# oc get mcp
NAME         CONFIG                                                 UPDATED   UPDATING   DEGRADED   MACHINECOUNT   READYMACHINECOUNT   UPDATEDMACHINECOUNT   DEGRADEDMACHINECOUNT   AGE
master       rendered-master-5750e45d70387c7d0df7a824eca28690       True      False      False      3              3                   3                     0                      9h
worker       rendered-worker-a8dcfdd7c435150f70dc3a557418e9f3       True      False      False      2              2                   2                     0                      9h
worker-cnf   rendered-worker-cnf-9665ba66c03c86f98fde3c54015a44cb   True      False      False      2              2                   2                     0                      4h49m

# oc get mcp worker-cnf -o yaml
apiVersion: machineconfiguration.openshift.io/v1
kind: MachineConfigPool
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"machineconfiguration.openshift.io/v1","kind":"MachineConfigPool","metadata":{"annotations":{},"labels":{"machineconfiguration.openshift.io/role":"worker-cnf"},"name":"worker-cnf"},"spec":{"machineConfigSelector":{"matchExpressions":[{"key":"machineconfiguration.openshift.io/role","operator":"In","values":["worker-cnf","worker"]}]},"maxUnavailable":null,"nodeSelector":{"matchLabels":{"node-role.kubernetes.io/worker-cnf":""}},"paused":false}}
  creationTimestamp: "2020-02-25T06:14:38Z"
  generation: 5
  labels:
    machineconfiguration.openshift.io/role: worker-cnf
  name: worker-cnf
  resourceVersion: "150323"
  selfLink: /apis/machineconfiguration.openshift.io/v1/machineconfigpools/worker-cnf
  uid: 19c3e3b3-7c4d-4785-bde8-5099ec708393
spec:
  configuration:
    name: rendered-worker-cnf-9665ba66c03c86f98fde3c54015a44cb
    source:
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 00-worker
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 00-worker-chronyd-custom
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 01-worker-container-runtime
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 01-worker-kubelet
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 98-worker-c77c686b-12fd-4e58-bbe1-aff3187bb6a2-kubelet
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 98-worker-cnf-19c3e3b3-7c4d-4785-bde8-5099ec708393-kubelet
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 99-worker-c77c686b-12fd-4e58-bbe1-aff3187bb6a2-registries
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 99-worker-cnf-19c3e3b3-7c4d-4785-bde8-5099ec708393-kubelet
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 99-worker-registries
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 99-worker-ssh
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: load-sctp-module
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: performance-performance
  machineConfigSelector:
    matchExpressions:
    - key: machineconfiguration.openshift.io/role
      operator: In
      values:
      - worker-cnf
      - worker
  nodeSelector:
    matchLabels:
      node-role.kubernetes.io/worker-cnf: ""
  paused: false
status:
  conditions:
  - lastTransitionTime: "2020-02-25T06:14:43Z"
    message: ""
    reason: ""
    status: "False"
    type: RenderDegraded
  - lastTransitionTime: "2020-02-25T06:14:48Z"
    message: ""
    reason: ""
    status: "False"
    type: NodeDegraded
  - lastTransitionTime: "2020-02-25T06:14:48Z"
    message: ""
    reason: ""
    status: "False"
    type: Degraded
  - lastTransitionTime: "2020-02-25T06:41:59Z"
    message: All nodes are updated with rendered-worker-cnf-9665ba66c03c86f98fde3c54015a44cb
    reason: ""
    status: "True"
    type: Updated
  - lastTransitionTime: "2020-02-25T06:41:59Z"
    message: ""
    reason: ""
    status: "False"
    type: Updating
  configuration:
    name: rendered-worker-cnf-9665ba66c03c86f98fde3c54015a44cb
    source:
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 00-worker
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 00-worker-chronyd-custom
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 01-worker-container-runtime
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 01-worker-kubelet
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 98-worker-c77c686b-12fd-4e58-bbe1-aff3187bb6a2-kubelet
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 98-worker-cnf-19c3e3b3-7c4d-4785-bde8-5099ec708393-kubelet
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 99-worker-c77c686b-12fd-4e58-bbe1-aff3187bb6a2-registries
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 99-worker-cnf-19c3e3b3-7c4d-4785-bde8-5099ec708393-kubelet
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 99-worker-registries
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: 99-worker-ssh
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: load-sctp-module
    - apiVersion: machineconfiguration.openshift.io/v1
      kind: MachineConfig
      name: performance-performance
  degradedMachineCount: 0
  machineCount: 2
  observedGeneration: 5
  readyMachineCount: 2
  unavailableMachineCount: 0
  updatedMachineCount: 2

I have a suspicion that it is related to race/sync. 

Since we have a system unit : pre-boot-tuning.service that runs:
rpm-ostree initramfs --enable --arg=-I --arg="/etc/systemd/system.conf /etc/systemd/system.conf.d/setAffinity.conf" 
touch /var/reboot 

And another system reboot.service unit that runs:
if [[ -f /var/reboot ]]; then 
    rm -f /var/reboot
    echo "File /var/reboot exists, initiate reboot"
    systemctl reboot
fi

We might need to remove reboot.service completely and just run under pre-boot-tuning.service script:
rpm-ostree initramfs --enable --arg=-I --arg="/etc/systemd/system.conf /etc/systemd/system.conf.d/setAffinity.conf" 
(Added a reboot -r flag)

Can you inspect the second worker node to see if the `pre-boot-tuning.service` actually ran?

Or if it is a race as suggested in comment#2, can you adjust the ordering of the systemd units so that `reboot.service` fires after `pre-boot-tuning.service`?

Though the suggestion of using `-r` as part of `rpm-ostree initramfs` would be a good path forward, too.

(In reply to Micah Abbott from comment #3)
> Can you inspect the second worker node to see if the
> `pre-boot-tuning.service` actually ran?

I have taken a look on the system where Gowrishankar Rajaiyan ran the deployment and saw that the pre-boot-tuning.service ran and the systemctl logs looks exactly the same as the one we have on a working node.
(shanks you can re-verify me on that - even by just running journalctl -u pre-boot-tuning.service on both nodes).


> Or if it is a race as suggested in comment#2, can you adjust the ordering of
> the systemd units so that `reboot.service` fires after
> `pre-boot-tuning.service`?


The reboot.service is a one shot run that should fire after pre-boot-tuning.service:

[Unit]
Description=Preboot tuning patch
Before=kubelet.service
Before=reboot.service


> Though the suggestion of using `-r` as part of `rpm-ostree initramfs` would
> be a good path forward, too.

I am working on a PR for that under the performance-addons-operator, but that is out of an assessment of what might have gone wrong.
Still need to understand if that is a sync issue or something else.

I am raising the severity as we are in trouble if the low latency deployment is not reliable. I do not see any workaround either.

Added a PR for using reboot directly in the rpm-ostree initramfs command: 
https://github.com/openshift-kni/performance-addon-operators/pull/104

I wonder if this could be related to the other initrd issue we have (file not being added): https://bugzilla.redhat.com/show_bug.cgi?id=1806588

>> Though the suggestion of using `-r` as part of `rpm-ostree initramfs` would
>> be a good path forward, too.

>I am working on a PR for that under the performance-addons-operator, but that is out of an assessment of what might have gone wrong.
>Still need to understand if that is a sync issue or something else.

Although using -r for rpm-ostree initramfs injection seem to pass CI in the suggest PR:   
https://github.com/openshift-kni/performance-addon-operators/pull/104

I still see inconsistent behaviour on my local machine using that fix.

--
I tried adding a sync+ sleep as well to see if possible race conditions can be solved using:

rpm-ostree initramfs --enable --arg=-I --arg="${SYSTEM_CONFIG_FILE} ${SYSTEM_CONFIG_CUSTOM_FILE}" 
sleep 10
sync
systemctl reboot

but still i see locally:
rpm-ostree status -v
State: idle
AutomaticUpdates: disabled
Deployments:
* pivot://quay.io/openshift-release-dev/ocp-v4.0-art-dev@sha256:43e531d989859972b6ec5ed6277ee98288b8a238a9da933d2a9bec6eb93528a6
              CustomOrigin: Managed by machine-config-operator
                   Version: 44.81.202001240931.0 (2020-01-24T09:36:54Z)
                BaseCommit: ad2b3d5d3595322bc00278bb0f09542e33f43d4c0d92c91757ea6996818912a1
                            |- art-rhcos-4.3 ((invalid timestamp))
                            |- art-rhcos-4.4 ((invalid timestamp))
                            |- rhel8-fast-datapath ((invalid timestamp))
                            |- rhel8-baseos ((invalid timestamp))
                            `- rhel8-appstream ((invalid timestamp))
                    Commit: b338e6fe85c5bb358c4111b13939d32ef198ad2601dea8b3797b80e1c643789f
                    Staged: no
                 StateRoot: rhcos
       RemovedBasePackages: kernel-core kernel-modules kernel kernel-modules-extra 4.18.0-147.3.1.el8_1
             LocalPackages: kernel-rt-modules-extra-4.18.0-147.3.1.rt24.96.el8_1.x86_64 kernel-rt-modules-4.18.0-147.3.1.rt24.96.el8_1.x86_64
                            kernel-rt-core-4.18.0-147.3.1.rt24.96.el8_1.x86_64

  pivot://quay.io/openshift-release-dev/ocp-v4.0-art-dev@sha256:43e531d989859972b6ec5ed6277ee98288b8a238a9da933d2a9bec6eb93528a6
              CustomOrigin: Managed by machine-config-operator
                   Version: 44.81.202001240931.0 (2020-01-24T09:36:54Z)
                    Commit: ad2b3d5d3595322bc00278bb0f09542e33f43d4c0d92c91757ea6996818912a1
                            |- art-rhcos-4.3 ((invalid timestamp))
                            |- art-rhcos-4.4 ((invalid timestamp))
                            |- rhel8-fast-datapath ((invalid timestamp))
                            |- rhel8-baseos ((invalid timestamp))
                            `- rhel8-appstream ((invalid timestamp))
                 StateRoot: rhcos

Service log:
 journalctl -u pre-boot-tuning.service 
-- Logs begin at Tue 2020-01-28 15:58:30 UTC, end at Thu 2020-02-27 09:13:20 UTC. --
Feb 27 07:35:29 test-1-84xwn-worker-0-kfpx4 systemd[1]: Starting Preboot tuning patch...
Feb 27 07:35:30 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: /etc/sysconfig/irqbalance
Feb 27 07:35:45 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Checking out tree ad2b3d5...done
Feb 27 07:35:45 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Enabled rpm-md repositories:
Feb 27 07:35:45 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Importing rpm-md...done
Feb 27 07:35:46 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Resolving dependencies...done
Feb 27 07:35:46 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Applying 4 overrides and 3 overlays
Feb 27 07:35:47 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Processing packages...done
Feb 27 07:35:47 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Running pre scripts...done
Feb 27 07:36:15 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Running post scripts...done
Feb 27 07:36:34 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Running posttrans scripts...done
Feb 27 07:36:36 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Writing rpmdb...done
Feb 27 07:41:16 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Generating initramfs...done
Feb 27 07:41:22 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Writing OSTree commit...done
Feb 27 07:41:28 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Staging deployment...done
Feb 27 07:41:32 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Freed: 913.4 MB (pkgcache branches: 0)
Feb 27 07:41:32 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1823]: Initramfs regeneration is now: enabled
Feb 27 07:41:43 test-1-84xwn-worker-0-kfpx4 systemd[1]: pre-boot-tuning.service: Main process exited, code=killed, status=15/TERM
Feb 27 07:41:43 test-1-84xwn-worker-0-kfpx4 systemd[1]: pre-boot-tuning.service: Failed with result 'signal'.
Feb 27 07:41:43 test-1-84xwn-worker-0-kfpx4 systemd[1]: Stopped Preboot tuning patch.
Feb 27 07:41:43 test-1-84xwn-worker-0-kfpx4 systemd[1]: pre-boot-tuning.service: Consumed 307ms CPU time
-- Reboot --
Feb 27 07:42:59 test-1-84xwn-worker-0-kfpx4 systemd[1]: Starting Preboot tuning patch...
Feb 27 07:43:00 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1812]: Pre boot tuning configuration already applied
Feb 27 07:43:00 test-1-84xwn-worker-0-kfpx4 pre-boot-tuning.sh[1812]: Setting kernel rcuo* threads to the housekeeping cpus


To rule out the possibility that realtime kernel have changed the rpm-ostree outcome- 
Realtime kernel updates :

Feb 27 00:37:51 localhost kernel: Linux version 4.18.0-147.3.1.el8_1.x86_64 (mockbuild.eng.bos.redhat.com) (gcc version 8.3.1 20190507 (Red Hat 8.3.1-4) (GCC)) #1 SMP Wed Nov 27 01:11:44 UTC 2019
Feb 27 07:34:23 localhost kernel: Linux version 4.18.0-147.3.1.rt24.96.el8_1.x86_64 (mockbuild.eng.bos.redhat.com) (gcc version 8.3.1 20190507 (Red Hat 8.3.1-4) (GCC)) #1 SMP PREEMPT RT Wed Nov 27 18:29:55 UTC 2019
Feb 27 07:42:01 localhost kernel: Linux version 4.18.0-147.3.1.rt24.96.el8_1.x86_64 (mockbuild.eng.bos.redhat.com) (gcc version 8.3.1 20190507 (Red Hat 8.3.1-4) (GCC)) #1 SMP PREEMPT RT Wed Nov 27 18:29:55 UTC 2019

Added a retry mechanism for initramfs injection under https://github.com/openshift-kni/performance-addon-operators/pull/104 
This can resolve the issue with an additional reboot

(In reply to Martin Sivák from comment #7)
> I wonder if this could be related to the other initrd issue we have (file
> not being added): https://bugzilla.redhat.com/show_bug.cgi?id=1806588

Yeah, I'm pretty sure this is a duplicate.

*** This bug has been marked as a duplicate of bug 1806588 ***

The needinfo request[s] on this closed bug have been removed as they have been unresolved for 1000 days