| Will | I have maas installed on an orange box. using juju I can load OpenStack. I want to use maas by itself to install/boot several nodes. However, on the nodes page I select a few nodes and run the bulk action but always get an error message: The action "Start selected nodes" could not be performed on 1 node because its state does not allow that action. | 04:10 |
|---|---|---|
| Will | Any idea about what state it is taking about? | 04:11 |
| mup | Bug #1415493 changed: MAAS raises 'Cluster Master' (some UUID); no connections available' when changing a cluster interface <MAAS:Invalid> <https://launchpad.net/bugs/1415493> | 10:44 |
| mup | Bug #1415493 was opened: MAAS raises 'Cluster Master' (some UUID); no connections available' when changing a cluster interface <MAAS:Invalid> <https://launchpad.net/bugs/1415493> | 10:50 |
| mup | Bug #1415493 changed: MAAS raises 'Cluster Master' (some UUID); no connections available' when changing a cluster interface <MAAS:Invalid> <https://launchpad.net/bugs/1415493> | 10:56 |
| === jfarschman is now known as MilesDenver | ||
| mup | Bug #1455560 was opened: when I move servers between tags, list in nodes view is out of sync <oil> <MAAS:New> <https://launchpad.net/bugs/1455560> | 16:12 |
| elurkki | What could be the problem when Commisioning new KVM virtual machine with MaaS the virtual machine gets IP nicely from DHCP, uses it, prints it on the screen, but the IP is not added to the Node. The IP can not seen in the node and from this DNS is not updated. | 20:34 |
| elurkki | The IP can be found from "Discovery Data" list aka. lsww:setting: ip: ip, value: 10.0.0.201 | 20:35 |
| elurkki | Thanks a lot for any tips | 20:35 |
| mup | Bug #1455643 was opened: 1.8 regression: Node listing extends past the edge of the screen <MAAS:Triaged by ubuntudotcom1> <https://launchpad.net/bugs/1455643> | 20:36 |
| mup | Bug #1455656 was opened: 1.8beta6: not filter for tags doesn't work <oil> <MAAS:New> <https://launchpad.net/bugs/1455656> | 21:16 |
| ahasenack | elurkki: commissioning is something temporary, after that the machine changes into Ready and gets powered off, so no IP | 21:18 |
| ahasenack | elurkki: do you get that far? To the Ready state and power off? | 21:18 |
| elurkki | There I get yes | 21:21 |
| bleepbloop | Is there a way to tell how many disks maas thinks a machine has? | 21:21 |
| elurkki | ahasenack: Should the MaaS handle the IP to the DNS ? | 21:21 |
| ahasenack | elurkki: during commissioning it will use an IP from the so called "dynamic range" that you configured in the cluster | 21:22 |
| ahasenack | elurkki: but after all that, with the machine Ready and off, the IP is free | 21:22 |
| elurkki | ahasenack: Yep, I can see it gets the IP from DHCP pool | 21:22 |
| ahasenack | elurkki: you need to start it | 21:22 |
| elurkki | Testing again | 21:22 |
| ahasenack | elurkki: once you start it (or deploy, in maas 1.8 terminology), it will get an OS installed and an IP from the "static range". That will be in DNS | 21:22 |
| elurkki | Commisioning with fast boot now | 21:23 |
| ahasenack | no | 21:23 |
| ahasenack | commissioning is only done once | 21:23 |
| ahasenack | if it's in the Ready state, you don't need to commission anymore | 21:23 |
| elurkki | Can not boot it | 21:23 |
| elurkki | Hangs in the PXE | 21:23 |
| elurkki | And timeouts | 21:23 |
| elurkki | So thought to commision it again | 21:24 |
| ahasenack | I thought you said the machine was in the "Ready" state | 21:24 |
| elurkki | It is | 21:24 |
| elurkki | Power OFf | 21:24 |
| elurkki | If I power it on | 21:24 |
| ahasenack | these are two things: Ready, and powered off | 21:24 |
| ahasenack | if it's Ready, it means it commissioned correctly before | 21:24 |
| elurkki | and the machine is powered off in ready state I assume ? | 21:25 |
| elurkki | pysically powered off | 21:25 |
| ahasenack | bleepbloop: in maas 1.7, go to the node page in maas and click on "discovered details", you will have to search a bit | 21:25 |
| ahasenack | elurkki: it's written in the node list page, it's a collumn. What does it say for this machine? | 21:25 |
| elurkki | Status: Ready | 21:25 |
| ahasenack | elurkki: ok, then select it and from the drop down action menu select "start" | 21:26 |
| elurkki | done | 21:26 |
| elurkki | "Deploying" | 21:26 |
| bleepbloop | ahasenack: okay, do you happen to know what canonical landscape looks for exactly there? I'm having an issue where landscape is attached to maas and is not seeing as many drives as there actually are | 21:26 |
| ahasenack | bleepbloop: are you sure it's drives, and not networks? That's what usually trips people (networks) | 21:27 |
| ahasenack | bleepbloop: there is one checklist item that wants both on a machine: two disks, and two nics. The later attached to actual networks (can be the same, but must be attached) | 21:28 |
| bleepbloop | ahasenack: the failing condition is "At least three machines with more than one disk have been commissioned" | 21:28 |
| ahasenack | bleepbloop: ok, then landscape will look for that hardware information from maas, in the "discovered details" | 21:28 |
| ahasenack | the disks need to be larger than 1Gb, and cannot be virtio disks | 21:28 |
| ahasenack | (if using a VM) | 21:28 |
| bleepbloop | Okay, they are all physical machines, I guess what is most likely messing it up is it is detecting one of the disks as 1gb even though it is larger | 21:29 |
| ahasenack | bleepbloop: there is a virtual disk that maas uses for commissioning, it appears in that output, but is ignored | 21:29 |
| ahasenack | it's a QEMU something something disk | 21:29 |
| ahasenack | bleepbloop: one other case I saw once was about raid disks | 21:30 |
| ahasenack | some confusion between the actual disks, and the raid device | 21:30 |
| bleepbloop | It is probably the raid disks, I have raid enabled, I'm guessing I should just completely disable raid for this case? | 21:30 |
| ahasenack | bleepbloop: is it a dell? | 21:30 |
| bleepbloop | It is 2 ibm's and an hp | 21:30 |
| mup | Bug #1455656 changed: 1.8beta6: not filter for tags doesn't work <oil> <MAAS:New> <https://launchpad.net/bugs/1455656> | 21:31 |
| ahasenack | bleepbloop: ok, try disabling raid | 21:31 |
| ahasenack | then commission again | 21:31 |
| ahasenack | if that still doesn't work, enlist again. I'm not sure commissioning picks up this kind of hardware change | 21:31 |
| bleepbloop | ahasenack: Will do, I don't have access to the hardware where I am right now, however I will definitely try that monday, thank you! Do you think it will be enough to tell the raid controller to treat each drive as a simple volume? I am only asking because it seems everyone has really struggled with completely disabling raid on this machine, its an IBM x3650 | 21:34 |
| ahasenack | bleepbloop: all that matters in the end is what linux sees once it boots | 21:35 |
| ahasenack | bleepbloop: the code runs the "lshw" tool, you can try it somewhere else if you want | 21:35 |
| ahasenack | it two disks are shown as something like /dev/md0, then that will count as one disk only, and the checklist will fail | 21:35 |
| ahasenack | unless you have 4 disks ;) | 21:35 |
| ahasenack | bleepbloop: the two disks are required because one is used for the OS, and the other is used entirely for storage (ceph or swift) | 21:36 |
| ahasenack | so access to the actual disk is needed (the block device) | 21:36 |
| mup | Bug #1455656 was opened: 1.8beta6: not filter for tags doesn't work <oil> <MAAS:New> <https://launchpad.net/bugs/1455656> | 21:37 |
| mup | Bug #1455658 was opened: 1.8beta6: There is no documentation for the filter syntax <oil> <MAAS:New> <https://launchpad.net/bugs/1455658> | 21:37 |
| mup | Bug #1455659 was opened: 1.8beta6: No way to filter to find nodes that don't have any of a list of tags <oil> <MAAS:New> <https://launchpad.net/bugs/1455659> | 21:37 |
| bleepbloop | ahasenack: Gotcha, okay I will have to play around with a livecd boot on all of the machines and see what it is really seeing, so out of curiousity, if the machine has 3 drives which are all detected, will it use them all or just use 1 for os and 1 for storage? | 21:39 |
| ahasenack | bleepbloop: I think it will use the other 2 for storage | 21:40 |
| ahasenack | bleepbloop: we pass it a list of devices from /dev/sda to /dev/sdz, and the storage charm tries them all | 21:40 |
| ahasenack | of course, it won't wipe /dev/sda if that's where the OS is :) | 21:41 |
| ahasenack | s/wipe/use/ | 21:41 |
| bleepbloop | ahasenack: very cool, okay thats really useful, does the os always get installed on the first one seen or is it smart about using the smaller one for the os and larger ones for storage? | 21:41 |
| ahasenack | bleepbloop: maas installs the OS for the autopilot, | 21:42 |
| ahasenack | bleepbloop: I believe it's very simple in that regard: something like alphabetical order of the block devices | 21:42 |
| bleepbloop | ahasenack: ah so thats pretty much up to maas at that point | 21:42 |
| ahasenack | right | 21:42 |
| ahasenack | juju asks maas for an ubuntu trusty machine, maas delivers | 21:43 |
| mup | Bug #1455658 changed: 1.8beta6: There is no documentation for the filter syntax <oil> <MAAS:New> <https://launchpad.net/bugs/1455658> | 21:43 |
| mup | Bug #1455659 changed: 1.8beta6: No way to filter to find nodes that don't have any of a list of tags <oil> <MAAS:New> <https://launchpad.net/bugs/1455659> | 21:43 |
| ahasenack | then the autopilot says what will be installed there in terms of services | 21:43 |
| ahasenack | I have seen a case where the machine had /dev/sda, /dev/sdb and /dev/nvmsomething | 21:43 |
| ahasenack | and /dev/nvm* was chosen for the OS | 21:43 |
| ahasenack | it was a high end ssd | 21:44 |
| ahasenack | had to customise a bit the installation step to have it use /dev/sda, which is what we wanted | 21:44 |
| mup | Bug #1455658 was opened: 1.8beta6: There is no documentation for the filter syntax <oil> <MAAS:New> <https://launchpad.net/bugs/1455658> | 21:46 |
| mup | Bug #1455659 was opened: 1.8beta6: No way to filter to find nodes that don't have any of a list of tags <oil> <MAAS:New> <https://launchpad.net/bugs/1455659> | 21:46 |
| bleepbloop | ahasenack: Interesting, really wish there was a bit more documentation on just how maas works, it does some stuff where it seems to try to abstract away as much stuff as possible from the user and leaves you wondering just what its going to do in a specific situtation | 21:46 |
| elurkki | ahasenack: This node which is trying to "Deploying" just hangs to the "Booting under MAAS direction". Probably my MaaS server just is not capable enough and does have something configured wrong. | 21:49 |
| ahasenack | bleepbloop: I think in general it's thought that if you have dozens of machines, grouped somehow, they are bound to be similar | 21:49 |
| ahasenack | elurkki: yeah, something else is wrong then, sorry | 21:50 |
| ahasenack | elurkki: did the vm do anything else since you clicked on start? | 21:50 |
| elurkki | ahasenack: Oh, after 7th boot try it started to go forward from the PXE | 21:50 |
| ahasenack | like, install the OS? You would have seen a big wget output here it downloaded images | 21:50 |
| ahasenack | elurkki: it's a pretty straight forward setup, we use it all the time (maas with VMs), it's probably a networking issue of some kind | 21:51 |
| elurkki | ahasenack: It just powered on, acquired IP from DHCP and then started to try PXE | 21:51 |
| ahasenack | don't forget to disable dhcp on the network where you attach maas to, for example | 21:51 |
| ahasenack | the only dhcp server should be maas | 21:51 |
| elurkki | Yes, there is only MaaS DHCP | 21:52 |
| bleepbloop | ahasenack: Well thanks for your help, I'll give disabling the raids as much as possible a go and hope it works! | 21:52 |
| elurkki | reading MaaS syslog to see | 21:52 |
| ahasenack | bleepbloop: good luck | 21:52 |
| elurkki | But I agree about other problems | 21:52 |
| ahasenack | bleepbloop: if you hit more issues, you can try askubuntu.com with the autopilot and/or landscape tags | 21:52 |
| ahasenack | bleepbloop: I rarely idle here | 21:52 |
| elurkki | I assume this old Fujitsu RX100 S4 with deadly old virtualization support is just pulling my leg | 21:53 |
| ahasenack | I mean, I *idle* here :) | 21:53 |
| ahasenack | elurkki: I created it all in virt-manager | 21:53 |
| bleepbloop | ahasenack: Will do, I was eyeing that up as well however figured I'd try here first :P | 21:53 |
| ahasenack | elurkki: network, vms | 21:53 |
| ahasenack | bleepbloop: sure | 21:54 |
| elurkki | ahasenack: I am using bridged network. It might be problem as well | 21:54 |
| ahasenack | hm | 21:54 |
| ahasenack | elurkki: I used NAT, what virt-manager calls NAT | 21:54 |
| ahasenack | elurkki: just disabled DHCP | 21:54 |
| ahasenack | elurkki: then created the vm for maas, and for the nodes, attached to this network | 21:55 |
| ahasenack | elurkki: that's the simplest case and should work. Later on you can try more advanced scenarios | 21:55 |
| elurkki | ahasenack: yep, probably have to test that scenario first | 21:55 |
| ahasenack | that's isolated enough, and allows the nodes to reach the internet without problems | 21:55 |
| elurkki | I agree | 21:56 |
| ahasenack | but you can only reach the nodes from your machine, of course | 21:56 |
| ahasenack | I'm gonna go | 21:57 |
| ahasenack | good luck guys | 21:57 |
| ahasenack | see you around | 21:57 |
| elurkki | thanks a lot for all the help | 21:57 |
| elurkki | cu later | 21:57 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!