets, I see two fabrics: one for WAN and one for LAN (which I suppose it's ok). BUT when I go to Nodes and choose Subnets (from the left sidebar) I can see only the LAN IPs. isn't this wrong? [12:44] Fixed my problem, i ran dpkg-reconfigure maas-cluster-controller and changed the address from 192.168.202.145 to 192.168.140.1 [12:44] imranh do you have any experience with setting MAAS? [12:45] Razva: none, sorry. I came on here tryng to setup my first cluster [12:45] hah, same here :)) [12:45] do you see your public IPs in the Nodes -> Subnets (sidebar)? [12:46] because I can see only my LAN IPs, which is...weird [12:46] Razva: I've only just got node enlisting working in my enviroment, before abnout 2 minutes ago i couldn't get anything to enlist [12:47] need a hand? [12:47] Razva: Just got it going, need to go down to the server room to switch the machines on now [12:48] don't. use PXE [12:48] or at least you should :D [12:52] Razva: switch the machines on to pxe boot so they appear in the node list is waht i meant [13:23] oh, ok. from my experience using KVM or iDRAC will be REALLY helpfull. I've installed Ubuntu about 25 times in order to have a fully functional MAAS + Juju + AutoPilot + OpenStack installation. [13:23] here are my hints [13:24] 1) DON'T use RAID. just use plain disks, without any kind of raid [13:24] 2) rename your nics from em1 or eno1 or whatever to eth0 and eth1 [13:24] ^^^ https://github.com/Razva/newbadmin/blob/master/device-management/nic-rename.sh [13:25] 3) install BIND Cache on your MAAS server if you want to deploy OpenStack or use any kind of networking stuff, as Juju is VERY bad sometimes at DNS (for reasons I cannot explain!) [13:27] Razva: cheers :) [13:28] I'm just waiting for access to the server room now [14:09] imranh: You don't have IPMI? AMT? Even wake-on-lan? [14:10] haasn: none of it is setup yet, fresh batch of servers [14:10] I don't even know the cpu model in them [14:12] All I know is that each one has: 56 threads, 128GB ram, 2x 320 drives, 2x 10G fiber cards and 1x 1G port [14:21] Bug #1548396 changed: 1.9.0 node stuck in broken status [14:23] woop just got the keys to the server room [14:23] time to go set these servers up [14:24] Bug #1548396 opened: 1.9.0 node stuck in broken status [14:25] imranh: You don't have your workstation located in the server room yet? [14:30] Bug #1548396 changed: 1.9.0 node stuck in broken status [14:30] haasn: If only it was a little bit more quieter [14:31] haasn: Also the temperature difference a few steps can take is unbelieveable [14:31] The noise is so monotonous that it's as good as silence, the upside being that it also drowns out any possible audio distractions [14:33] haasn: :) [14:45] (Such as coworkers furiously yelling at you) [15:33] hmmm maas seems to think my servers are on when they are not... [15:34] imranh: wait for a minute or two till maas updates power states [15:34] imranh: maas doesn't do second by second power state update [15:34] imranh: otherwise your BMC may lock up [15:35] roaksoax: ok :) [15:35] It's funny when you think about it. The handbook for my servers contains an entire section on what to do if the BMC locks itself up (in various ways) [15:35] How are we, in 2016, programming work-arounds for BMCs locking themselves up? [15:36] haasn: that's due to the crappy BMC implementations out there!! You'd be surprised how bad sometimes they are :) [15:36] Imagine if Linux had code to throttle its floating point computations on intel processors to avoid locking up the floating point chip [15:37] It's just bizarre. How can these companies be _so_ bad at programming their devices? [15:37] I mean I get that in the case of our servers it's Sun/Oracle writing/maintaining it, and therefore it probably also runs on Java which explains 90% of that crap, but still [15:37] haasn: i guess the lack of rigorous testing against all the interfaces out there ? [15:38] roaksoax: But even their built-in web interface can lock itself up. You'd think they'd test at least that? [15:39] It seems like a case of “well, if the user power cycles the machine it fixes itself so let's just ship it with a line of documentation alluding to that” [15:39] haasn: i'm sure they do! You'd imagine that these things, being so important, take more attention, but I guess that shipping so many different firwmare version with so many different systems, takes a toll [15:40] (They would probably be more stable if they figured out some way of getting BSD-like microkernel + a small existing httpd to run on them instead of writing their own NIH web server :p) [15:41] Maybe even faster [15:47] hehe [15:53] https://www.youtube.com/watch?v=GZeUntdObCA a great talk on BMCs [16:02] roaksoax, is there a way to deploy currently a node with a bond0 interface without any network associated to it, but have VLAN interfaces on different networks? [16:02] I don't seem to be able to click through this via the UI. [16:22] hmmm the openstack installer is currently telling me i have no machines enlisted with power control [16:37] Nevermind, me being silly [16:40] [ERROR: 02-23 16:20:18, ev.py:130] Exception in ev.run(): Traceback (most recent call last): [16:40] fun === matsubara_ is now known as matsubara [21:30] Hey guys .. Any [21:30] Anyone got this bug recently? [21:31] https://bugs.launchpad.net/maas/+bug/1520645 [21:32] @roaksoak [21:41] bbaqar1: not really [21:42] bbaqar1: what do you see exactly? VM's failing to enlist ? [21:42] Yes. [21:42] bbaqar1: do you have console logs to look ato? maas logs (regiond.log, clusterd.log) [21:42] Which exact logs would you like to see. I can get them once I get access to the setup again. [21:43] bbaqar: 1. console log of the machine trying to enlist 2. regiond.log 3. clusterd.log (the latter 2 around the time you were trying to enlist) [21:44] Okay got it. Will check them out. [21:44] bbaqar: thanks! [21:44] bbaqar: do share1, that will allow to see what may be wrong [21:46] Where to get the console.log of the enlisting node? [21:46] bbaqar: from the console of your enlisting node : [21:46] ) [21:46] bbaqar: we don't record that info [21:47] aah i get it. thats what I was confused about. Will get back to you soon. [21:47] thanks. [21:53] roaksoax: Connecting a node to two subnets with both managing DHCP and DNS should not be a problem right? === lutostag_ is now known as lutostag [22:07] bbaqar: that may indeed be the problem dependending on how dhcp is configuring them [22:10] roaksoax: I did disconnect all cables from the node apart from just one but still got the error. Where can I remove all node specific information from in maas? dhcp.leases? [22:11] bbaqar: has the node been added to MAAS ? [22:11] bbaqar: if the node is not added to MAAS then you don't need to worry about dhcpd.leases [22:11] Basically it fails with the following error: "failed to enlist system maas server 'compute.maas'" where compute.maas is the fqdn of a already added node [22:11] no it hasnt. the enlistment was never succesful. [22:12] bbaqar: interesting! so what's happening is that the enlistment is doing a dig to try to find out the hostname [22:12] if theres any [22:12] bbaqar: and tries to enlist it [22:13] bbaqar: do you have an external dhcp server ? [22:13] err [22:13] DNS server ? [22:13] not a dhcp serve. [22:13] i can check if there is dns server running or not. [22:14] but why does it get the wrong hostname [22:14] i mean right at that time its a baremetal node so doesnt really have a hostname [22:15] bbaqar: try this: http://paste.ubuntu.com/15183875/ on /etc/maas/preseeds/enlist_userdata [22:15] bbaqar: and try to re-enslit. That should fix your issue [22:18] Okay cool. so i have to remove those three lines. [22:18] Great I ll try it out soon. Thanks a bunch.