| opendevreview | zhouxinyong proposed openstack/magnum master: Use new get_rpc_client API from oslo.messaging https://review.opendev.org/c/openstack/magnum/+/883820 | 01:35 |
|---|---|---|
| opendevreview | Merged openstack/magnum stable/zed: Fix pep8 gate https://review.opendev.org/c/openstack/magnum/+/883104 | 03:21 |
| opendevreview | Jake Yip proposed openstack/magnum stable/zed: Add `-p` param to `mkdir` in agent startup script https://review.opendev.org/c/openstack/magnum/+/883996 | 04:04 |
| opendevreview | Jake Yip proposed openstack/magnum stable/zed: Fix pods unable to send traffic to ClusterIP https://review.opendev.org/c/openstack/magnum/+/883997 | 04:04 |
| jakeyip | I'm here if anyone needs me | 08:56 |
| mkjpryor | hi all | 09:04 |
| travisholton | hi! | 09:04 |
| mkjpryor | jakeyip: we think that the first batch of patches for the Cluster API driver are ready for review now | 09:06 |
| mkjpryor | it would be good to start getting eyes on them from outside the team | 09:06 |
| travisholton | nice! | 09:07 |
| jakeyip | nice! whre should i start? | 09:12 |
| dalees | o/ | 09:12 |
| jakeyip | hi dalees | 09:12 |
| mkjpryor | jakeyip: https://review.opendev.org/c/openstack/magnum/+/815521 is the bottom of the patch chain | 09:15 |
| jakeyip | ok will take a look | 09:17 |
| mkjpryor | The patches so far implement create, delete, config, resize, add/remove/resize nodegroup | 09:18 |
| mkjpryor | All the patches marked Active are ready for review | 09:18 |
| mkjpryor | There will be additional patches on top to implement upgrade, to respect more of the networking parameters and for autoscaling | 09:19 |
| dalees | mkjpryor: nice, i was just about to ask feature list. Are Cluster resource status messages surfaced/updated into Magnum? | 09:19 |
| mkjpryor | Plus some other stuff probably | 09:19 |
| mkjpryor | dalees: Not at the moment, but would be another easy patch I think | 09:19 |
| mkjpryor | Because Cluster API uses a reconciliation loop, resources don't actually go into a failed state very often | 09:20 |
| mkjpryor | So actually error reporting in general is a little tricky | 09:20 |
| dalees | and where do the Helm charts live - I last saw these hardcoded as pointing at github? Where does the OpenStack Magnum project need these external resources to be? (We need these locally hosted, but can work on that) | 09:20 |
| mkjpryor | We have made the Helm chart repo, name and default version config options | 09:21 |
| mkjpryor | They default to our Helm charts on GitHub | 09:21 |
| dalees | great! | 09:21 |
| mkjpryor | The version can be overridden in a template label, if you want to peg a template to a particular chart version | 09:21 |
| mkjpryor | (not the repo or chart name though, as that is a recipe for code injection!) | 09:22 |
| dalees | mkjpryor: yeah, state will need to be polled or watched from Magnum? It can go to failed state and recover through reconcile, or go into a failed state within ClusterAPI without Magnum being involved in the action. Reconcile will help many, but I'm sure there are at least esoteric permanent failure states to surface ;) We can work on this though, | 09:24 |
| dalees | once we've all got it running. | 09:24 |
| mkjpryor | That is the plan :D | 09:24 |
| mkjpryor | The weird thing is that the two states in the Magnum API are really meant to indicate "did the last change apply correctly" and "is the cluster healthy" | 09:25 |
| mkjpryor | So actually, what we might implement is a custom monitor that looks at the Cluster API resources to determine their current health | 09:26 |
| mkjpryor | And keep the status as "did we transition into a ready state after the last change was made" | 09:27 |
| mkjpryor | Basically, the health state in Magnum is more like the reconciliation loop | 09:28 |
| mkjpryor | :shrugs: | 09:28 |
| dalees | yeah, i kinda agree - it may change often back and forth. Surfacing in Magnum API is the primary goal, so customers who only interact there can see what ClusterAPI might be doing or failing on. I've found most status in the Cluster object are bubbled up from others, so it may be that we only need to surface one thing. | 09:30 |
| dalees | I'll give it a bit more thought as I start using. | 09:31 |
| mkjpryor | Sounds good | 09:32 |
| jakeyip | something aside, i just got thru first of the rbac changes by ricolin. would appreciate another pair of eyes as this has big impact if we mess up. | 09:32 |
| dalees | great to see resize is included | 09:32 |
| mnasiadka | morning | 09:37 |
| dalees | jakeyip: I'll see what I can do with rbac. It's something we've customised a lot, though not much with Magnum. | 09:37 |
| mnasiadka | jakeyip: anything else needs reviewing? ;-) | 09:37 |
| travisholton | mkjpryor: sounds really promising. I'll try and spend some time testing over the next few days | 09:37 |
| jakeyip | mnasiadka, dalees: https://review.opendev.org/c/openstack/magnum/+/874945 for rbac. | 09:42 |
| mkjpryor | Who is going to Vancouver? Would be good to arrange to see people in person and put faces to names. | 09:44 |
| mnasiadka | jakeyip: commented, but we really need answers from the author ;-) | 09:46 |
| jakeyip | mnasiadka: thanks! | 09:51 |
| jakeyip | kind of hit the nail on theh head there with that comment mnasiadka :D | 09:52 |
| jakeyip | https://stackhpc.github.io/capi-helm-charts is 404 ? | 09:56 |
| supamatt | https://github.com/stackhpc/capi-helm-charts | 09:57 |
| jakeyip | oh nvm got it | 09:57 |
| mnasiadka | well, helm needs to have them published as files, it doesn't support fetching from git? | 09:59 |
| mkjpryor | https://stackhpc.github.io/capi-helm-charts/index.yaml is the Helm chart repo | 09:59 |
| mkjpryor | When you do https://stackhpc.github.io/capi-helm-charts it looks for index.html, which doesn't exist | 09:59 |
| mkjpryor | But when you tell Helm the repo, you just say https://stackhpc.github.io/capi-helm-charts because Helm knows to add the index.yaml | 10:00 |
| jakeyip | yeah the words in changelog "from this repo" means helm repo not git repo (or chart file in repo to be exact), and I just clicked expecting git repo | 10:03 |
| jakeyip | all good | 10:04 |
| mnasiadka | I guess it should be moved to opendev, but then we would need to somehow sort out publishing | 10:38 |
| opendevreview | Matt Pryor proposed openstack/magnum master: ClusterAPI: implement update_cluster_status https://review.opendev.org/c/openstack/magnum/+/851076 | 10:44 |
| opendevreview | Matt Pryor proposed openstack/magnum master: Implement cluster update for Cluster API driver https://review.opendev.org/c/openstack/magnum/+/880805 | 12:04 |
| opendevreview | Matt Pryor proposed openstack/magnum master: Cluster API: Implement upgrade to a new template https://review.opendev.org/c/openstack/magnum/+/884254 | 17:04 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!