[07:20] does anyone know why a bunch of machines would go allocated but not start deploying? We had 5 servers from 25 deploy, but the rest are stuck in allocated === frankban|afk is now known as frankban [09:19] we seems to be having an issue with MAAS not releasing Ip's it has observed, does anyone have an ideas on how to flush observed ip's from a subnet's ip use? [10:10] we hacked the DB to make it work [11:12] Hi all, I is it possible to increase a VM RAMS after the maas deployment ? [11:45] mah, probably, but may require recommissioning for maas to detect it [11:45] maybe not [11:46] when I tried to edit it from the maas UI [11:47] I found it blocked can not be edited [11:47] So is there another way to configure it [11:47] or I have to deploy MAAS again from begining ? [11:48] mah, oh well maas does not control the RAM, you would have to do that with the VM management sw [11:48] maas is not creating the machines [11:48] hmmm [11:48] mah, which virtualization sw are you using? [11:48] and after that I have to do commissioning ? [11:49] mah, maybe - maas may just detect the change, but i doubt that it will unless the machines changes state, e.g. by recommissioning or deploying [11:50] in that case I need to do recommissionig the changed machine only not all the nodes [11:50] right ? [11:50] mah, oh for sure [11:50] ok fine [11:50] mah, it will only be needed for that node [11:50] Thanks a lot [11:51] I have node 0 and it has 3 VMs [11:51] but I will do these changes on one VM only [11:51] so just recommissiong that VM not all the 3 [11:52] I am using Virtual Machine Manager [11:53] mah, right, so kvm - you can either modify the ram through the ui or use the virsh command line tool [11:53] Ok [11:53] thanks [11:56] mah, bonus - here's how to change the memory on a running machine: [11:57] virsh destroy vm-name [11:57] virsh setmaxmem vm-name 4GB [11:57] it is not running :) [11:57] virsh start vm-name [11:57] but thanks for that bonus too [11:57] virsh setmem vm-name 4GB [11:58] setmem only works when it's running [11:58] hmmm [11:58] and setmaxmem doesn't [11:59] so setmaxmem works when it is not running ? === Guest21645 is now known as ahasenack === ahasenack is now known as Guest91977 [12:19] mah, yes - you can't setmaxmem on a running vm [13:23] Bug #1637182 opened: Help and documentation 'list of unicodes' inconsistent [13:23] Bug #1637183 opened: New subnets and not_subnets allocate params missing from help and docs [13:32] Bug #1637192 opened: Allocate using subnets or not_subnets with space fails === frankban is now known as frankban|afk [13:53] Bug #1637183 changed: New subnets and not_subnets allocate params missing from help and docs [13:53] Bug #1637198 opened: Unrecognised params silently ignored over api [13:53] Bug #1637203 opened: Allocate using subnets params gives unhelpful errors === frankban|afk is now known as frankban [15:31] so, a slightly philosophical question here... should i have MAAS do as much setup as possible or should i have Ansible or Puppet do things after the server boots up? [15:31] I've always been a proponent of having a server boot up in a ready state, ready to go, but i am having issues achieving some things with MAAS. [15:32] for example, I need to install Mellanox 40 gigabit interface drivers but ,unfortunately, it bombs in MAAS because Mellanox installer needs to compile a bunch of stuff [15:34] solefald: i'd suggest get maas to do as much setup as possible [15:34] and without having 40 gig network, I cant configure and install a few things, which holds me back from installing some other things. [15:34] solefald: that would match what MAAS knows about vs what the machine actually reflects [15:38] Bug #1637198 changed: Unrecognised params silently ignored over api [16:03] Bug #1637246 opened: MaaS should use configured names for VLAN interfaces [16:03] Bug #1637246 opened: MaaS should use configured names for VLAN interfaces [16:30] Bug #1637256 opened: MaaS might disrupt network connectivity by probing DHCP services === frankban is now known as frankban|afk [16:45] Bug #1637198 opened: Unrecognised params silently ignored over api [16:51] Bug #1637198 changed: Unrecognised params silently ignored over api [17:12] Bug #1637198 opened: Unrecognised params silently ignored over api [17:15] Bug #1637198 changed: Unrecognised params silently ignored over api === zz_CyberJacob is now known as CyberJacob === CyberJacob is now known as Guest22220 [18:31] solefald: im finding it easier to have maas just do a minimal install and ansible do everything else [18:35] shubjero: yeah, im leaning this way. i can get way fancier with Ansible and keep everything in one place. [18:36] same [18:37] some of the stuff maas uses like curtin is poorly documented, i cant make heads or tails of things [18:37] shubjero: i think you meant "not documented at all" :) [18:37] or the way maas does certain things like nic bonding, jumbo frames.. doesnt work properly [18:37] so we deploy it with maas but then use ansible to overwrite /etc/network/interfaces with a config that seems to work better [18:38] thats exactly where im stuck now. converting it all to Ansible at the moment [18:38] lots of little things [18:40] installing mellanox 40 gig drives and bonding into 80 gig interface, and changing MTU's everywhere, etc. MAAS just bombs. [18:40] plus i hate the fact that there is no easy way to test the preseed file. have to release/deploy my test box literally 100 times a day [18:41] 1 typo and 20 minutes are gone [18:43] haha yeah, espescially painful with bare metal [18:43] you can only test so many things with a vm [18:43] at some point you need to move to the real thing where you have like you say, unique 40g interfaces & etc [18:43] btw, im curious what you are doing with servers equipped with 80Gbps [18:44] I work in cancer research and admin a large openstack cluster, our servers are 10Gbps and 40(4x10) connected [18:44] 40Gbps for ceph nodes [18:44] 10Gbps for compute nodes [18:47] deep learning cluster. rendering and compute nodes [19:29] Hi, I'm trying to pass a "mount" command using /etc/maas/presseed/curtin_userdata, but - it fails the deployment. I've tried multiple syntax'es, no success though. [19:30] e.g mount_data: ["curtin", "in-target", "--", "sh", "-c", "mount", "192.168.1.1:", "/data", "/data"] [19:30] I was able to install nfs-common via presseed, so the nfs client is there (the provisioned system). [19:31] oh, curtin [19:58] GA: try specifying filesystem [21:01] Bug #1621507 opened: initramfs-tools configure_networking() fails to dhcp ipv6 addresses (Ubuntu):Fix Released> [21:01] [21:04] Bug #1621507 changed: initramfs-tools configure_networking() fails to dhcp ipv6 addresses (Ubuntu):Fix Released> [21:04] [21:07] Bug #1621507 opened: initramfs-tools configure_networking() fails to dhcp ipv6 addresses (Ubuntu):Fix Released> [21:07] === Guest22220 is now known as CyberJacob [22:16] Bug #1636969 opened: [1.9] Multiple negative spaces constraints given and rejected by MAAS [22:28] Bug #1636969 changed: [1.9] Multiple negative spaces constraints given and rejected by MAAS [22:34] Bug #1636969 opened: [1.9] Multiple negative spaces constraints given and rejected by MAAS === CyberJacob is now known as zz_CyberJacob