http://bugzilla.opensuse.org/show_bug.cgi?id=1186125 http://bugzilla.opensuse.org/show_bug.cgi?id=1186125#c3 --- Comment #3 from OBSbugzilla Bot --- This is an autogenerated message for OBS integration: This bug (1186125) was mentioned in https://build.opensuse.org/request/show/893739 Factory / kubernetes1.21 -- You are receiving this mail because: You are the assignee for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1186125 http://bugzilla.opensuse.org/show_bug.cgi?id=1186125#c5 Anthony Rabbito changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |hello@anthonyrabbito.com --- Comment #5 from Anthony Rabbito --- Would be interested in knowing the root cause of this as this hack breaks Cilium pod networking if kubeadm service is restarted.. The sysctls the cilium pod sets are reverted, thus breaking connectivity. -- You are receiving this mail because: You are the assignee for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1186125 http://bugzilla.opensuse.org/show_bug.cgi?id=1186125#c13 --- Comment #13 from OBSbugzilla Bot --- This is an autogenerated message for OBS integration: This bug (1186125) was mentioned in https://build.opensuse.org/request/show/894324 Factory / kubernetes1.21 -- You are receiving this mail because: You are the assignee for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1186125 Richard Brown changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |IN_PROGRESS -- You are receiving this mail because: You are the assignee for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1186125 http://bugzilla.opensuse.org/show_bug.cgi?id=1186125#c19 --- Comment #19 from Marc Balmer --- The latest snapshot from today again broke the cluster. By running "sysctl -a --system" on each node, I can get it somewhat running, but Rancher still not access the cluster. DNS resolution works, but the cluster manager can not access the ranger system, maybe a firewall issue? kubic-admin:~ # kubectl get pod -n cattle-system -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES cattle-cluster-agent-54879c4bb9-hdmn4 0/1 CrashLoopBackOff 4 5m23s 10.42.0.34 kubic-3 <none> <none> cattle-cluster-agent-69d5bb5845-w7wtx 0/1 CrashLoopBackOff 7 11d 10.32.0.61 kubic-admin <none> <none> -- You are receiving this mail because: You are the assignee for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1186125 http://bugzilla.opensuse.org/show_bug.cgi?id=1186125#c21 --- Comment #21 from Marc Balmer --- Kubic snapshot 20210521 is still affected by this issue. -- You are receiving this mail because: You are the assignee for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1186125 http://bugzilla.opensuse.org/show_bug.cgi?id=1186125#c23 --- Comment #23 from Richard Brown --- (In reply to Marc Balmer from comment #21)

Kubic snapshot 20210521 is still affected by this issue.

Somewhat - the fixes in 0521 resolved the situation in some cases, but not in yours. Thanks to your report (and subsequent discussions on the list) we now have a more robust fix otw, which should toggle the forwarding sysctl on kubelet start, ensuring that all ethernet devices get it properly enabled. Doing my best to speed it through stagings but as this week has some major Tumbleweed work going on a workaround meanwhile would be to copy /usr/lib/systemd/system/kubelet.service.d/10-kubeadm.conf to /etc/systemd/system/kubelet.service.d/10-kubeadm.conf and make it's contents match the linked file https://build.opensuse.org/package/view_file/devel:kubic/kubernetes1.21/10-k... -- You are receiving this mail because: You are the assignee for the bug.