-
Bug
-
Resolution: Done
-
Major
-
None
-
4.13.0
-
None
-
Moderate
-
No
-
SDN Sprint 234
-
1
-
Rejected
-
False
-
This is a clone of issue OCPBUGS-10889. The following is the description of the original issue:
—
When there is an error initializing the healthz server, ovnkube-node panics and the process hangs without closing up.
This is due to an invalid interface reference to the network controller which is initialized to a nil implementation instead of a nil interface. The panic happens when it is attempted to be used on Stop(). On that panic, defer for that go-routine are resolved, which includes a wait on the metric server wait group which holds forever as the main context has not been canceled .
This is a stack trace showcasing the problem:
runtime.gopark(proc.go:364) runtime.goparkunlock(proc.go:369) runtime.semacquire1(sema.go:150) sync.runtime_Semacquire(sema.go:62) sync.(*WaitGroup).Wait(waitgroup.go:139) main.startOvnKube.func5(ovnkube.go:293) runtime.gopanic(panic.go:890) runtime.panicmem(panic.go:260) runtime.sigpanic(signal_unix.go:835) github.com/ovn-org/ovn-kubernetes/go-controller/pkg/node.(*DefaultNodeNetworkController).Stop(default_node_network_controller.go:785) github.com/ovn-org/ovn-kubernetes/go-controller/pkg/network-controller-manager.(*nodeNetworkControllerManager).Stop(node_network_controller_manager.go:163) github.com/ovn-org/ovn-kubernetes/go-controller/pkg/network-controller-manager.(*nodeNetworkControllerManager).Start.func1(node_network_controller_manager.go:129) github.com/ovn-org/ovn-kubernetes/go-controller/pkg/network-controller-manager.(*nodeNetworkControllerManager).Start(node_network_controller_manager.go:142) main.runOvnKube(ovnkube.go:504) main.startOvnKube(ovnkube.go:297) main.main.func1(ovnkube.go:112) github.com/urfave/cli/v2.(*App).RunContext(app.go:315) main.main(ovnkube.go:136) runtime.main(proc.go:250) runtime.goexit(asm_amd64.s:1594) runtime.newproc(<autogenerated>:1)
- clones
-
-
- Closed
-
- is blocked by
-
-
- Closed
-
- links to
