RHSA-2016:2778 - Security Advisory

Topic

An update for openshift-ansible and ansible is now available for OpenShift Container Platform 3.2 and 3.3.

Red Hat Product Security has rated this update as having a security impact of Moderate. A Common Vulnerability Scoring System (CVSS) base score, which gives a detailed severity rating, is available for each vulnerability from the CVE link(s) in the References section.

Description

Red Hat OpenShift Container Platform is the company's cloud computing Platform-as-a-Service (PaaS) solution designed for on-premise or private cloud deployments.

Security Fix(es):

• Ansible fails to properly sanitize fact variables sent from the Ansible controller. An attacker with the ability to create special variables on the controller could execute arbitrary commands on Ansible clients as the user Ansible runs as. (CVE-2016-8628)

This issue was discovered by Michael Scherer (Red Hat).

Bug Fix(es):

• Previous versions of the openshift-ansible code base were not compatible with the latest Ansible 2.2.0.0 release. This bug fix resolves several compatibility issues with the GA version of Ansible 2.2.0.0. (BZ#1389928) (BZ#1389275)
• The hosts.ose.example inventory file had the incorrect openshift_release version set. This bug fix updates the version to match the channel in which it is shipped. (BZ#1386333)
• The etcd certificate authority created by the installer had an expiry date one year in the future. With this bug fix, the expiry date has been updated to five years, matching the lifespan of other certificate authorities created by the installer. (BZ#1391548)
• After restarting systemd-journal, master controllers and API services stopped working. This bug fix updates the installer to set Restart=always for the master controllers and API services, and this issue no longer occurs for new installations. For existing clusters, see https://access.redhat.com/solutions/2749571. (BZ#1378929)
• When using the quick installer to install a cluster with a single master, the installer messaging suggested that an embedded etcd would be deployed. In newer versions of the quick installer, this is no longer the case, and a stand-alone etcd datastore is deployed in this scenario. This bug fix updates the quick installer messaging to match those changes. (BZ#1383961)
• Upgrades would fail if the /etc/ansible/facts.d/openshift.fact cache was missing on the system, particularly for co-located master and etcd hosts. This bug fix improves etcd fact checking during upgrades, and the issue no longer occurs. (BZ#1391608)
• Containerized upgrades from OpenShift Container Platform 3.2 to 3.3 would fail to properly create the service signing certificate due to an invalid path being used in containerized environments. This bug fix corrects that error, and containerized upgrades now create service signer certificates as a result. (BZ#1391865)
• Upgrades from OpenShift Container Platform 3.2 to 3.3 could fail with the error "AnsibleUndefinedVariable: 'dict object' has no attribute 'debug_level'". This bug fix sets missing defaults for debug_level, and as a result the upgrade error no longer occurs. (BZ#1392276)
• Previously in embedded environments, etcd 2.x was used to backup the etcd data before performing an upgrade. However, etcd 2.x has a bug that prevents backups from working properly, preventing the upgrade playbooks from running to completion. With this bug fix, etcd 3.0 is now installed for embedded etcd environments, which resolves the bug allowing upgrades to proceed normally. This bug only presents itself when using the embedded etcd service on single master environments. (BZ#1382634)
• Pacemaker clusters are no longer supported, but related code that remained could in some cases cause upgrade failures. This bug fix removes the Pacemaker restart logic from the installer to avoid these issues. (BZ#1382936)
• Previously, upgrades from OpenShift Container Platform 3.1 to 3.2 could fail due to erroneous host names being added for etcd hosts during backup. This bug fix addresses issues with conditionals and loops in templates that caused this problem, and as a result the upgrade errors no longer occur. (BZ#1392169)

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

Solution

For details on how to apply this update, which includes the changes described in this advisory, refer to:

https://access.redhat.com/articles/11258

To apply this update, run the following on all hosts where you intend to initiate Ansible-based installation or upgrade procedures:

# yum update atomic-openshift-utils

Affected Products

• Red Hat OpenShift Container Platform 3.3 x86_64

Fixes

• BZ - 1378929 - [3.3.0.32] Restarting systemd-journald causes master controllers to die
• BZ - 1382634 - Asynchronous errata upgrade to OSE 3.3.0.34 fails
• BZ - 1382936 - upgrade failed for single master
• BZ - 1383961 - it should be an embedded etcd instead of a separate etcd instance when quick install with one master
• BZ - 1386333 - Wrong default openshift_release version
• BZ - 1388113 - CVE-2016-8628 ansible: Command injection by compromised server via fact variables
• BZ - 1389275 - Installation failed when enabling OpenStack cloudprovider
• BZ - 1389928 - AWS/OpenStack cloudprovider wasn't configured in master/node config
• BZ - 1391548 - Increase default CA lifetime (advanced installation)(https://github.com/openshift/openshift-ansible/pull/2703)
• BZ - 1391608 - Upgrade Playbook from 3.3.0.35 to 3.3.1.3 failed on checking embedded etcd on multi-master/etcd environment
• BZ - 1391865 - Trouble creating service signer certificate while running upgrade in containerized environment
• BZ - 1392169 - Evaluate etcd_hosts_to_backup task causes upgrade failure from 3.1 to 3.2
• BZ - 1392276 - Upgrade fails with AnsibleUndefinedVariable

CVEs

• CVE-2016-8628

References

• https://access.redhat.com/security/updates/classification/#moderate