modify fact gathering to use local facts

This change makes the retention gathering operation faster by storing
the retention values as "local facts". The local facts are then
referenced in templates loading from the local fact file instead of
running repetitive queries which are slow making very large deployments
cumbersome. To make the retention policy fact gathering process smarter
it will now automatically refresh if undefined or should the
elasticsearch cluster size change. This will ensure we're improving
speed of execution while also catering to the needs of deployers to
grow, or shrink, elasticsearch cluster sizes.

Documentation has been added regarding the new option and why it may be
of use to depoyers.

Change-Id: I3936ee94461ac39fb8bc78dc2c873e6067552461
Signed-off-by: Kevin Carter <kevin.carter@rackspace.com>
This commit is contained in:
Kevin Carter 2018-11-27 13:42:42 -06:00
parent 2923df2295
commit e08c58dd15
7 changed files with 112 additions and 35 deletions

View File

@ -570,6 +570,22 @@ state variable, `elk_package_state`, to latest.
ansible-playbook site.yml $USER_VARS -e 'elk_package_state="latest"'
Forcing the Elasticsearch cluster retention policy to refresh
-------------------------------------------------------------
To force the cluster retention policy to refresh set `elastic_retention_refresh`, to
"yes". When setting `elastic_retention_refresh` to "yes" the retention policy will forcibly
be refresh across all hosts. This option should only be used when the Elasticsearch storage
array is modified on an existing cluster. Should the Elasticseach cluster size change
(nodes added or removed) the retention policy will automatically be refreshed on playbook
execution.
.. code-block:: bash
cd /opt/openstack-ansible-ops/elk_metrics_6x
ansible-playbook site.yml $USER_VARS -e 'elastic_retention_refresh="yes"'
Trouble shooting
----------------

View File

@ -30,6 +30,13 @@
when:
ansible_service_mgr != 'systemd'
- name: Refresh local facts
setup:
filter: ansible_local
gather_subset: "!all"
tags:
- always
- name: Ensure virtualenv is installed
package:
name: "{{ curator_distro_packages }}"

View File

@ -15,10 +15,10 @@
{% set action_items = [] -%}
{# Delete index loop #}
{% for key in elastic_beat_retention_policy_keys -%}
{% for key in (ansible_local['elastic']['retention']['elastic_beat_retention_policy_keys'] | from_yaml) -%}
{% set delete_indices = {} -%}
{# Total retention size in days #}
{% set _index_retention = hostvars[inventory_hostname]['elastic_' + key + '_retention'] -%}
{% set _index_retention = ansible_local['elastic']['retention']['elastic_' + key + '_retention'] -%}
{% set index_retention = ((_index_retention | int) > 0) | ternary(_index_retention, 1) | int %}
{% set _ = delete_indices.update(
{

View File

@ -15,10 +15,10 @@
{% set action_items = [] -%}
{# Delete index loop #}
{% for key in elastic_beat_retention_policy_keys -%}
{% for key in (ansible_local['elastic']['retention']['elastic_beat_retention_policy_keys'] | from_yaml) -%}
{% set delete_indices = {} -%}
{# Total retention size in gigabytes #}
{% set _index_size = ((hostvars[inventory_hostname]['elastic_' + key + '_size'] | int) // 1024) -%}
{% set _index_size = ((ansible_local['elastic']['retention']['elastic_' + key + '_size'] | int) // 1024) -%}
{% set index_size = ((_index_size | int) > 0) | ternary(_index_size, 1) | int %}
{% set _ = delete_indices.update(
{

View File

@ -94,3 +94,6 @@ elastic_beat_retention_policy_hosts:
packetbeat:
weight: 1
hosts: "{{ groups['hosts'] | default([]) }}"
# Refresh the elasticsearch retention policy local facts.
elastic_retention_refresh: false

View File

@ -13,40 +13,90 @@
# See the License for the specific language governing permissions and
# limitations under the License.
- name: Query es storage
uri:
url: "http://{{ coordination_nodes[0] }}/_nodes/{{ (data_nodes | map('extract', hostvars, 'ansible_host') | list) | join(',') }}/stats/fs"
method: GET
register: elk_data
environment:
no_proxy: "{{ coordination_nodes[0].split(':')[0] }}"
until:
- elk_data is success and elk_data['json'] is defined
retries: 5
delay: 30
run_once: true
- name: Ensure local facts directory exists
file:
dest: "/etc/ansible/facts.d"
state: directory
group: "root"
owner: "root"
mode: "0755"
recurse: no
- name: Set retention keys fact
set_fact:
es_storage_json: "{{ elk_data['json'] }}"
- name: Initialize local facts
ini_file:
dest: "/etc/ansible/facts.d/elastic.fact"
section: "retention"
option: cacheable
value: true
- name: Load retention algo variables
include_vars: "calculate_index_retention_{{ elastic_index_retention_algorithm }}.yml"
- name: Refresh local facts
setup:
filter: ansible_local
gather_subset: "!all"
tags:
- always
- name: Set retention facts (mb size)
set_fact: "elastic_{{ item.key }}_size={{ item.value }}"
when:
- hostvars[inventory_hostname]["elastic_" ~ item.key ~ "_size"] is undefined
with_dict: "{{ es_storage_per_index }}"
- name: Retention storage block
block:
- name: Query es storage
uri:
url: "http://{{ coordination_nodes[0] }}/_nodes/{{ (data_nodes | map('extract', hostvars, 'ansible_host') | list) | join(',') }}/stats/fs"
method: GET
register: elk_data
environment:
no_proxy: "{{ coordination_nodes[0].split(':')[0] }}"
until:
- elk_data is success and elk_data['json'] is defined
retries: 5
delay: 30
run_once: true
- name: Set retention facts (days)
set_fact: "elastic_{{ item.key }}_retention={{ item.value }}"
when:
- hostvars[inventory_hostname]["elastic_" ~ item.key ~ "_retention"] is undefined
with_dict: "{{ es_days_per_index }}"
- name: Set retention keys fact
set_fact:
es_storage_json: "{{ elk_data['json'] }}"
- name: Set retention keys fact
set_fact:
elastic_beat_retention_policy_keys: "{{ elastic_beat_retention_policy_hosts.keys() | list }}"
- name: Load retention algo variables
include_vars: "calculate_index_retention_{{ elastic_index_retention_algorithm }}.yml"
tags:
- always
- name: Set storage fact
ini_file:
dest: "/etc/ansible/facts.d/elastic.fact"
section: "retention"
option: "cluster_nodes"
value: "{{ groups['elastic-logstash'] | length }}"
- name: Set retention policy fact
ini_file:
dest: "/etc/ansible/facts.d/elastic.fact"
section: "retention"
option: "elastic_beat_retention_policy_keys"
value: "{{ elastic_beat_retention_policy_hosts.keys() | list }}"
- name: Set size fact
ini_file:
dest: "/etc/ansible/facts.d/elastic.fact"
section: "retention"
option: "elastic_{{ item.key }}_size"
value: "{{ item.value }}"
with_dict: "{{ es_storage_per_index }}"
- name: Set retention fact
ini_file:
dest: "/etc/ansible/facts.d/elastic.fact"
section: "retention"
option: "elastic_{{ item.key }}_retention"
value: "{{ item.value }}"
with_dict: "{{ es_days_per_index }}"
- name: Refresh local facts
setup:
filter: ansible_local
gather_subset: "!all"
tags:
- always
when:
- (ansible_local['elastic']['retention']['cluster_nodes'] is undefined) or
((groups['elastic-logstash'] | length) != (ansible_local['elastic']['retention']['cluster_nodes'] | int)) or
(elastic_retention_refresh | bool)

View File

@ -33,6 +33,7 @@
OK with the old rollup job being purged, add the following option
`elastic_allow_rollup_purge=yes` to the command line and rerun the
playbook.
run_once: true
when:
- check_rollup['json']['jobs'] | length > 0
- not elastic_allow_rollup_purge | bool
@ -44,7 +45,7 @@
min_days_until_rollup: |-
{% set index_retention = [] %}
{% for item in ansible_play_hosts %}
{% set _ = index_retention.append(hostvars[item]['elastic_' + index_name + '_retention'] | int) %}
{% set _ = index_retention.append(ansible_local['elastic']['retention']['elastic_' + index_name + '_retention'] | int) %}
{% endfor %}
{{ index_retention | min }}
run_once: true