StarlingX

  1. Bug #1914291
  2. Activity log

Activity log for bug #1914291

Date Who What changed Old value New value Message
2021-02-02 20:43:00 Jerry Sun bug added bug
2021-02-02 20:43:21 Jerry Sun starlingx: assignee Jerry Sun (jerry-sun-u)
2021-02-03 03:38:46 Ghada Khalil starlingx: importance Undecided High
2021-02-03 03:38:48 Ghada Khalil starlingx: status New Triaged
2021-02-03 03:39:20 Ghada Khalil description Brief Description ----------------- Changing kube-apiserver parameters causes the kube-apiserver to fail to start up again due to missing etcd parameters. The root cause seems to be introduced by the fix to this bug: https://bugs.launchpad.net/starlingx/+bug/1894870 the etcd parameters introduced seems to be mandatory, but gets lost when changing kube-apiserver parameters. This is due to the kubernetes puppet class having no default for the new etcd parameters, and thus the "undef" defaults gets passed down to the script to change the kube-apiserver parameters. Severity -------- Major Steps to Reproduce ------------------ system service-parameter-add kubernetes kube_apiserver admission_plugins=PodSecurityPolicy system service-parameter-apply kubernetes Expected Behavior ------------------ kube apiserver restarts, with the PodSecurityPolicy plugin enabled Actual Behavior ---------------- kube apiserver fails to restart. Manually starting kube-apiserver on the command line using the same parameters as before changing the service parameters shows the kube-apiserver pods crashing. Reproducibility --------------- 100% reproducible System Configuration -------------------- Multi node system Branch/Pull Time/Commit ----------------------- pull on 2021-02-02 Last Pass --------- never Brief Description ----------------- Changing kube-apiserver parameters causes the kube-apiserver to fail to start up again due to missing etcd parameters. The root cause seems to be introduced by the fix to this bug: https://bugs.launchpad.net/starlingx/+bug/1894870 the etcd parameters introduced seems to be mandatory, but gets lost when changing kube-apiserver parameters. This is due to the kubernetes puppet class having no default for the new etcd parameters, and thus the "undef" defaults gets passed down to the script to change the kube-apiserver parameters. Severity -------- Major Steps to Reproduce ------------------ system service-parameter-add kubernetes kube_apiserver admission_plugins=PodSecurityPolicy system service-parameter-apply kubernetes Expected Behavior ------------------ kube apiserver restarts, with the PodSecurityPolicy plugin enabled Actual Behavior ---------------- kube apiserver fails to restart. Manually starting kube-apiserver on the command line using the same parameters as before changing the service parameters shows the kube-apiserver pods crashing. Reproducibility --------------- 100% reproducible System Configuration -------------------- Multi node system Branch/Pull Time/Commit ----------------------- pull on 2021-02-02 Last Pass --------- Before the etcd changes were merged
2021-02-03 03:39:38 Ghada Khalil tags stx.5.0 stx.config stx.containers
2021-02-10 13:20:38 Jerry Sun starlingx: status Triaged Fix Released
2021-06-03 18:37:05 OpenStack Infra tags stx.5.0 stx.config stx.containers in-f-centos8 stx.5.0 stx.config stx.containers
2021-06-03 18:37:05 OpenStack Infra cve linked 2020-15705