1.8.1

Release Notes

1.8.1

Bug Fixes

• Add boolean type for deploy_osp in custom OSD fact gathering module.

1.8.0

Release Notes

1.8.0

New Features

• This PR allows AIO creation for RPCR osp-13. The condition is ${RE_JOB_ACTION} == osp_13_deploy
• Ceph commands for cech checks needs to run on the host not, not inside ceph docker container on osp-13. This PR fix that and enable ceph checks for osp-13 gates.
• Now that embedded ansible is required to deploy rpc-maas playbooks against RPC-O and/or RPC-R deployments, the magnanimous-turbo-chainsaw (MTC), tooling repo has been added as a submodule of rpc-maas. This repository contains resources intended to be used across all RPC Toolstack components. As part of the initial commit, the generate-environment-vars.yml playbook and bootstrap-embedded-ansible.sh wrapper have been symlinked into the main directory structure.
• 1. Added missing supported memcached version 1.4.39 for osp13
  2. Adjusted client.my.conf tempalte to take internal ip for osp13
  3. Rearranged maas-agent vars variables.
• Add agent/poller installation for rpcr-maas
• Add support for providing deployment overrides into MaaS deploy environment
• Add support Glance, Cinder and Heat support for OSP13
• Nova, neutron, and swift checks are now supporting rpcr OSP13
• osp-13 dynamic inventory now includes logic automatically loading up stackrc env file. maas_stackrc will be set by dynamic inventory too.
• The rpcr-dynamic inventory now is able to set OS_CACERT environment variable automatically.
• Adjust ceph plugin code and check templates to make maas ceph cheks working with rpcr osp-13
• This PR enables osp-13 gate for MAAS. It does so by using a wrapper script test-ansible-functional-osp-mnaio.sh to invoke tests on OSP-13 MNAIO director node. It also fixed some discrepencies between several checks and other missing steps in ansible dynamic inventory scripts as well.
• Horizon and keystone checks are now supporting rpcr osp13.
• Host and infra checks are now supporting rpcr OSP13. Also a new check type of pacemaker has been added.

Upgrade Notes

• Deployment processes can skip the step sourcing /home/stack/stackrc. Also maas_stackrc can be removed from variable overrides
• Deployment process can skip the step setting OS_CACERT.

Bug Fixes

• Add agent token cleanup to raxmon and move execution of entity delete and agent token delete functions to the test-ansible-functional.sh trap.
• Enable but not start rax-mon service in maas-agent-setup
• Fix nfs_check.py to properly detect nfs file systems
• Fix rpc-maas-tool.py to report excludedchecks as the check will always be created.
• Gating - Remove 'enable_ironic' from aio-create.sh as the functionality is not currently supported and it is causing numerous critical alarms preventing maas-verify.yml from passing. This can be added back to the necessary release versions when required.
• Gating - Remove maas-tigkstack-all.yml from test playbooks. This functionality is no longer supported as a product addition.
• Gating - Enable 'maas_verify_registration' in gating to validate alarms are created in the api.
• Gating - Enable 'maas_verify_status' in gating to validate alarms are not critical after deployments.
• Gating - Tweak existing variable overrides for different gating release versions and actually use these overrides.
• Gating - Remove tests/vars alarm exclusions, frequency changes, and update excluded checks to include the proper checks per release version.
• Gating - Disable 'maas_remote_check' to exclude remote.http checks from testing as they will always fail in AIOs because of private endpoints.
• Gating - Properly enable overrides for user_$VERSION_vars.yml based on the gate release.
• Gating - Run maas-restart.yml after deploying rally performance checks.
• Move the RH subscription-unregister to the gate exit tasks.
• Fix a missed instantiation of ansible_host in rally plays without a container_address default.

1.7.10

Release Notes

1.7.10

New Features

• Managed Kubernetes k8s API checks have been moved to the maas agent inside the kubernetes cluster to align better with the device id of the cluster under monitoring and simplify management of alerts.

Bug Fixes

• • The conntrack_count.py plugin is now checking for network namespaces listed at /var/run/netns and retreives the iptables connection tracking infomation for each namespace. This ensures that embedded network namespaces are alerted in case connection tracking hashes are about to exceed a configurable threshold. Due to the limited availability of MAAS metrics per alarm, only the namespace with the higest connection tracking count is reported.
• Latest influxdb-relay installation requires newer version of go.

1.7.9

Release Notes

1.7.9

New Features

• Added tags to Run MaaS verify local task in maas-verify.yml for ASC test integration

Bug Fixes

• The maas_common.py is setting OS_VOLUME_API_VERSION and OS_IMAGE_API_VERSION with a default of None. This caused calls to get_auth_details to fail a key check for 'None' causing nova and neutron checks to fail of those variables are not being set in the environment anywhere.
• Some 1.7.8 checks (nova api) breaks in absence of maasrc file. This fix it.

1.7.8

Release Notes

1.7.8

New Features

• Add maas rally scheme defaulting to the same as maas

Upgrade Notes

• Refactored maas-common keystone libs to work with RPC-R

Bug Fixes

• Fixes new queens gate
• The rpc-r additions were unsetting a variable causing a set_fact task to fail on rpc-o environments. This is to make sure both checks use a different registered variable and set the fact based on the rpc-o vs rpc-r conditionals.
• Map nova_console_type and nova_console_port back to a maas_* variable. These are defaults in OSA and aren't pulled in properly from the ansible role. This will allow overrides to be configured in a less obtrusive way.

1.7.7

Release Notes

1.7.7

Bug Fixes

• Fix the bonding mii split logic to remove leading and trailing spaces and calculate the slave_count outside the loop determining if a slave is down.

1.7.6

Release Notes

1.7.6

New Features

• Adds support for designatwe checks in RPC-R

Upgrade Notes

• Add slave_count to ensure the bond has redundancy and check explicitly for not "up" This ensures a bond slave that is marked down (should only occur in a maintenance window) triggers an alert.

Bug Fixes

• Remove - vars/maas-{{ ansible_distribution | lower }}.yml include as it is redundant and problematic for RPC-R Maas work

1.7.5

Release Notes

1.7.5

Prelude

This is a barebones script to add RPC-R support for octavia and mk8s

New Features

• Added protocol and port option support for ironic checks, so it can support ssl and/or different port.
• • Added new bonding interface metric host_bonding_iface_<iface>_slave_down
• nova/neutron service check plugins now produces string metrics. When API is down and agents/service cannot reach it, monitoring agent will generate a warning of message "cannot reach API" by emitting out a metric with a string metric "<service/agent> cannot reach API". And in other scenarios, the plugin will emit "Yes"/"No" metric to indicate service/agent is up/down.
• 'site-osp.yml' will prepare and install octavia and mk8s maas checks on an RPC-R environment. This feature is very limited.

Known Issues

• After the script is run the maas agent needs to be restarted manually for changes taking effect.

Upgrade Notes

• The elasticsearch maas check is not doing a whole lot and is more noise than value. This is on top of the fact that the elasticsearch checks are not made for a deployment where elasticsearch/logstash are deployed in a high performance cluster. This change also removes the filebeat check.

Bug Fixes

• notification queue matching regex is not working properly. Some messages may or may not start with '/' and may or may not have 'versioned_'. This PR fixes it.

1.7.4

Release Notes

1.7.4

New Features

• Added the ability to override max_row_limit in influxdb.conf
• Add the ability to specify the Ceph rados Gateway listen address for MaaS monitoring. Using the existing ceph-ansible vars which are, in order of precedence. radosgw_address which is used to specify the address. radosgw_address_block which can be used to specify the address block in CIDR format/ radosgw_interface which can be used to specify the interface on which RadosGW is listening, for example br-storage. If none of these options are specified the ansible_host address will be used.

Bug Fixes

• Fixes the wrong auth servcie and the typo when calling the managed_k8s_services_local_check.py script.

1.7.3

Release Notes

1.7.3

New Features

• Managed Kubernetes installs several lxc-conatiners for various auhentication related functionality (etp, etg, auth, ui). This will monitor if the respective processes are up and the API endpoints are accessible. UI also includes checks for the LB whereas the other services only have local checks.

Bug Fixes

• Add an extra maas_env_product, defaults to 'rpco'. In ceph environment it should be set to 'ceph'. Also maas_rpco_dir has been renamed to maas_product_dir, and osa version will be left out in the case of ceph environment
• Update metadata macro to include lb_ssl_expiry_check and private_lb_ssl_expiry_check in the API category.