| axw | stokachu hallyn: yep I agree, it's horrible (predates my involvement), but I couldn't see a way around it. the OVA is fetched locally, split into OVF & VMDK, then uploaded to vsphere. AFAICT there's no way to get vsphere to do that for you | 00:15 |
|---|---|---|
| stokachu | axw: interesting | 00:17 |
| stokachu | axw: and it happens prior to bootstrap so we couldnt cache it anyway | 00:18 |
| axw | stokachu: right. one thing we *can* do, which I replied to an email about the other day, is reuse the OVF/VMDK cached in the datastore after bootstrapping | 00:18 |
| axw | to avoid having hte controller download the OVA for each new machine | 00:19 |
| axw | that bit's doable, but doesn't affect UX nearly as much | 00:19 |
| stokachu | axw: ah ok | 00:19 |
| stormmore | I wonder if I can get k8s to run on top of openstack on lxd | 01:20 |
| === skay is now known as Guest63699 | ||
| Ting_ | Does some have use juju on aws with federated user (short-live uesr) ? | 07:53 |
| === tinwood_ is now known as tinwood | ||
| D4RKS1D3 | Hi someone knows why when I add a unit in a machine that the state is ready, this machine waste around 200 seconds to running up? | 10:34 |
| D4RKS1D3 | If I execute more machines the time waiting between ready and deploying usually is more than 35 minutes | 10:34 |
| magicaltrout | internet bandwidth? | 10:40 |
| D4RKS1D3 | magicaltrout, this answer is for me? | 10:40 |
| magicaltrout | well, its a guess for you | 10:43 |
| magicaltrout | i dunno, but of course when you spin up new nodes, they all download a bunch of stuff. juju, updates etc | 10:43 |
| === salmankhan1 is now known as salmankhan | ||
| D4RKS1D3 | But I am not talking about preparing the node... I am talking about turn on the node | 11:19 |
| === Guest63699 is now known as skay | ||
| orf | hey, is there any way to use the JuJu VSphere provider with a specificed Resource Pool? | 13:44 |
| orf | we have a namespaced system and need to create everything within a 'Development' resource pool | 13:44 |
| orf | I had a browse of the source, and it seems to be able to: https://github.com/juju/juju/blob/d28f92df4051a02602347b73335467a29f5e7327/provider/vsphere/internal/vsphereclient/createvm.go#L139 | 13:44 |
| orf | there just is no documentation as to how to configure that | 13:44 |
| hallyn | the docs are in createvm.go... around line 140 iirc | 14:47 |
| hallyn | j/k | 14:47 |
| hallyn | yeah i'm hoping the vsphere controller gets some love | 14:47 |
| === primeribz is now known as rahworkx | ||
| orf | I gave up hallyn | 15:24 |
| orf | Cant work out how to define `ComputeResource.ResourcePool` | 15:24 |
| hallyn | :( | 15:29 |
| === frankban is now known as frankban|afk | ||
| stormmore | o/ | 17:42 |
| === mjs0 is now known as menn0 | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!