Friday, 2018-02-09

kata-dev-irc-bot<eric.ernst> no more K8S Helm talk?00:09
kata-dev-irc-bot<eric.ernst> Or just needing one in addition?00:10
kata-dev-irc-bot<eric.ernst> @tpepper00:10
kata-dev-irc-bot<eric.ernst> That's not too soon.00:10
kata-dev-irc-bot<tpepper> I’m not sure if it was a …Helmer? or Helmet? … who dropped or if you’d be talk 2 of 2?00:11
*** mylinux has quit IRC00:22
kata-dev-irc-bot<mayank.kumar> @eric.ernst i do see the containerId from the pod yaml in the cc-runtime list . you were referring to some annotations which decide in the workload, can you tell me which annotation are you referring to00:43
*** mylinux_ has joined #kata-general00:47
kata-dev-irc-bot<eric.ernst> In CRIO it is io.kubernetes.cri-o.TrustedSandbox: "false"01:06
kata-dev-irc-bot<eric.ernst> example pod yaml: https://github.com/egernst/k8s-testing-scripts/blob/master/nginx-untrusted.yaml01:06
kata-dev-irc-bot<eric.ernst> there's a -trusted one as well01:06
kata-dev-irc-bot<eric.ernst> but how that annotation is interpreted then also depends on how crio is configured, specifically default trust level, etc.  This is all described on that medium post, IIRC01:07
*** mylinux_ has quit IRC01:30
*** sjas_ has joined #kata-general01:35
*** mylinux_ has joined #kata-general01:37
*** sjas has quit IRC01:38
kata-dev-irc-bot<mayank.kumar> thanks @eric.ernst01:46
kata-dev-irc-bot<mayank.kumar> i have the following setting  ``` cat /etc/crio/crio.conf | grep ^default_work   default_workload_trust = "untrusted" root@worker-3:~/go/src/github.com/clearcontainers/tests/integration/kubernetes# cat /etc/crio/crio.conf | grep ^run   runroot = "/var/run/containers/storage" runtime = "/usr/local/bin/crio-runc" runtime_untrusted_workload = "/usr/local/bin/cc-runtime" ```01:46
kata-dev-irc-bot<mayank.kumar> but when i launch the pod with annotation trusted it still launches with the cc-runtime01:46
kata-dev-irc-bot<mayank.kumar> how can i fix it ?01:46
kata-dev-irc-bot<mayank.kumar> i am using the following yaml  ``` apiVersion: v1 kind: Pod metadata:   annotations:     io.kubernetes.cri-o.TrustedSandbox: "true"   name: ubuntu-pod2 spec:   containers:   - name: ubuntu     image: "ubuntu:14.04"     command: ["top"]     stdin: true     tty: true ```01:47
kata-dev-irc-bot<eric.ernst> @mayank.kumar pasted from that blog write-up: When the default workload type is set to untrusted, the provided untrusted runtime in the CRI-O configuration will be used for all non-privileged containers regardless of the value of io.kubernetes.cri-o.TrustedSandbox. This rule ensures all workloads can be run using Clear Containers without any changes to default payload definitions. The result could be running all02:09
kata-dev-irc-botnon-infrastructure pods in Clear Containers with relative ease. In the event that an untrusted runtime is not defined when configuring CRI-O, all containers will fall back to the trusted runtime, which is configured by default as runc.02:09
kata-dev-irc-bot<eric.ernst> Hope that clarifies!02:09
kata-dev-irc-bot<mayank.kumar> if i change that to trusted(in crio.conf) do i need to restart something  , i want to show a demo where i show one running through runc and other running through cc-runtime02:10
kata-dev-irc-bot<mayank.kumar> trying `systemctl restart crio` hopefully that fixes it02:19
*** liujiong has joined #kata-general02:30
*** mylinux_ has quit IRC02:30
kata-dev-irc-bot<mayank.kumar> hmm i think that fixxed it but `cc-runtime list` stopped working it seems02:37
kata-dev-irc-bot<mayank.kumar> i started the untrsuted pod and immediately saw a new qemu process but nithing in the cc-runtime list02:37
kata-dev-irc-bot<mayank.kumar> it seems there is some cleanup needed  ``` cc-runtime list stat /var/lib/containers/storage/overlay/52e633703a812c8be925096afaba4d5346b92835e5b0a91e54c89dbb7ad311d6/merged: no such file or directory```02:38
*** mylinux has joined #kata-general03:19
*** mylinux has quit IRC03:25
*** liujiong has quit IRC03:46
*** mylinux has joined #kata-general04:21
*** mylinux has quit IRC04:26
*** jodh has joined #kata-general07:38
*** jodh has joined #kata-general07:38
*** gwhaley has joined #kata-general09:06
kata-dev-irc-bot<samuel.ortiz> @mayank.kumar It makes more sense to set the default to trusted and annotate your pods with "untrusted" to show the mixed runtime use case.09:43
*** mylinux has joined #kata-general10:20
*** mylinux has quit IRC10:25
*** gwhaley has quit IRC11:58
*** gwhaley has joined #kata-general13:20
*** mylinux_ has joined #kata-general13:39
*** jodh has quit IRC15:00
*** jodh has joined #kata-general15:00
*** jodh has joined #kata-general15:00
kata-dev-irc-bot<mayank.kumar> thanks @samuel.ortiz16:28
kata-dev-irc-bot<mayank.kumar> for some reason cc-runtime has stopped working and it just shows the above message16:28
kata-dev-irc-bot<mayank.kumar> do you know a way to clean that up16:29
kata-dev-irc-bot<mayank.kumar> i see the new qemu processes but nothing in the cc-runtime list16:29
kata-dev-irc-bot<eric.ernst> @mayank.kumar I think this has to do with how kubeadm reset is being handled.17:12
kata-dev-irc-bot<eric.ernst> AFAIU, in that older version of CRIO/K8S, the reset is attempting to happen with help from docker, rather than using the crio socket.17:12
kata-dev-irc-bot<eric.ernst> So appropriate kill isn't ever coming down to the cc-runtime, resulting in some artifacts being left about.17:13
kata-dev-irc-bot<eric.ernst> I *think* that may be what you're running into.17:13
kata-dev-irc-bot<eric.ernst> Thankfully our testing just (this morning) moved to using latest (ish) K8S/CRIO now17:14
*** leadfoot has joined #kata-general17:18
*** jodh has quit IRC18:02
*** gwhaley has quit IRC18:33
*** justJanne has quit IRC20:04
*** justJanne has joined #kata-general20:06
*** mylinux_ has quit IRC20:30
*** mylinux has joined #kata-general20:31
*** mylinux has quit IRC20:50
*** mylinux_ has joined #kata-general20:59
*** mylinux_ has quit IRC21:12
*** mylinux has joined #kata-general21:21
*** mylinux has quit IRC23:11

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!