Addressed issues¶

The MCP 2019.2.18 update contains fixes for the following MCP components:

• DriveTrain:
  • [36767] Fixed a number of security vulnerabilities in Docker containers.
  • [36761] Fixed the issue that caused the inoperability of periodical backups for Salt Master nodes.
  • [36754] Set the default Neutron dhcp_lease_duration parameter to 86400 to avoid the overloading of DHCP agents.
  • [36736] Removed a deprecated HAProxy user from MySQL.
  • [36734] Fixed the issue that caused the inability to prepare the Jenkins worker node for executing any Salt-based Jenkins pipeline job after upgrading DriveTrain in an offline MCP deployment.
  • [36710] Fixed the issues wherein Fluentd failed to keep up with the volume of logs it was expected to process. Enhanced the Fluentd Salt formula to support the multi-worker configuration.
• OpenStack:
  • [36763][Queens] Fixed the issue wherein Cinder ignored the availability zone when retyping a volume in a multi-availability zone environment. Added the scheduler_retype_use_az availability zone filter that you can use for retyping.
  • [36716][Queens] Fixed the issue wherein the heat_cloudwatch_api backends were down after upgrading OpenStack Pike to Queens.
  • [36719][Pike, Queens] Fixed the wrong time stamp format in the td-agent configuration, which caused failure to ship logs from Panko, Aodh, and Barbican to Elasticsearch.
• Ceph:
  • [36764] Fixed the issue with upmap in Jenkins pipeline jobs that caused an error and proceeded to a standard rebalancing when adding Ceph OSDs into undersized placement groups.