Currently, I have the following playbook:
- hosts: "{{env}}_{{product}}"
name: "tools"
sudo: yes
vars_prompt:
product: "Which product would you like to deploy [all|proxy|query|rest]?"
env: "Which environment should we deploy to [dev|qa|test|prod]?"
roles:
- { role: proxy, when: "product == 'all' or product == 'proxy'" }
- { role: query, when: "product == 'all' or product == 'query'" }
- { role: rest, when: "product == 'all' or product == 'rest'" }
All of my groups are stored in a single ./inventory/all file, e.g.,:
# -----------------------------------------------------------------------------------
# --------------
# ### DEV ###
# --------------
# -----------------------------------------------------------------------------------
[dev_all:children]
dev_redis
dev_query
dev_rest
dev_proxy
[dev_redis]
.hosts
[dev_query]
.hosts
[dev_rest]
.hosts
[dev_proxy:children]
dev_proxy-dc1
dev_proxy-dc2
dev_proxy-dc3
[dev_proxy-dc1]
.hosts
[dev_proxy-dc2]
.hosts
[dev_proxy-dc3]
.hosts
# -----------------------------------------------------------------------------------
# --------------
# ### PROD ###
# --------------
# -----------------------------------------------------------------------------------
[prod_all:children]
prod_redis
prod_query
prod_rest
prod_proxy
[prod_redis]
.hosts
[prod_query]
.hosts
[prod_rest]
.hosts
[prod_proxy:children]
prod_proxy-dc1
prod_proxy-dc2
prod_proxy-dc3
[prod_proxy-dc1]
.hosts
[prod_proxy-dc2]
.hosts
[prod_proxy-dc3]
.hosts
I can't help but feel like I'm making this overly complex though. I'm trying to avoid requiring people to pass in tags or inventory files. But now I'm not sure what the best way to go about allowing deploys for things like redis hosts, which aren't really what we would consider a "product", but still requires it's own group since it's hosted on it's own set of hosts. I could just add it to the current list, [all|proxy|query|rest|redis], but... it seems like there should be a way to specify redis and another product, but at the same time not requiring both... I don't know how though.
I wanted to have something where you could say
"I want to deploy proxy to dev, and let's update redis while we're at it"
-- is this possible with my current setup?
This doesn't feel like the best way to handle this.
Instead, I'd structure my playbooks individually to target specific roles and I'd equally set my inventories to only cover a certain environment rather than everything.
If you want to make it so that people don't have to pass in the inventory file or the specific playbook then I'd wrap the ansible-playbook command(s) in some form of wrapper script.
This allows people to use your playbooks a lot more granularly and flexible as you need it but still offer the same ability to offer presets through your wrapper script.
Related
I want to be able to setup via ansible different "environments" (sets of same variables), e.g. production, testing, staging, etc.
I would like to:
define which environments to set up (setup_environments) at some global place
run a role r1 once for each environment in setup_environments
run a role r2 without setting its environment, but r2 should have access both to the variable setup_environments AND be able to load the defined environments for certain tasks.
Is this possible? 3. should be possible with include_vars, but I mention it to show that I really need the "setup_environments" variable and can't simply run ansible multiple times including each environment separately with -i.
Reading your requirement it seems that you need to structure your inventory only. Let's assume your inventory looks like
inventory.ini
[environment:children]
prod
qa
dev
webserver
database
[environment:vars]
# No global variables here / yet
[prod:children]
webserver_prod
database_prod
[qa:children]
webserver_qa
database_qa
[dev:children]
webserver_dev
database_dev
# Logical services nodes
[webserver:children]
webserver_prod
webserver_qa
webserver_dev
[database:children]
database_prod
database_qa
database_dev
# List of all hosts
# [prod:children]
[webserver_prod]
web.prod.example.com
[database_prod]
primary.prod.example.com
secondary.prod.example.com
# [qa:children]
[webserver_qa]
web.qa.example.com
[database_qa]
primary.qa.example.com
secondary.qa.example.com
# [dev:children]
[webserver_dev]
web.dev.example.com
[database_dev]
primary.dev.example.com
secondary.dev.example.com
you may get familar with the structure than by
ansible-inventory dev --graph
ansible-inventory database_dev --graph
...
It will be possible to apply playbooks, tasks, etc. just to certain nodes via
ansible-playbook database_dev ...
Furthermore you should introduce group_vars like
all
prod
qa
dev
webserver_prod
database_prod
...
All depdends on the structure of the inventory but makes it easier afterwards.
I have a problem with groups variables.
Example: I have two inventory groups group_A and group_B, and also have the same name files in group_vars:
inventories/
hosts.inv
[group_A]
server1
server2
[group_B]
server3
server4
group_vars/
group_A - file
var_port: 9001
group_B - file
var_port: 9002
The problem is when i execute:
ansible-playbook playbooks/playbook.yml -i inventories/hosts.inv -l group_B
playbook was executed for proper scope of servers (server3, server4) but it takes variables from group variables file group_A.
expected result: var_port: 9002
in realty : var_port: 9001
ansible 2.4.2.0
BR Oleg
I included ANSIBLE_DEBUG , and what i have found:
2018-05-03 15:23:23,663 p=129458 u=user | 129458 1525353803.66336: Loading data from /ansible/inventories/prod/group_vars/group_B.yml
2018-05-03 15:23:23,663 p=129458 u=user | 129661 1525353803.66060: in run() - task 00505680-eccc-d94e-2b1b-0000000000f4
2018-05-03 15:23:23,664 p=129458 u=user | 129661 1525353803.66458: calling self._execute()
2018-05-03 15:23:23,665 p=129458 u=user | 129458 1525353803.66589: Loading data from /ansible/inventories/prod/group_vars/group_A.yml
on playbook execution ansible scan all files with variables in folder group_vars which have variable "var_port", last will win.....
as you can found in another topic:
Ansible servers/groups in development/production
and from documentation:
http://docs.ansible.com/ansible/latest/user_guide/playbooks_variables.html#variable-precedence-where-should-i-put-a-variable
Note
Within any section, redefining a var will overwrite the previous instance. If multiple groups have the same variable, **the last one loaded wins**. If you define a variable twice in a play’s vars: section, the **2nd one wins**.
For me now NOT clear how to manage configuration files. In this case I must use unique variables names for each group, but it is not possible regarding roles, or should I use include_vars when i call playbook?
Super example how to manage variables files in multistage environment from DigitalOcean
How to Manage Multistage Environments with Ansible
I believe that the problem here, while not explicitly stated in the original question, is that Server{1,2} and Server{3,4} are actually the same servers in 2 different groups at the same level.
I ran into this problem which caused me to do some digging. I don't agree with it, but it is as designed. This was even fixed with full compatibility and the pull request was rejected
Discussion
Pull Request
I am pretty new to Jinja2, and I'm wondering how to achieve this.
Say I have the following vars:
---
servers:
192.168.0.1:
names:
- foo.example.com
- foo
exports:
data:
foo1: /disks/foo1
foo2: /disks/foo2
192.168.0.2:
...
I want to create a symlink /data/foo1 to /disks/foo1 and /data/foo2 to /disks/foo2, but only on foo server; on other servers, make symlinks to their respective exports. So I thought file status=link with_items=... would be the correct thing to do. In Python, I can get the array I need using the following logic:
[
{ 'mount': mount, 'export': export }
for ip, server in servers.iteritems()
if ansible_hostname in server['names']
and 'exports' in server
and 'data' in server['exports']
for mount, export in server['exports']['data'].iteritems()'
]
I don't know how to do this in Jinja2. I wanted to do something like
{{ servers | select('ansible_hostname in self.names') | ... }}
but that doesn't work. Would I need to create a plugin for this logic? Or is my approach all wrong and I should rethink the structure of my servers data?
Answer from my comment:
Usually you want to use inventory_hostname variable – it is what you use as host name in inventory.
servers[ansible_hostname] will access servers' key with name of ansible_hostname's value.
Just for curiosity, you can check out this (complex filter chain) and this (runtime object construction).
Say the minion host has a default yaml configuration named myconf.yaml. What I want to do is to edit parts of those yaml entries using values from a pillar. I can't even begin to think how to do this on Salt. The only think I can think of is to run a custom python script on the host via cmd.run and feed it with input via arguments, but this seems overcomplicated.
I want to avoid file.managed. I cannot use a template, since the .yaml file is big, and can change by external means. I just want to edit a few parameters in it. I suppose a python script could do it but I thought salt could do it without writing s/w
I have found salt.states.file.serialize with the merge_if_exists option, I will try this and report.
You want file.serialize with the merge_if_exists option.
# states/my_app.sls
something_conf_file:
file.serialize:
- name: /etc/my_app.yaml
- dataset_pillar: my_app:mergeconf
- formatter: yaml
- merge_if_exists: true
# pillar/my_app.sls
my_app:
mergeconf:
options:
opt3: 100
opt4: 200
On the target, /etc/my_app.yaml might start out looking like this (before the state is applied):
# /etc/my_app.yaml
creds:
user: a
pass: b
options:
opt1: 1
opt2: 2
opt3: 3
opt4: 4
And would look like this after the state is applied:
creds:
user: a
pass: b
options:
opt1: 1
opt2: 2
opt3: 100
opt4: 200
As far as I can tell this uses the same algorithm as pillar merges, so e.g. you can merge or partially overwrite dictionaries, but not lists; lists can only be replaced whole.
This can be done for both json and yaml with file.serialize. Input can be inline on the state or come from a pillar. A short excerpt follows:
state:
cassandra_yaml:
file:
- serialize
# - dataset:
# concurrent_reads: 8
- dataset_pillar: cassandra_yaml
- name: /etc/cassandra/conf/cassandra.yaml
- formatter: yaml
- merge_if_exists: True
- require:
- pkg: cassandra-pkgs
pillar:
cassandra_yaml:
concurrent_reads: "8"
I'm configuring /etc/security/limits.conf with Ansible' new module pam_limits.
What I've succeeded at:
Setting values for specific domain and type in the default limits.conf. (A new string is appended to the end of the file).
Changing values (the string gets rewritten).
The problem is when I want to completely remove the setting. E.g. I don't want to save core dumps anymore. How should I use pam_limits to remove the string completely?
I've managed to develop the following workaround, but I don't consider it good. It doesn't remove the string but rather sets the limit to 0, which may be not the same.
roles/myrole/tasks/main.yaml
...
- name: enable core dumps for myservice
pam_limits: domain='*' limit_type='-' limit_item=core value="{{ 'unlimited' if myrole_save_core_dumps else 0 }}"
...
group_vars/myhosts.yaml:
myrole_save_core_dumps: true
myservice.yaml
hosts: myhosts
become: yes
roles:
- myrole
I believe this would be an feature which is currently not implemented. But there is a feature request on github for this feature.