rick_h | skay: "values in the services file"? | 05:58 |
---|---|---|
nvalerkos | hi | 09:28 |
stickupkid | nvalerkos, hi | 09:28 |
nvalerkos | I've deployed a controller for kubernetes charm | 09:28 |
nvalerkos | i got a problem which seem a bit odd | 09:28 |
nvalerkos | Stopped services: kube-proxy | 09:30 |
nvalerkos | status of the kubernetes master is blocked | 09:30 |
nvalerkos | I am using LXC on a single machine, localhosdt | 09:30 |
nvalerkos | I am not sure why this is happening because I just deployed a charm which should work without any problems | 09:31 |
stickupkid | nvalerkos, can you pastebin a juju status here | 09:32 |
nvalerkos | App Version Status Scale Charm Store Rev OS Notescontainerd active 5 containerd jujucharms 54 ubuntu easyrsa 3.0.1 active 1 easyrsa jujucharms 296 ubuntu etcd 3.3.15 active 3 etcd | 09:33 |
nvalerkos | jujucharms 488 ubuntu flannel 0.11.0 active 5 flannel jujucharms 468 ubuntu kubeapi-load-balancer 1.14.0 active 1 kubeapi-load-balancer jujucharms 704 ubuntu exposedkubernetes-master 1.17.2 blocked 2 kubernetes-master jujucharms 792 ubuntu kubernetes-worker 1.17.2 | 09:33 |
nvalerkos | active 3 kubernetes-worker jujucharms 634 ubuntu exposedUnit Workload Agent Machine Public address Ports Messageeasyrsa/0* active idle 1 10.38.12.8 Certificate Authority connected.etcd/0 active idle 3 10.38.12.103 | 09:33 |
nvalerkos | 2379/tcp Healthy with 3 known peersetcd/1* active idle 0 10.38.12.236 2379/tcp Healthy with 3 known peersetcd/2 active idle 5 10.38.12.60 2379/tcp Healthy with 3 known peerskubeapi-load-balancer/0* active idle 7 10.38.12.39 443/tcp | 09:33 |
nvalerkos | Loadbalancer ready.kubernetes-master/0 waiting idle 9 10.38.12.10 6443/tcp Waiting for 6 kube-system pods to start containerd/4 active idle 10.38.12.10 Container runtime available flannel/4 active idle 10.38.12.10 Flannel | 09:33 |
nvalerkos | subnet 10.1.4.1/24kubernetes-master/1* blocked idle 2 10.38.12.113 6443/tcp Stopped services: kube-proxy containerd/3 active idle 10.38.12.113 Container runtime available flannel/3 active idle 10.38.12.113 Flannel subnet | 09:33 |
nvalerkos | 10.1.93.1/24kubernetes-worker/0 active idle 8 10.38.12.54 80/tcp,443/tcp Kubernetes worker running. containerd/1 active idle 10.38.12.54 Container runtime available flannel/1 active idle 10.38.12.54 Flannel subnet | 09:33 |
nvalerkos | 10.1.11.1/24kubernetes-worker/1 active idle 6 10.38.12.237 80/tcp,443/tcp Kubernetes worker running. containerd/2 active idle 10.38.12.237 Container runtime available flannel/2 active idle 10.38.12.237 Flannel subnet | 09:33 |
nvalerkos | 10.1.45.1/24kubernetes-worker/2* active idle 4 10.38.12.89 80/tcp,443/tcp Kubernetes worker running. containerd/0* active idle 10.38.12.89 Container runtime available flannel/0* active idle 10.38.12.89 Flannel subnet | 09:33 |
nvalerkos | 10.1.96.1/24Machine State DNS Inst id Series AZ Message0 started 10.38.12.236 juju-0a24b5-0 bionic Running1 started 10.38.12.8 juju-0a24b5-1 bionic Running2 started 10.38.12.113 juju-0a24b5-2 bionic Running3 started 10.38.12.103 juju-0a24b5-3 bionic Running4 | 09:33 |
nvalerkos | started 10.38.12.89 juju-0a24b5-4 bionic Running5 started 10.38.12.60 juju-0a24b5-5 bionic Running6 started 10.38.12.237 juju-0a24b5-6 bionic Running7 started 10.38.12.39 juju-0a24b5-7 bionic Running8 started 10.38.12.54 juju-0a24b5-8 bionic Running9 started 10.38.12.10 | 09:33 |
nvalerkos | juju-0a24b5-9 bionic Running | 09:33 |
stickupkid | nvalerkos, to here https://pastebin.canonical.com/ | 09:33 |
stickupkid | :) | 09:33 |
nvalerkos | https://pasteboard.co/IRhbIoA.png | 09:33 |
nvalerkos | I dont have access to that one | 09:34 |
stickupkid | nvalerkos, it seems like there is an issue with kube-proxy which is blocking the kube-master | 09:34 |
stickupkid | nvalerkos, I'm not sure why that would be | 09:35 |
stickupkid | nvalerkos, it might be worth asking here https://discourse.jujucharms.com/ | 09:35 |
manadart | stickupkid nvalerkos: pastebin.canonical.com is internal pastebin.ubuntu.com is public. | 09:42 |
stickupkid | d'ho | 09:42 |
nammn_de | rick_h: morning, with provider are the clouds meant, right? (Maas, aws ..) | 09:46 |
nvalerkos | localhost | 09:46 |
nvalerkos | Warning FailedCreatePodSandBox 3m48s (x125 over 31m) kubelet, juju-0a24b5-6 (combined from similar events): Failed to create pod sandbox: rpc error: code = Unknown desc = failed to start sandbox container: failed to create containerd task: failed to mount rootfs component &{overlay overlay | 09:46 |
nvalerkos | [workdir=/var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/1745/work upperdir=/var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/1745/fs lowerdir=/var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/1/fs]}: invalid argument: unknown | 09:46 |
nvalerkos | All pods are stuck at containercreating | 09:47 |
nvalerkos | something to do with the storage volume | 09:47 |
nvalerkos | I am using lxd with zfs | 09:47 |
stickupkid | manadart, achilleasa CR please https://github.com/juju/juju/pull/11140 | 10:45 |
manadart | stickupkid: Yep. | 11:25 |
rick_h | nammn_de: right, I'm hinting there's a cloud-based specific trick to that PR | 13:12 |
nammn_de | rick_h: ahh, yes I was planning to write code where specific cloud providers support renaming (aws does, maas does not) | 13:12 |
nammn_de | something along the line | 13:12 |
rick_h | nammn_de: ok cool just tossing a hint your way since I didn't see anything in the PR around that | 13:13 |
nammn_de | rick_h: yeah thanks! | 13:13 |
nammn_de | better safe than sorry =D | 13:13 |
nammn_de | my pr is still in the early phase, just out for people to give fundamental hints. Like you did. Still constantly updated | 13:14 |
rick_h | nammn_de: understand | 13:14 |
nammn_de | rick_h: i answered your concerns in my old pr (show-space). As well here in chat, late though. Does that answer your concers? | 13:14 |
nammn_de | concerns | 13:14 |
rick_h | nammn_de: I think so, I'm just trying to find time to play with it myself then | 13:15 |
skay | rick_h: services file == yaml file that mojo uses. anyway, I figured out the problem. The yaml was not correct. all good. The magic thing happened where I found the error right after asking for help | 13:42 |
rick_h | skay: ah ok awesome | 14:03 |
manadart | Patch for systemd file relocation: https://github.com/juju/juju/pull/11147 | 14:51 |
stickupkid | manadart, I'll give it look now | 14:51 |
manadart | stickupkid: Thanks. | 14:54 |
stickupkid | manadart, i'm assuming that this needs regression testing | 14:54 |
manadart | stickupkid: I tested on Focal and Bionic. I assume that covers back to Xenial as well. | 14:56 |
stickupkid | manadart, yeah, what's supported under ESM | 14:56 |
stickupkid | manadart, answering my own questions - 14.04 is still in esm, but that's not systemd right, that's upstart? | 14:57 |
stickupkid | manadart, so we need to test xenial onwards? | 14:57 |
manadart | manadart: Series upgrades definitely now need rework. I will follow with a patch for that. At the same time I will test/fix the old upgrade steps. | 14:57 |
manadart | stickupkid: Trusty is Upstart IIRC. | 14:58 |
stickupkid | yeah, thought as much | 14:58 |
=== narindergupta is now known as narinderguptamac | ||
stickupkid | manadart, so upgrading series will now need to move files right? | 14:59 |
manadart | stickupkid: Actually thinking about it, upgrading from Trusty should work out of the box (I will test it). Aside from that, we handle in a Juju upgrade step. | 15:04 |
manadart | I think. Testing to follow. | 15:04 |
manadart | Series upgrade leaves the files alone if the upgrade is systemd -> systemd. | 15:05 |
stickupkid | but that's not what we want if a restart happens though right? | 15:06 |
manadart | It should be OK. If you do a series upgrade with Juju < 2.7.2, you will not be able to go up to Focal. | 15:09 |
stickupkid | interesting | 15:09 |
manadart | So you upgrade Juju, then do it. | 15:09 |
manadart | If you are > Trusty and you upgrade Juju, it just needs to relocate the files. | 15:09 |
stickupkid | > sudo: setrlimit(RLIMIT_CORE): Operation not permitted | 15:10 |
manadart | stickupkid: That is a symptom of sudo on Focal (at least on LXD). it doesn't come from Juju. | 15:11 |
stickupkid | yeah, really didn't like that | 15:11 |
manadart | stickupkid: You can test it on the container directly - just sudo something. | 15:11 |
stickupkid | anyway, it finished :D | 15:12 |
stickupkid | manadart, done | 15:22 |
nammn_de | achilleasa manadart: is there an easy way to bootstrap an controller on aws with --config juju-ha-space=<any space> ? I thought alpha should work. But it fails for me | 15:23 |
manadart | nammn_de: I think there might be chicken/egg thing there with AWS, because you are trying to land a controller on a space that doesn't come from the provider. Juju hasn't loaded the subnets and created the space yet, so it can't satisfy. | 15:29 |
nammn_de | manadart: hmmm makes sense. Just wanted to set the config setting so that i can test the rename-code | 15:29 |
manadart | nammn_de: Try the other config value (juju-mgmt-space). | 15:29 |
nammn_de | I could, maybe, just add a key-value to the collection :D? | 15:29 |
nammn_de | manadart: same error | 15:30 |
manadart | nammn_de: OK, both should work if you set it post bootstrap with `juju controller-config juju-ha-space=alpha` | 15:30 |
nammn_de | manadart: oh yes, hmm tried before. now works after trying again.. weird. Thanks! | 15:31 |
nammn_de | manadart: regarding your comment | 16:32 |
nammn_de | didnt we discuss last time that we cannot remove them and need to add them as empty https://github.com/juju/juju/pull/11088#discussion_r370208805 | 16:32 |
nammn_de | because we need to fulfill the interface? | 16:32 |
manadart | Wrap network backing stub in a new type for the old facade tests and put them on that. | 16:44 |
manadart | The old facade tests will be strangled out in favour of the new mock suite and they will ultimately disappeared. Those methods have nothing to do with the common backing. | 16:45 |
achilleasa | hml: can you take a look at https://github.com/juju/description/pull/72 ? | 16:50 |
hml | achilleasa: looking | 16:50 |
nammn_de | manadart: not sure if I can follow 100% do you have an example for me? | 16:52 |
nammn_de | because afaict the spaces.NewAPIWithBacking(...) takes a backing interface. That's why it initially used those implements. But Actually there is no networkbacking usage in that test anymore | 17:00 |
hml | achilleasa: approved | 17:10 |
achilleasa | hml: tyvm | 17:10 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!