RHBA-2019:0794 - Bug Fix Advisory

Topic

Red Hat OpenShift Container Platform release 3.11.104 is now available with
updates to packages and images that fix several bugs.

Description

Red Hat OpenShift Container Platform is Red Hat's cloud computing
Kubernetes application platform solution designed for on-premise or private
cloud deployments.

This advisory contains the RPM packages for Red Hat OpenShift Container
Platform 3.11.104. See the following advisory for the container images for
this release:

https://access.redhat.com/errata/RHBA-2019:0795

This update fixes the following bugs:

• Director-deployed pods would stop in the `CrashLoopBackOff` state after a

rolling reboot of a node. This was because the `READY` sequence would
display a node before it had started. Now, the `READY` indicator allows
components to come online before displaying as a ready state. (BZ#1654044)

• Ansible playbook `health.yml` assumed `curator` was a `deploymentconfig` instead of a `cronjob`. Now, the `health.yml` playbook checks for `curator` as a `cronjob`. This change properly evaluates the `curator` status. (BZ#1676720)
• The `NetworkPolicy` plugin did not clean up rules from deleted namespaces. Now, all `OpenVSwitch` flows associated with a namespace are deleted properly when a namespace is deleted. (BZ#1686025)
• During `Satellite` registry-based installations of OpenShift Container Platform 3.11, the example template URLs were installed at the wrong file path. This would cause example resources to be configured incorrectly. Now, a condition has been added to replace the image URL with the `Satellite` location. As a result, the example resources are configured with valid image URLs. (BZ#1689848)
• Previously, cluster logging did not store secret names in service accounts. When secrets were required to be whitelisted, the logging service accounts were unable to access their required secrets. Now, secret names are added appropriately to their service accounts. (BZ#1690605)
• `oc cp` commands were not checking links from tar files used to copy

files between pods and user's workstations. The `oc cp` command could cause
a directory traversal and replace or delete files on a user's workstation.
Now, escaping links are not permitted. As a result, the `oc cp` command
verifies files copied between pods and workstations without allowing escape
from directories. (BZ#1693315)

• During previous upgrades, the `tuned` package and profiles could have

been removed. The `tuned` role was not being applied during an upgrade, but
only during a fresh install. Now, the `tuned` role is applied during
upgrades to ensure `tuned` profiles are applied appropriately. (BZ#1694131)

• `NetworkPolicy` rules were not updated reliably after service restarts. A bug in the re-initialization process of the plugin state after a restart of the SDN service would ignore changes in a namespace. Now, updates to `NetworkPolicies` are now correctly tracked at all times. (BZ#1694704)
• Monitoring certificates were not updated after certificate redeployment. As a result, `prometheus`, `grafana`, and `alertmanager` user interfaces were inoperable. Now the TLS secrets and pods are removed during certificate redeployment and the user interfaces work correctly after certificate redeployment. (BZ#1696198)

All OpenShift Container Platform 3.11 users are advised to upgrade to these
updated packages and images.

Solution

Before applying this update, ensure all previously released errata relevant to your system are applied.

See the following documentation, which will be updated shortly for release 3.11.104, for important instructions on how to upgrade your cluster and fully apply this asynchronous errata update:

https://docs.openshift.com/container-platform/3.11/release_notes/ocp_3_11_release_notes.html

This update is available via the Red Hat Network. Details on how to use the
Red Hat Network to apply this update are available at
https://access.redhat.com/articles/11258.

Affected Products

• Red Hat OpenShift Container Platform 3.11 x86_64
• Red Hat OpenShift Container Platform for Power 3.11 ppc64le

Fixes

• BZ - 1427274 - Kibana header "container-brand" image should be properly branded for the deployment
• BZ - 1633892 - OCP 3.11: CRI-O cluster install with openshift-ansible fails on large AWS instances m5.12xlarge and m4.16xlarge due to Master node openshift-sdn pod in CrashLoopBackOff state
• BZ - 1634151 - [3.11.z] CRI-O cluster install with openshift-ansible fails on large AWS instances m5.12xlarge and m4.16xlarge due to Master node openshift-sdn pod in CrashLoopBackOff state
• BZ - 1651393 - rotate-server-certificates gets set to false when set separately
• BZ - 1654044 - OCP 3.11: pods end up in CrashLoopBackOff state after a rolling reboot of the node
• BZ - 1670418 - inventory variable which points out to IP address of NFS server is not being expanded, so that the installer creates the PVC with wrongs data.
• BZ - 1680063 - python2-certifi points to the wrong location
• BZ - 1686025 - [3.11] Network Policy Plugin does not clean up flows from deleted namespaces
• BZ - 1689000 - Custom managed nic causing install to fail
• BZ - 1689848 - example template image url does not work on Satellite
• BZ - 1690605 - Aggregated Logging installation does not add secret to serviceaccount [3.11.z]
• BZ - 1690900 - etcd rpm is installed when etcd is co-located on a master
• BZ - 1690951 - [3.11.z] KubeletTooManyPods statically compares against 100 instead of --max-pods (-10)
• BZ - 1691893 - KubeCPUOvercommit factors in Completed and Failed pods
• BZ - 1693035 - Fluentd doesn't output it's logs to STDOUT when LOGGING_FILE_PATH=console
• BZ - 1694106 - [crio-tool] Installing or upgrading OCP 3.11 when using crio, keeps changing the crictl.yaml file with the incorrect pathvim
• BZ - 1694131 - Tuned profiles are not applied to OpenShift clusters that get upgraded from 3.6
• BZ - 1694704 - [3.11] NetworkPolicy rules don't update reliably after a service restart
• BZ - 1694899 - The kibana was authorized as CN=system.logging.kibana by mistake
• BZ - 1695271 - [3.11] Wildcard routes get 503 intermittently
• BZ - 1695856 - Task failure restart docker while running redeploy-certificates
• BZ - 1696198 - after running redeploy-certificates.yml playbook grafana, prometheus and alertmanager routes are not accessible anymore
• BZ - 1697169 - Addition of custom ca bundle in Grafana throws error
• BZ - 1697295 - Prometheus shows different monitoring history with Grafana dashboard refresh

CVEs

(none)

References

(none)