[07:56] <kklimonda> my maas 2.0 throws "A physical interface can only belong to an untagged VLAN" when I try to assign a physical... interface.. to a tagged VLAN
[07:56] <kklimonda> I mean, I understand the error, just not the reasoning behind it
[07:57] <kklimonda> it seems like LP#1572070 was actually supposed to fix it, so I'm wondering if there is a bug, or I'm doing it wrong.
[08:04] <kklimonda> ok, I guess it makes sense.. need more coffee
[08:09] <kklimonda> to use a tagged VLAN, I need a vlan interface on the node - that part is actually fine. But what I need is a way to assign different untagged vlans to different nics and spaces - as I understand that will fail with MAAS 2.2
[09:43] <BlackDex> is there a quick script to change the bond-mode on all the maas nodes?
[09:43] <BlackDex> or per maas node?
[09:43] <BlackDex> i mean per node
[09:43] <BlackDex> i have a lot of systems with the wrong bond-type
[09:44] <BlackDex> and it seems i can't change the type without destroying the bond-interface via the gui
[09:59] <cnf> any reason why maas would stop logging?
[09:59] <cnf> i no longer have a maas.log
[11:12] <netore> hello, i got this error     [  failed to detect a valid IP address from -1 ] i looked it up and it looks like an old bug but i never seen an explanation of a solution
[11:12] <netore> hello
[11:13] <netore> anyone have a tutorial for phpvirtualbox on ubuntu maas? it seems the ubuntu maas installas a bunch of things that the tutorials i have been following dont accomidate for
[11:23] <zeestrat> BlackDex: Check out the MAAS CLI command "maas $profile interfaces create-bond --help"
[11:24] <zeestrat> BlackDex: You'll need to loop through your nodes to get their system ID and then get the correct interface ID but should be OK
[11:25] <zeestrat> BlackDex: If you're into Ansible, I have a short playbook for that.
[11:27] <netore> im lost already
[11:27] <netore> oh
[11:27] <netore> nm
[11:30] <netore> who wants a beer.
[11:30] <netore> i gotta have this runnin by 1 today.
[11:31] <zeestrat> BlackDex: Apologies, make that command "maas $profile interface update --help". The first one is for creating bonds, not updating their settings.
[11:32] <zeestrat> netore: Would love some beer, but am out of the door. Might be able to take a look in a hour. What's your TZ?
[11:33] <netore> mountain im not sleeping
[11:33] <netore> just before you leave because this could solve my whole isssue... is ubuntu 16 ready for maas, or should i run 14.4
[11:34] <zeestrat> Ubuntu 16.04 (Xenial) is preferred if you want to run an up to date MAAS (2.1.x).
[11:35] <netore> aiight thats what im runnin and im already having issues that i wanst prepaired for.
[11:35] <netore> if your on in an hr i got you a beer.
[11:35] <zeestrat> Gotcha. I've got a HA cluster running on bare metal and have a test setup with KVM so if you're doing something similiar I might be able to help a bit along the way
[11:36] <netore> aiight... im runnin 16 on a laptop as my controller
[11:36] <netore> and just trying to add a test vbox as a node and
[11:36] <netore> getting ready to present this and im not happy
[11:37] <netore> i can send loot thru fb
[11:37] <brendand> netore, can you elaborate on what's going wrong?
[11:37] <zeestrat> Aha. I tried using virtualbox in the beginning but ran into some issues as well. KVM made it a lot easier at least for me.
[11:37] <zeestrat> Back in a bit.
[11:37] <netore> aiight.
[11:37] <netore> brendand
[11:37] <brendand> netore, if it has to be virtualbox, that might make it a bit harder. kvm is easy
[11:37] <netore> what is kvm
[11:38] <brendand> netore, just another hypervisor
[11:38] <netore> im not planning on runnin any vbox nodes
[11:38] <netore> i was just testing i can throw another laptop up
[11:38] <brendand> netore, https://help.ubuntu.com/community/KVM/Installation
[11:39] <netore> failed to detect a valid IP address from -1
[11:39] <netore> thats on the controller on a laptop
[11:40] <netore> my end goal is to run phpvirtualbox on the maas
[11:40] <netore> i have about 10 virtual box;s i need to span across a few machines.
[11:45] <brendand> netore, i'm afraid i don't quite understand what you want to do. do you have physical hardware that you want to add to maas?
[11:46] <netore> yeah
[11:46] <brendand> netore, i was thinking you wanted the virtualbox vm's to be the machines in maas...
[11:46] <netore> nah that was just testing adding nodes.
[11:46] <netore> i will grab another laptop for a node.
[11:47] <brendand> netore, i'd go straight on to that, it should work out of the box as long as the laptop is set to pxe boot
[11:47] <netore> my end goal is to be able to run 10 vbox's across 3 machines, and manage them with phpvirtualbox.
[11:47] <netore> so whats this error failed to detect a valid IP address from -1
[11:47] <brendand> netore, when does it happen and where do you see it?
[11:47] <netore> http://IP/MAAS/#/dashboard
[11:50] <brendand> just there, on the dashboard?
[11:50] <netore> yup
[11:50] <netore> too early for errors, its a fresh install.
[11:50] <brendand> strange. does it look like some kind of notification?
[11:50] <netore> nah, says error occurred.
[11:51] <netore> was i supposed to install as a rackcontroller or just regular controller.
[11:51] <brendand> netore, how did you install?
[11:52] <netore> i ran the setup as regional controller
[11:52] <netore> from usb... :|
[11:52] <brendand> netore, oh right
[11:53] <brendand> can you do dpkg -l | grep maas-rack-controller?
[11:53] <netore> yeah
[11:54] <netore> -> % dpkg -l | grep maas-rack-controller
[11:54] <netore> ii  maas-rack-controller               2.1.3+bzr5573-0ubuntu1~16.04.1     all          Rack Controller for MAAS
[11:54] <brendand> ah that should be ok then
[11:54] <brendand> i've not seen this error before though
[11:55] <brendand> it should be harmless i think..
[11:55] <brendand> if you can send a screenshot, just to make sure
[11:56] <netore> http://imgur.com/a/QpOqr
[11:57] <netore> that gray over nodes is misleading... my mouse was just over nodes
[11:57] <netore> http://imgur.com/rGDiZpo
[12:00] <brendand> cat /etc/maas/regiond.conf
[12:00] <netore> database_host: localhost
[12:00] <netore> database_name: maasdb
[12:00] <netore> database_pass: x1hfpl9sQLR8
[12:00] <netore> database_user: maas
[12:00] <netore> maas_url: http://192.168.0.24:5240/MAAS
[12:01] <netore> you think its just laptop hardware being laptop hardware?
[12:01] <netore> its a toshiba satellite
[12:02] <brendand> netore, can you 'database_port: 5432' to that file and run 'sudo systemctl restart maas-regiond'
[12:03] <netore> how do i do that first part? database prot
[12:04] <netore> just see if it is accepting connections
[12:05] <brendand> netore, word missing 'can you *add* database_port: 5432 to that file' i meant to say
[12:05] <netore> what file
[12:06] <netore> oh ok
[12:06] <netore> region...
[12:06] <brendand> to /etc/maas/regiond.conf
[12:07] <netore> aiight i got that... then it did device discovery... and errored...
[12:07] <netore> maybe i just need to throw another node up and then try the device discovery?
[12:08] <netore> but all i am doing is hitting the MAAS buttin on the page... and it returns an error... maybe its erroring because its a controller without any nodes? maybe its just harmless like you said.
[12:09] <netore> im gonna throw up a pxe boot lappy n see if it still trips.
[12:12] <netore> fml who only has one enet chord in the house. aiight one sec... i still need to run phpvbox on maas and im too drunk and got too much riding on this. if it works i still got you 5 $ for the bit of info you already provided.
[12:14] <netore> if you get me running phpvbox i got you some more $, i can send the 5 as soon as i get a node attatched or this error gone.
[12:14] <netore> but if you get me runnin phpvbox which i dont understand why i cant run... half a case maybe but either way i appreciate the help n i got some loot on my fb
[12:25] <netore> fuhk i just tried to pxe boot another machine and nothing
[12:43] <zeestrat> netore: I'm back. You still alive?
[12:43] <netore> hell yeah
[12:43] <netore> bit drunker
[12:48] <zeestrat> sounds like a great idea. quick question, what is your host os? Did I understand correctly that you have a laptop with some vbox guests and one of them is running ubuntu as your MAAS controller?
[12:48] <netore> nah
[12:48] <netore> im running a toshiba satelliete, with ubuntu 16 on it... thats it.
[12:48] <netore> i also have another laptop ready to go with no os
[12:49] <netore> i want to run about 10 vboxes across the 2 maas laptops. and manage them from phpvbox
[12:52] <zeestrat> gotcha. do you have a hard requirement to use phpvbox and other laptops? as you're under a deadline and I have no experience with MAAS controller vbox nodes, I was going to recommend a tutorial that uses KVM (pretty much the same as virtualbox, but MAAS has support for controlling KVM guests out of the box) on a single node.
[12:53] <netore> i dont need maas in a virtualbox.
[12:53] <netore> i am creating maas to host vbox.
[12:53] <netore> there will never be a node that is a virtualbox.
[12:54] <netore> i am creating the cluster so that i can host 10 vbox's across 2 machines.
[12:55] <netore> lets just forget the whole virtualbox thing completely.
[12:55] <netore> i just need to make maas work on two laptops.
[12:56] <netore> Error occurred                        failed to detect a valid IP address from -1
[12:59] <zeestrat> I haven't seen that error message before either.
[12:59] <netore> yeah im starting to think its because its a laptop.
[13:00] <netore> i tried pxe booting another machine.. so its not harmless because it just wont boot
[13:00] <zeestrat> I've used this tutorial (https://www.youtube.com/watch?v=ojTTgrtl-RU&index=2&list=PLvn2jxYHUxFlxNmc1dAbw524aoPmHxNpC) to setup a MAAS environment (MAAS controller and some nodes to deploy) with KVM which worked really well.
[13:01] <netore> basically that is running a few maas across kvm, so like if i had amazon aws, i could make a few kvm and have them all work together.
[13:01] <netore> i am using real hardware.
[13:03] <netore> i want to dedicate the whole machines to maas not just a kvm virtual enviornment... once i get maas with some nodes, i plan to INSTALL vbox on the cluster.
[13:03] <zeestrat> You can also have the laptop be your KVM host. It makes it a bit easier to control/experiment with an isolated network that MAAS runs DHCP on.
[13:03] <netore> i cant, i am going to be scaling this to racks. each rack will be in the cluster, all real hardware.
[13:04] <netore> so lets say i have 3 laptops... each one will be completly dedicated to maas.
[13:04] <zeestrat> gotcha. just a recommendation if you needed to get a proof of concept out of the door under a deadline instead of troubleshooting network issues.
[13:05] <netore> i have 2 laptops and a dell server, no point in running tests at this point, i need at least 1 node and one controller.
[13:06] <netore> then im hoping to get more machines and just pxe boot them into the cluster and keep scaling.
[13:06] <netore> if i understand right, maas will let me use all of the hardware for one process.
[13:07] <netore> so if i run... /usr/bin/minetest all of the computers will lend ram and proccessor to that process.
[13:08] <netore> thats what this is really for so i can play minetest with the rescources of 3 computers lololol nah its not lololololol
[13:09] <zeestrat> Not sure if I'm following you on the process part. MAAS lets you manage and deploy operating systems such as ubuntu to machines (most often bare metal).
[13:11] <netore> hold on.
[13:11] <netore> a cluster means that all of the computers share rescources
[13:11] <netore> 3 computers = 1 combined hd, 1 combined ram, 1 combined processor.
[13:14] <zeestrat> I'd say that depends quite on the type of cluster. Anyway, MAAS just spins up individual machines for you. The rest you have to do yourself.
[13:15] <netore> so wait, if i have a maas controller, its just going to allow me to auto pxe boot n install machines on each comp i set to pxe boot?
[13:16] <brendand> netore, not quite that automatically
[13:16] <netore> i want a cluster, a sharing of rescources
[13:17] <netore> fml
[13:17] <netore> MAAS delivers the fastest OS installation times in the industry thanks to its optimised image-based installer. Setup the RAID and Network configuration you want through the MAAS web UI or CLI, then press a button and come back in minutes to a fully-deployed OS.
[13:18] <brendand> netore, ?
[13:20] <netore> astrix?
[13:20] <netore> i was wanting a cluster.
[13:20] <netore> fml
[13:22] <andrew-ii> I think maas will happily manage the metal in your cluster, but you'll need to put cluster software on it
[13:23] <netore> cluster software.
[13:23] <netore> ?
[13:24] <andrew-ii> Like, it'll let you manage the machines, but you'll need to use it to deploy a cluster onto that metal (like, it'll let you manage the network for the cluster and the OS the cluster will run on, but you still need software)
[13:24] <netore> HA cluster
[13:24] <andrew-ii> For example, I'm fiddling around with Juju and Openstack
[13:24] <netore> yes
[13:24] <netore> maybe thats why i am getting confused is openstack what lets you cluster.
[13:24] <andrew-ii> I use MAAS to manage the machine configurations like network cards, hard drive designations, and what the subnets and VLANs and such are
[13:25] <andrew-ii> Then Juju asks MAAS for a machine to be a controller, and that controller asks MAAS for more machines
[13:25] <andrew-ii> Then I ask Juju to put Openstack bits on the machines it requested
[13:26] <andrew-ii> And then Openstack lets me do the cool cloud shenanigans
[13:26] <brendand> netore, sort of. You'd use something like juju to deploy the openstack components on a MAAS (MAAS locates and manages the bare metal systems for you)
[13:26] <netore> ok so i am in the right area... but i just cant stop at maas, i need to have juju and openstack...
[13:26] <netore> here is my end goal.
[13:26] <andrew-ii> For cluster compute? Probably. I think there's some bundles set up for that...
[13:27] <netore> i want to be able to run 3 computers, and have 10 vbox's share the rescources across those 3 computers.
[13:28] <andrew-ii> I haven't tried virtualizing across a maas before. Sounds neat. Openstack sorta breaks the problem up a bit more than that I think
[13:29] <andrew-ii> Part of the issue is that Openstack kinda needs a pile of machines to really get set up easily
[13:29] <andrew-ii> You can do three though... one sec
[13:30] <andrew-ii> See if this is at all informative: http://marcoceppi.com/2014/06/deploying-openstack-with-just-two-machines/
[13:30] <netore> aiight i got three machines to throw at this
[13:30] <tychicus> netore: do you need to virtualize any windows machines?
[13:30] <netore> possibly.
[13:30] <andrew-ii> Also I was wrong - the Juju charm in the charmstore is Minecraft, not minetest, sorry
[13:31] <netore> lololol
[13:31] <tychicus> ok, if you did not, you could deploy nova-lxd
[13:31] <netore> i dont want to not be able to run any vbox i want...
[13:31] <netore> my end game is just to take all of these laptops and comps i got laying around, and make one machine out of them
[13:32] <netore> even if i go buy a brand new server, im still going to want to add another brand new server and have them share rescources
[13:32] <tychicus> may be a little tricky, as maas has specific hardware requirements that it needs for provisioning
[13:32] <andrew-ii> I think you're still going to end up with stuff running on one machine, but if that machine dies, it'll likely be able to spin up elsewhere
[13:33] <tychicus> I am guessing that your laptops don't have out of band management, intel AMT or impi
[13:33] <netore> i have run a process on freebsd and had two computers have at it... but it was limited... i was sure ubuntu openstack and juju solved this...
[13:33] <netore> i dunno but i can get machines specific for this.
[13:34] <netore> i thought maas was the cluster, and juju was for deploying apps like wordpress and openstack was the management
[13:35] <tychicus> people have done really well with intel nuc, http://blog.dustinkirkland.com/2013/11/review-ubuntu-and-intel-nuc.html
[13:36] <tychicus> or any of the xeon D based server's should work
[13:37] <netore> let me chek the specs on that machine i just got
[13:37] <brendand> tychicus, you have to be careful in choosing your nuc
[13:38] <netore> thats fine... as long as when i run 1 new vbox it shares the ram and cpu of all the nodes
[13:38] <tychicus> netore: it doesn't quite work that way
[13:39] <netore> int3l zeon e3-1270 v5
[13:39] <netore> fml
[13:39] <zeestrat> netore: Pooling resources does not really work that way outside of supercomputers. Most clusters are based on the smallest common denominator (i.e. your individual servers)
[13:40] <tychicus> :brendand, I think the only one that supports AMT is the DC53427HYE
[13:41] <netore> daym.
[13:41] <zeestrat> I'd look at your workload and see how you can split it up as smaller individual hosts/jobs and then run them in a kubernetes cluster or openstack (though note that you really need to be sure that is something you want as it requires a bunch of resources and knowledge)
[13:41] <netore> kubernetes cluster let me look that up rq
[13:44] <netore> beowulf
[13:44] <netore> is what i ran in freebsd.
[13:45] <netore> but it was limited to only a few programs
[13:46] <tychicus> netore: MaaS does not solve that particular problem for you.  MaaS (Metal as a Service) used for provisioning an OS onto bare metal, juju service orchestration, allows you to quickly install complex applications across multiple machines, whether physical or virtual
[13:47] <netore> aiight fellas, sorry for wasting your time... i thought maas was the new equivilent to beowulf.
[13:47] <netore> fml
[13:48] <netore> i just dont understand why maas wouldnt solve this.
[13:48] <netore> especially when i seen how juju works, i thought it was a no brainer.
[13:48] <tychicus> you could write a juju charm, that installs beowulf on all of the machines that have been commissioned by MaaS
[13:49] <netore> really lolol
[13:49] <netore> your grabbing me bak in... lolol
[13:50] <tychicus> but nothing in maas or juju will automagically aggregate the compute resources for all of the machines
[13:50] <tychicus> they just provide a platform that makes managing machines easier
[13:50] <netore> and as far as i know beowulf is limited to gcc  processes
[13:51] <netore> yeah, i gotta figure out something before 1, i fuhken had this in the bag i thought.
[13:56] <netore> The cluster is now complete. To take advantage of the clustering, open a terminal and type "mpirun -np # PROGRAM" where "#" is the number of processes/threads to create and "PROGRAM" is the program or script to run on the cluster.
[13:57] <netore> with mpi i could run the command mpirun....# virtualbox.
[14:42] <zeestrat> Hey, what's an easy way to list out the system id of all nodes that are ready or new (not including the rack/region controllers) via the CLI?
[14:52] <BlackDex> zeestrat: Thanks for your answer btw.. I resolved it by using some jq magic ;)
[14:53] <BlackDex> i only needed to give the correct machine-id and it changed all bond interfaces to an other mode-mode
[14:53] <zeestrat> BlackDex: np :) Glad to hear you got something working
[14:54] <BlackDex> would have been a pain in the a** using the gui
[14:54] <BlackDex> you first need to remove the interface and re-create the bond again if you want to change the bond-mode
[14:54] <roaksoax> zeestrat: try maas admin machines read status=<status>
[14:54] <BlackDex> doing that at 9 machines, with 4 bonded interfaces.
[14:54] <zeestrat> yeah, I created a bug on the gui stuff for editing bonds. We're mostly using playbooks with CLI to configure everything anyway
[14:55] <zeestrat> roaksoax: thanks, I'll give it a go
[14:55] <roaksoax> zeestrat: new=0, ready=4
[14:55] <roaksoax> zeestrat: so maas admin machines read status=0
[14:55] <roaksoax> zeestrat: so maas admin machines read status=4
[14:55] <BlackDex> you have a playbook for me to look at?
[15:01] <zeestrat> BlackDex: EOD here, but pm me and I can get back to you :) They are not optimal, but get the job done.
[15:02] <BlackDex> sure
[15:02] <BlackDex> :)
[15:03] <BlackDex> done :) have a nice weekend :)
[15:04] <zeestrat> Cheers, same to you!
[15:40] <mup> Bug #1675823 opened: [2.2] ProcessGroupLeaderMixin races with spawnProcess <MAAS:In Progress by allenap> <https://launchpad.net/bugs/1675823>
[16:25] <mup> Bug #1675838 opened: [2.2 ui] OS/Release section of Filter By panel doesn't show text for the different options <MAAS:New> <https://launchpad.net/bugs/1675838>
[16:37] <mup> Bug #1675844 opened: MAAS takes ipv6 ULA address as a primary one <MAAS:Triaged> <https://launchpad.net/bugs/1675844>
[16:46] <mup> Bug #1675844 changed: MAAS takes ipv6 ULA address as a primary one <MAAS:Triaged> <https://launchpad.net/bugs/1675844>
[16:55] <mup> Bug #1675844 opened: [2.1] MAAS takes ipv6 ULA address as a primary one <MAAS:Triaged> <https://launchpad.net/bugs/1675844>
[18:34] <mup> Bug #1675887 opened: [2.1] Commissioning fails to detect drives. <MAAS:Incomplete> <https://launchpad.net/bugs/1675887>
[19:58] <ybaumy> https://bugs.launchpad.net/maas/+bug/1673724
[19:58] <ybaumy> guys i need that patch in devel/next
[19:58] <ybaumy> that would be cool
[20:07] <ybaumy> it would be more than cool because i need to show a working maas environment. and everything works for me except this and this is really important
[20:07] <ybaumy> so. could someone release the patch into next
[20:08] <ybaumy> is andres rodriguez here?
[20:17] <pmatulis> ybaumy, as a hack, i believe you can make a one-line change to a single Python script
[20:17] <ybaumy> pmatulis: which script?
[20:18] <pmatulis> /usr/lib/python3/dist-packages/provisioningserver/refresh/node_info_scripts.py
[20:19] <pmatulis> line 9 here:
[20:19] <pmatulis> http://paste.ubuntu.com/24237785/
[20:19] <pmatulis> you will need to re-add the node
[20:19] <pmatulis> afterwards
[20:20] <ybaumy> gonna try it. give me a few minutes
[20:23] <pmatulis> last minute friday hack job!
[20:26] <ybaumy> ok commission is running.
[20:27] <ybaumy> that will take just a few minutes
[20:31] <ybaumy> pmatulis: do i have to restart daemons for that to work because i didnt restart anything and it didnt work .. its still /dev/sdb as boot device
[20:31] <pmatulis> ybaumy, i didn't hear about such a requirement
[20:31] <ybaumy> pmatulis: then it doesnt work
[20:31] <pmatulis> ybaumy, this is for failed Deployments right?
[20:32] <ybaumy> no its for commissioning
[20:32] <pmatulis> damn, sorry
[20:32] <ybaumy> i get /dev/sdb as boot or the last of the disks
[20:32] <pmatulis> possible to go with just single disk machines for your demo?
[20:33] <ybaumy> pmatulis: nope there are ceph-osd machines i want to show
[20:33] <pmatulis> darn
[20:33] <ybaumy> yep this sucks
[20:33] <pmatulis> what PPA are you using? maybe there is something bleeding edge that works
[20:33] <ybaumy> next ppa
[20:33] <pmatulis> not sure where next-proposed is at
[20:34] <ybaumy> bleeding edge sounds like it would break something else. and currently else is working you undestand that i dont want to mess with it if i could
[20:34] <pmatulis> fwiw, https://launchpad.net/~maas/+archive/ubuntu/next-proposed
[20:34] <pmatulis> yes, i understand
[20:35] <ybaumy> is it fixed in proposed?
[20:35] <pmatulis> i'm looking at the dates, march 13
[20:36] <pmatulis> roaksoax, any comment? ^^^
[20:36] <pmatulis> next and next-proposed look identical to me
[20:36] <pmatulis> https://launchpad.net/~maas/+archive/ubuntu/next
[20:37] <ybaumy> then i dont have to change the ppa
[20:37] <mup> Bug #1675915 opened: [2.2] Script selection doesn't show scroll bar <hardware-testing> <ui> <MAAS:Triaged> <https://launchpad.net/bugs/1675915>
[20:37] <pmatulis> right
[20:37] <mup> Bug #1675919 opened: [2.2] Commissioning and Testing tabs should show spinner while running <hardware-testing> <ui> <MAAS:Triaged> <https://launchpad.net/bugs/1675919>
[20:38] <pmatulis> i gotta run. be back later tonight
[20:38] <pmatulis> good luck ybaumy
[20:38] <ybaumy> pmatulis: thanks dude
[20:38] <ybaumy> later
[20:48] <ybaumy> what is the nick of andres rodriguez
[20:55] <brendand> ybaumy, roaksoax
[20:58] <ybaumy> ok
[20:58] <ybaumy> in which timezone is he in
[21:02] <brendand> ybaumy, if you want it now you can use https://launchpad.net/~maas-maintainers/+archive/ubuntu/experimental3 but that's a testing ppa and i would *not* recommend using that for any setup you want to be reasonably stable
[21:03] <brendand> ybaumy, the release is not going to be expedited because you need one patch
[21:03] <ybaumy> brendand: no i wont. because i dont want to break other things now that its working for me
[21:03] <brendand> ybaumy, well atm that's what's going into /next anyway
[21:04] <brendand> ybaumy, early next week after a little more testing
[21:04] <ybaumy> brendand: hmm that would be cool early next week. on wednesday i would need it
[21:05] <brendand> ybaumy, fyi releases are done in a cadence, not ad-hoc/on-demand
[21:05] <ybaumy> brendand: i fully understand this.
[21:05] <roaksoax> ybaumy: 2.2beta4 is cut
[21:06] <roaksoax> should help fix a few issues
[21:06] <ybaumy> brendand: i dont like to pressure too. but i want to show management that we have something here we can use .. you must understand
[21:06] <roaksoax> ybaumy: you can use either ppa:maas/proposed -> 2.1.5
[21:06] <roaksoax> ybaumy: or 2.2b4 will soon land in ppa:maas/next-proposed
[21:07] <roaksoax> just doing a final round of testing
[21:07] <ybaumy> roaksoax: ok
[21:07] <ybaumy> roaksoax: thanks for the information
[21:08] <roaksoax> ybaumy: 2.1.5 is ready to be released, just prioritizing 2.2b4 atm
[21:08] <ybaumy> roaksoax: can i downgrade from 2.2.0~beta3+bzr5815
[21:09] <roaksoax> ybaumy: uhmmm i can't remember but i think there has been migrations from 2.2b2 to 2.2b3, which means you probably cant
[21:09] <roaksoax> we dont do backward migrations
[21:09] <ybaumy> roaksoax: then i have to wait
[21:09] <roaksoax> ybaumy: what are your issues with 2.2b3 ?
[21:10] <ybaumy> roaksoax: /dev/sdb or the last disk is the boot disk
[21:10] <roaksoax> ybaumy: you can easily change that
[21:10] <ybaumy> roaksoax: not /dev/sda
[21:11] <roaksoax> ybaumy: maas <user> block-device set-boot-disk <system-id> <block-id>
[21:11] <roaksoax> maas <user> machine set-storage-layout <system-id> storage_layout=flat
[21:11] <roaksoax> where block-id is the id of /dev/sda
[21:13] <ybaumy> roaksoax: that what im doing right now already. you have to see that i need a management ready presentation where i just add the nodes and click commission and then show them how to deploy cloud with juju
[21:13] <ybaumy> roaksoax: if i do workarounds they will say that its not working
[21:13] <ybaumy> roaksoax: i know them
[21:13] <roaksoax> ybaumy: this is not a workaound tough. This is valid user configuration
[21:13] <roaksoax> there's no way to know which one is the boot disk
[21:14] <roaksoax> set by the bios
[21:14] <roaksoax> maas just makes an assumption
[21:14] <ybaumy> roaksoax: hmm thats not good at all
[21:14] <roaksoax> and this assumption is now the last disk because of lsblk
[21:14] <ybaumy> im using vmware
[21:14] <ybaumy> and you have the scsi ids or not?
[21:15] <roaksoax> ybaumy: yup, that's lsblk listing sdb before sda
[21:15] <roaksoax> ybaumy: while MAAS knows that it is poewr managing vmware VM's, it uses them as regular machines
[21:16] <ybaumy> roaksoax: hmm so there is not way to change that behaviour?
[21:17] <roaksoax> lsblk's ? yeah we are investigating on it
[21:17] <ybaumy> ybaumy: i have to write an own website for that .. triggering the correct commands then
[21:18] <roaksoax> either way, what we did in maas is try to ensure that we always return the first discovered disk, so thta it gets set as the book device
[21:21] <ybaumy> hmm i dont know what to do right now
[21:21] <ybaumy> will have to think this over
[21:21] <roaksoax> ybaumy: you can patch maas yourself directly if you want
[21:21] <ybaumy> roaksoax: im not a programmer
[21:22] <ybaumy> roaksoax: i dont want to mess with things i cant understand or dont understand fully
[21:22] <roaksoax> ack!
[21:23] <roaksoax> ybaumy: either way, I just pushed b4 to next-proposed
[21:23] <ybaumy> roaksoax: i have to ask a colleague from work .. he knows python and maybe he can do something
[21:23] <roaksoax> ybaumy: it should be available in the next 30 mins
[21:23] <ybaumy> roaksoax: thanks
[21:24] <roaksoax> ybaumy: you will need to re-add the machine and re-commission it, and do please let me know if that made any idfferent
[21:24] <roaksoax> we've tested against hardware and VM's here, and tests were successful
[21:24] <ybaumy> roaksoax: will try as soon as i have the new relsase
[21:25] <roaksoax> ybaumy: sudo lsblk --exclude 1,2,7 -d -P -o NAME,RO,RM,MODEL,ROTA,MAJ:MIn -x MAJ:MIN -> if you can try that on the VM to see how it lists the disk
[21:26] <roaksoax> then i'll know if it works for you or not
[21:31] <ybaumy> roaksoax: i ran the command on already deployed VM's and sda is always first
[21:32] <roaksoax> ybaumy: ok cool, so that will work
[21:32] <roaksoax> ybaumy: what if you run: sudo lsblk --exclude 1,2,7 -d -P -o NAME,RO,RM,MODEL,ROTA
[21:33] <ybaumy> roaksoax: then same
[21:34] <roaksoax> uhm maybe this is not an lsblk issue after all and a kernel thing
[21:35] <ybaumy> roaksoax: we will see when i got the patch applied .. readded the VM's and commissioned them. i have 10 VM's so it should be enough to see whats going on
[21:42] <zeestrat> roaksoax: Dumb question, but does that bug show up only during commissioning? We've hit something like this in https://bugs.launchpad.net/maas/+bug/1644856 in 2.0.0, but there MAAS correctly commissioned them, but upon switched it up on deployment and chose the last device device as boot device.
[21:42] <roaksoax> ybaumy: ok, i just tested this against a kvm machine where sdb would be the root disk, now it is making sda as the root disk
[21:42] <roaksoax> ybaumy: so you should be good to go too
[21:43] <ybaumy> roaksoax: lets see. im waiting
[21:44] <roaksoax> zeestrat: without looking at it, I'd say that thiscould be hardware related. For example, during commissioning the OS finds hd0 as sda, but in a subsequent boot, hdx becomes sda but MAAS depends on the serials and such
[21:44] <roaksoax> zeestrat: but would be indeed interesting to investigate
[21:46] <roaksoax> zeestrat: so maas , at deployment time knows that what it know as sda, is no longer sda and it i something else
[21:46] <roaksoax> zeestrat: but that's just a wild guess
[21:46] <roaksoax> zeestrat: if you could also attach this, it would be great:
[21:46] <roaksoax>  maas <user> machine get-curtin-config <system_id_of_machine>
[21:46] <roaksoax>  - maas <user> node-results read system_id=<system_id> name=00-maas-07-block-devices.out | grep "\"data\"" | cut -d"\"" -f4 | base64 --decode
[21:46] <roaksoax>  - maas <user> node-results read system_id=<your_system> result_type=1 | grep "\"data\"" | cut -d"\"" -f4 | base64 --decode
[21:59] <zeestrat> roaksoax: Thanks I'll note that down and try to replicate it. Only happens some times which is annoying.
[22:01] <zeestrat> roaksoax: Anyways, thanks for all the help. Have a good weekend!
[22:02] <ybaumy> roaksoax: patch is here. commissioning is running ..
[22:06] <ybaumy> roaksoax: :D it works .. yes sir
[22:06] <ybaumy> roaksoax: all ten are /dev/sda
[22:06] <ybaumy> roaksoax: im so happy right now
[22:07] <ybaumy> roaksoax: there are machine with 2 4 8 disks and all are correctly set
[22:10] <ybaumy> running juju deploy ceph-osd right now
[22:26] <ybaumy> roaksoax: i have to call it quit right now. deploy works now too. will double check tomorow. will drop you a PM if i have problems but really looks good right now. thanks for the quick help. i really appreciate this
[22:27] <ybaumy> will be here around 0800 tomorow
[22:53] <mup> Bug #1675953 opened: [2.2] Space constraint matching should allow matching the undefined space <MAAS:In Progress by mpontillo> <https://launchpad.net/bugs/1675953>