| sncder | hi, guys, I want to build my own img used for application deploy. does anybody know how to that? | 02:52 |
|---|---|---|
| sncder | any one here? | 03:04 |
| mup | Bug #1704517 changed: Deploy status of 2 nodes <logs> <MAAS:Expired> <https://launchpad.net/bugs/1704517> | 04:18 |
| mup | Bug #1719271 opened: grub package in xenial-proposed will change boot order for a MaaS deployed system <apport-bug> <arm64> <package-from-proposed> <uec-images> <xenial> <MAAS:New> <grub2 (Ubuntu):New> <https://launchpad.net/bugs/1719271> | 09:27 |
| BlackDex | Hello there | 11:15 |
| BlackDex | I'm trying to pxe boot using a vlan configured on the nic, but it seems the systems isn't possible to connect to the iscsi then | 11:15 |
| BlackDex | it does start the commisioning image etc... but after that it tries to connect to the network, and that failes. I Think because it doesn't use a VLAN. | 11:30 |
| BlackDex | How can i fix this? | 11:30 |
| mup | Bug #1719316 opened: Unable to commision/deploy(/enlist) when having a tagged only network setup <MAAS:New> <https://launchpad.net/bugs/1719316> | 14:10 |
| mup | Bug #1719353 opened: [2.3a3, Machine listing] Improve the information presentation of the exact tasks MAAS is running when running hardware testing <ui> <MAAS:New> <https://launchpad.net/bugs/1719353> | 15:46 |
| mup | Bug #1719353 changed: [2.3a3, Machine listing] Improve the information presentation of the exact tasks MAAS is running when running hardware testing <ui> <MAAS:New> <https://launchpad.net/bugs/1719353> | 15:52 |
| mup | Bug #1719353 opened: [2.3a3, Machine listing] Improve the information presentation of the exact tasks MAAS is running when running hardware testing <ui> <MAAS:New> <https://launchpad.net/bugs/1719353> | 16:04 |
| bdx | few architectural questions | 16:15 |
| bdx | the name "rackd" almost makes this overly obvious, that a maas-rackd should run in each physical rack .... but what is the underlying reasoning that I should use to determine when another rackd is needed? | 16:18 |
| bdx | I'm thinking network spaces might be a reason to have separate rack servers | 16:19 |
| bdx | like you have desperate networks and need to put a rack server in each local to facilitate the maas management of those networks | 16:20 |
| bdx | but if you have networks that span racks | 16:21 |
| bdx | and (hypothetically) the servers on each rack are all on the same flat maas-mgmt network | 16:21 |
| bdx | then would I still want a rackd server on each physical rack? | 16:22 |
| bdx | ahh, I think I see | 16:25 |
| bdx | the reasoning for having a rackd on each rack, is for performance | 16:25 |
| bdx | not to reach different address spaces or locals | 16:26 |
| bdx | just simply, performance | 16:26 |
| bdx | cool | 16:26 |
| bdx | hmmm, but if I have multiple rackd on the same mgmt-net, do they coordinate which one offers each service e.g dhcp, pxe, iscsi, ntp? | 16:28 |
| bdx | or do I need to have separate maas-mgmt network for each rack, so each rackd can have an affinity to a address space? | 16:29 |
| bdx | my blades aren't really super high density, each node is a 1U, and I have ~30 nodes per rack | 16:31 |
| bdx | so would 30 nodes really constitute the need for an affinity for their own rackd server? | 16:32 |
| roaksoax | bdx: you dont need 1 rackd for each rack, you can have 1 rackd for multiple racks -> wich is what ppl usually do | 16:33 |
| mup | Bug #1719361 opened: [2.3 alpha 3, Hardware testing, machine listing] Remove success icons for components that passed the tests fill the screen and surface this information in a better way <ui> <MAAS:New> <https://launchpad.net/bugs/1719361> | 16:34 |
| roaksoax | bdx: you can have 2 racks for the same management network if you want HA and things like that | 16:34 |
| bdx | roaksoax: but the rackd on the first rack, and the rackd running on the second rack won't balance the load? its only for HA? | 16:36 |
| bdx | or do they get a vip endpoint between them and get loadbalanced to - something to this effect? | 16:37 |
| roaksoax | bdx: not only for ha, but for pxe booting can happen | 16:37 |
| roaksoax | bdx: you need to enable DHCP for it to work | 16:37 |
| roaksoax | bdx: it wont be loadbalanced | 16:38 |
| roaksoax | bdx: it uses nuderlying isc-dhcp HA feature | 16:38 |
| bdx | roaksoax: I have 4 racks ... ~100 servers that I'll be checking into this maas | 16:39 |
| bdx | I have a 24core/64GBram node I had set aside for a region + rack controller | 16:41 |
| bdx | roaksoax: do you think with just 4 racks, I will have a need to decouple and HA the things? | 16:42 |
| bdx | I was thinking I would just setup another maas for whatever is added next and just leave this maas region+rack to handle these ~100 nodes | 16:45 |
| bdx | so, possibly I don't need to plan this maas to scale really | 16:45 |
| bdx | I'm going to go for an all in 1 region+rack controller for the 4 racks and see if I hit my head on anything | 16:47 |
| bdx | I'll report back here this week sometime | 16:47 |
| roaksoax | bdx you could handle those 100 machines with a single region+rack no problems | 18:16 |
| mup | Bug #1719440 opened: [2.3, UI, regression] Expanding Description text box on Vlan Details page, causes all other editable fields to expand <MAAS:Triaged> <https://launchpad.net/bugs/1719440> | 20:31 |
| mup | Bug #1719473 opened: MAAS adding domains to dns-search field that are unnecessary <cpe-onsite> <MAAS:New> <https://launchpad.net/bugs/1719473> | 22:32 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!