This repository was archived by the owner on Jan 30, 2020. It is now read-only.
This repository was archived by the owner on Jan 30, 2020. It is now read-only.
systemd unit with fleet commands fails after CoreOS update #1331
Description
Hi,
after updating CoreOS to version 774.0.0, which contains a new fleet version (0.11.4) according to the release notes, a systemd unit does not work anymore, which runs several fleet commands. Since there are no problems with CoreOS 773.1.0, I assume fleet is the problem.
systemd unit (part of cloud init):
- name: kubernetes.service
content: |
[Unit]
Description=Kubernetes Start script
Requires=etcd2.service fleet.service
After=etcd2.service fleet.service
[Service]
Type=oneshot
RemainAfterExit=yes
# wait for fleet...
ExecStart=/usr/bin/sleep 5
ExecStart=/usr/bin/fleetctl start --no-block /home/core/fleet-units/kube-apiserver.service
ExecStart=/usr/bin/fleetctl start --no-block /home/core/fleet-units/kube-proxy.service
ExecStart=/usr/bin/fleetctl start --no-block /home/core/fleet-units/kube-scheduler.service
ExecStart=/usr/bin/fleetctl start --no-block /home/core/fleet-units/kube-controller-manager.service
ExecStart=/usr/bin/fleetctl start --no-block /home/core/fleet-units/kube-kubelet.service
ExecStart=/home/core/bin/create-k8s-env.sh
kube-apiserver.service:
[Unit]
Description=Kubernetes API Server
Documentation=https://github.com/GoogleCloudPlatform/kubernetes
[Service]
ExecStart=/opt/bin/kube-apiserver \
--insecure-bind-address=0.0.0.0 \
--insecure-port=10260 \
--bind-address=0.0.0.0 \
--secure-port=10270 \
--kubelet-https=true \
--etcd-servers=http://172.17.8.20:2379 \
--service-cluster-ip-range=10.0.0.0/16 \
--runtime-config=api/v1 \
--allow-privileged=true \
--logtostderr=true \
--v=3
Restart=always
RestartSec=10
[X-Fleet]
MachineMetadata=role=etcd
kube-proxy.service:
[Unit]
Description=Kubernetes Proxy
Documentation=https://github.com/GoogleCloudPlatform/kubernetes
[Service]
ExecStart=/opt/bin/kube-proxy \
--master=http://172.17.8.20:10260 \
--logtostderr=true \
--v=3
Restart=always
RestartSec=10
[X-Fleet]
Global=true
With fleet 0.11.4 only kube-apiserver.service is started, the other ExecStart commands are not executed:
core@controller ~ $ journalctl -u kubernetes
-- Logs begin at Tue 2015-08-18 16:07:58 UTC, end at Tue 2015-08-18 16:26:05 UTC. --
Aug 18 16:09:03 controller systemd[1]: Starting Kubernetes Start script...
Aug 18 16:09:08 controller fleetctl[1372]: Unit kube-apiserver.service inactive
Aug 18 16:09:08 controller fleetctl[1372]: Triggered unit kube-apiserver.service start
core@controller ~ $ journalctl -u fleet
-- Logs begin at Tue 2015-08-18 16:07:58 UTC, end at Tue 2015-08-18 16:26:05 UTC. --
Aug 18 16:09:03 controller systemd[1]: Started fleet daemon.
Aug 18 16:09:03 controller systemd[1]: Starting fleet daemon...
Aug 18 16:09:03 controller fleetd[1273]: INFO fleetd.go:64: Starting fleetd version 0.11.4
Aug 18 16:09:03 controller fleetd[1273]: INFO fleetd.go:168: No provided or default config file found - proceeding without
Aug 18 16:09:03 controller fleetd[1273]: INFO server.go:157: Establishing etcd connectivity
Aug 18 16:09:04 controller fleetd[1273]: INFO server.go:168: Starting server components
Aug 18 16:09:04 controller fleetd[1273]: INFO engine.go:159: Updated cluster engine version from 0 to 1
Aug 18 16:09:04 controller fleetd[1273]: INFO engine.go:185: Engine leadership acquired
Aug 18 16:09:08 controller fleetd[1273]: INFO engine.go:271: Scheduled Unit(kube-apiserver.service) to Machine(f770c5bb3a5a44788e811cf439ee9b81)
Aug 18 16:09:08 controller fleetd[1273]: INFO reconciler.go:161: EngineReconciler completed task: {Type: AttemptScheduleUnit, JobName: kube-apiserver.service, MachineID: f770c5bb3a5a44788e811cf439ee9b81, Re
Aug 18 16:09:09 controller fleetd[1273]: INFO manager.go:246: Writing systemd unit kube-apiserver.service (540b)
Aug 18 16:09:09 controller fleetd[1273]: INFO manager.go:182: Instructing systemd to reload units
Aug 18 16:09:09 controller fleetd[1273]: INFO manager.go:127: Triggered systemd unit kube-apiserver.service start: job=3002
Aug 18 16:09:09 controller fleetd[1273]: INFO reconcile.go:330: AgentReconciler completed task: type=LoadUnit job=kube-apiserver.service reason="unit scheduled here but not loaded"
Aug 18 16:09:09 controller fleetd[1273]: INFO reconcile.go:330: AgentReconciler completed task: type=ReloadUnitFiles job=N/A reason="always reload unit files"
Aug 18 16:09:09 controller fleetd[1273]: INFO reconcile.go:330: AgentReconciler completed task: type=StartUnit job=kube-apiserver.service reason="unit currently loaded but desired state is launched"
core@controller ~ $ fleetctl list-units
UNIT MACHINE ACTIVE SUB
kube-apiserver.service f770c5bb.../172.17.8.20 active running
Any ideas what is going wrong?
Thanks and greetings,
Marc