rackerlabs / auter-manager Goto Github PK

Re-review #28 The logic is incorrect

Create new logmons when auter is deployed

I have started working on reformatting the Ansible playbooks, bringing some modifications along as well.

This is an enhancement / work in progress, not finished yet but opening issue to let people aware of on-going work. The PR will be created once I have finished the rewrite.

Current state of things:

• no matter what is set in the auter_config.csv file, the installation of Auter currently blindly enforce the yum excludes settings.

Problem:
It is pretty easily to overlook this and end up with no yum exclusions after the installation of auter through auter-manager.

Suggestion:

• using a none value into the CSV file when we want to ensure that any yum excludes are removed during auter installation
• if no value is set at all, do not overwrite yum.conf

Questions ? Comments ?

Having a valid ID in there is asking for trouble whether they are test devices or not. It should be swapped for a basic placeholder that will error out if someone tries to run the playbook without modifying the config.

Something along the lines of {{ device_id }}.

BUG:

Auter is configured to run and apply updates before midnight say at 11:30pm on 20th July 1997 , it creates its .pre snapshot files before midnight passes and auter is then instructed to post midnight reboot (18th July 1997), on reboot the .post snapshot fails and is unable to run since its looking for the same timestamp .pre entries ( from the previous day) to check against which it believes does not exist.becuase a new day has already come into effect.

Originally spotted on modular branch, PR coming soon for master. modular is being updated as well.

In some particular cases, epel check (yum repolist | grep epel | wc -l) does not return an stdout variable.
We need to catch this case to request manual check on the device for those cases.

Raised initially as part of PR #48

The playbook also set up the extra nimbus config always at the same place:

- name: Create nimbus monitors in "{{ nimbus_path }}"
  tags: configure,nimbus
  blockinfile:
    backup: yes
    insertbefore: "</profiles>"
    marker: "# {mark} ANSIBLE MANAGED AUTER BLOCK"
    dest: /opt/nimsoft/probes/system/logmon/logmon.cfg

If an auter patching starts on one days, say around 23:50, and the apply/ reboot take more than a few minutes, we can end up in 2 different configsnap tags/directory created, which also means the pre/post comparison does not run.

Related to #38

TASK [add the epel repo] ****************************************************************************************************************************************************************************************************************************************************************
fatal: []: FAILED! => {"msg": "The conditional check '( 'no package' in result.msg|lower ) and epelPresent.stdout|int == 0' failed. The error was: error while evaluating conditional (( 'no package' in result.msg|lower ) and epelPresent.stdout|int == 0): 'dict object' has no attribute 'msg'\n\nThe error appears to have been in '/home/nich1145/Downloads/auter-manager/installer/auter_installer.yml': line 45, column 7, but may\nbe elsewhere in the file depending on the exact syntax problem.\n\nThe offending line appears to be:\n\n\n    - name: add the epel repo\n      ^ here\n"}
fatal: []: FAILED! => {"msg": "The conditional check '( 'no package' in result.msg|lower ) and epelPresent.stdout|int == 0' failed. The error was: error while evaluating conditional (( 'no package' in result.msg|lower ) and epelPresent.stdout|int == 0): 'dict object' has no attribute 'msg'\n\nThe error appears to have been in '/home/nich1145/Downloads/auter-manager/installer/auter_installer.yml': line 45, column 7, but may\nbe elsewhere in the file depending on the exact syntax problem.\n\nThe offending line appears to be:\n\n\n    - name: add the epel repo\n      ^ here\n"}

Looks like I never tested this before!

PR to fix this coming up.