I have the following code which isn't throwing an error but the fact is empty
- shell: echo '{{ p }}'
register: results
- debug:
var: results
- set_fact:
myrepo: "{{ results.stdout | regex_search(regexp,'\\1') | default ( {'0':'global'} ) }}"
vars:
regexp: '(.*)/(.*)'
Here is the output
TASK [command] **************************************************************************************************************************************************************************************
changed: [localhost]
TASK [debug] ****************************************************************************************************************************************************************************************
ok: [localhost] => {
"results": {
"changed": true,
"cmd": "echo 'tim'",
"delta": "0:00:00.095831",
"end": "2017-09-06 16:37:19.977023",
"rc": 0,
"start": "2017-09-06 16:37:19.881192",
"stderr": "",
"stderr_lines": [],
"stdout": "tim",
"stdout_lines": [
"tim"
]
}
}
TASK [set_fact] *************************************************************************************************************************************************************************************
ok: [localhost]
TASK [debug] ****************************************************************************************************************************************************************************************
ok: [localhost] => {
"myrepo": ""
}
The command is ansible-playbook -i hosts -c local file.yml --extra-vars "p=tim" I want myrepo to be global if the regex results are empty
Without any parameters default filter is triggered only when value is Undefined. But result of unmatched regexp is an empty string, which is not Undefined. You may want to set boolean flag:
- set_fact:
myrepo: "{{ results.stdout | regex_search(regexp,'\\1') | default('global', boolean=True) }}"
vars:
regexp: '(.*)/(.*)'
Related
I have a question about dicts and lists.
What I want to achieve is, have the key and value from a seperate list saved as fact/hostvar for each matching host.
I'm getting a list from a Confluence API that looks like this (abbreviated):
[
{
"title": "MACHINE1",
"_links": {
"tinyui": "/x/1234"
}
},
{
"title": "MACHINE2",
"_links": {
"tinyui": "/x/5678"
}
},
{
"title": "MACHINE3",
"_links": {
"tinyui": "/x/9876"
}
}
]
What worked to get each individual item (just to debug, and show that the loop itself works) is:
- name: DEBUG specific item in list of get_children.json.results
debug:
msg: "{{ item.title }} {{ item._links.tinyui }}"
loop:
"{{ get_children.json.results }}"
delegate_to: 127.0.0.1
Ansible Output (here: output for only one machine):
"msg": "MACHINE1 /x/1234"
Machine Hostnames:
Yes, they are lowercase in my inventory, and in the above list output they are uppercase. But I guess a simple item.title|lower would do fine.
machine1
machine2
machine3
How can I now match the item.title with ansible_hostname and save above API Output as a fact for each machine?
And for clarification: item.title|lower == ansible_hostname
I hope it gets clear to what I want to achieve and thanks to everyone in advance :)
EDIT: Thanks to both answers I managed to get it to work. Using '(?i)^'+VAR+'$' and some other conditional checks you guys posted definitely helped. :)
In a nutshell, given the inventories/tinyui/main.yml inventory:
---
all:
hosts:
machine1:
machine2:
machine3:
i.do.not.exist:
The folowing tinyui.yml playbook:
---
- hosts: all
gather_facts: false
vars:
# In real life, this is returned by your API call
get_children:
json:
results: [
{
"title": "MACHINE1",
"_links": {
"tinyui": "/x/1234"
}
},
{
"title": "MACHINE2",
"_links": {
"tinyui": "/x/5678"
}
},
{
"title": "MACHINE3",
"_links": {
"tinyui": "/x/9876"
}
}
]
# This won't be defined before you call the API which
# returns and registers the correct result. If there is
# no match for host in the returned json, '!no uri!' will
# be returned below. Adapt with a default uri if needed
tinyui: "{{ get_children.json.results
| selectattr('title', '==', inventory_hostname | upper)
| map(attribute='_links.tinyui')
| default(['!no uri!'], true) | first }}"
tasks:
# In real life, you would have called your API
# and registered the result in `get_children` e.g.
# - name: get info from confluence
# uri:
# url: <confluence api endpoint url>
# <more parameters here>
# run_once: true
# delegate_to: localhost
# register: get_children
- name: Display tinyui for host
debug:
msg: "tinyui for host {{ inventory_hostname }} is {{ tinyui }}"
Gives:
$ ansible-playbook -i inventories/tinyui/ tinyui.yml
PLAY [all] ***********************************************************************************************************************
TASK [Display tinyui for host] ***************************************************************************************************
ok: [machine1] => {
"msg": "tinyui for host machine1 is /x/1234"
}
ok: [machine2] => {
"msg": "tinyui for host machine2 is /x/5678"
}
ok: [machine3] => {
"msg": "tinyui for host machine3 is /x/9876"
}
ok: [i.do.not.exist] => {
"msg": "tinyui for host i.do.not.exist is !no uri!"
}
PLAY RECAP ***********************************************************************************************************************
i.do.not.exist : ok=1 changed=0 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
machine1 : ok=1 changed=0 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
machine2 : ok=1 changed=0 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
machine3 : ok=1 changed=0 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
You can filter the list of dictionaries by title, taking out the right dict.
You can do this with the following line:
"{{ hostsdata | selectattr('title', 'match', '(?i)^'+host+'$') | first }}"
With selectattr you filter your list of dicts by the title, where this must match '(?i)^'+host+'$'.
The (?i) is the ignore case inline flag for pattern matching, concatenated with the hostname (case does not matter because of ignore case flag). ^...$ specifies that the whole string must match, from start to end.
Since selectattr returns a list as result, you can use first to take out the first element of the list.
Instead of using (?i) you can also set the ignorecase parameter, which will look like this:
"{{ hostsdata | selectattr('title', 'match', '^'+host+'$', 'ignorecase=true') | first }}"
Both variants work equivalently.
Entire playbook:
---
- hosts: localhost
gather_facts: no
vars:
hostsdata:
- {
"title": "MACHINE1",
"_links": {
"tinyui": "/x/1234"
}
}
- {
"title": "MACHINE2",
"_links": {
"tinyui": "/x/5678"
}
}
- {
"title": "MACHINE3",
"_links": {
"tinyui": "/x/9876"
}
}
tasks:
- debug:
var: hostsdata
- name: Pick out host specific dict
set_fact:
machine_data: "{{ hostsdata | selectattr('title', 'match', '(?i)'+host) | first }}"
vars:
host: machine3
- debug:
var: machine_data
- debug:
msg: "{{ machine_data.title }} {{ machine_data._links.tinyui }}"
Resulting output:
TASK [debug] ***********************************************************************************************************
ok: [localhost] => {
"hostsdata": [
{
"_links": {
"tinyui": "/x/1234"
},
"title": "MACHINE1"
},
{
"_links": {
"tinyui": "/x/5678"
},
"title": "MACHINE2"
},
{
"_links": {
"tinyui": "/x/9876"
},
"title": "MACHINE3"
}
]
}
TASK [Pick out host specific dict] *************************************************************************************
ok: [localhost]
TASK [debug] ***********************************************************************************************************
ok: [localhost] => {
"machine_data": {
"_links": {
"tinyui": "/x/9876"
},
"title": "MACHINE3"
}
}
TASK [debug] ***********************************************************************************************************
ok: [localhost] => {
"msg": "MACHINE3 /x/9876"
}
To filter multiple machines, here is another example:
- debug:
msg: "{{ md.title }} {{ md._links.tinyui }}"
when: md | length
vars:
md: "{{ hostsdata | selectattr('title', 'match', '(?i)^'+item+'$') | first | default('') }}"
with_items:
- MachINE1
- MACHINE2
- machine3
- unknown
Add a default('') and a when: to skip a non-existent hostname.
Output:
TASK [debug] ***********************************************************************************************************
ok: [localhost] => (item=MachINE1) => {
"msg": "MACHINE1 /x/1234"
}
ok: [localhost] => (item=MACHINE2) => {
"msg": "MACHINE2 /x/5678"
}
ok: [localhost] => (item=machine3) => {
"msg": "MACHINE3 /x/9876"
}
skipping: [localhost] => (item=unknown)
Convert the titles to lowercase
titles: "{{ get_children.json.results|
map(attribute='title')|
map('lower')|
map('community.general.dict_kv', 'title')|
list }}"
gives
titles:
- title: machine1
- title: machine2
- title: machine3
Replace the lowercase titles and create a dictionary
title_links: "{{ get_children.json.results|
zip(titles)|
map('combine')|
items2dict(key_name='title', value_name='_links') }}"
gives
title_links:
machine1:
tinyui: /x/1234
machine2:
tinyui: /x/5678
machine3:
tinyui: /x/9876
Put these declarations, for example, into the group_vars
shell> cat group_vars/all/title_links.yml
titles: "{{ get_children.json.results|
map(attribute='title')|
map('lower')|
map('community.general.dict_kv', 'title')|
list }}"
title_links: "{{ get_children.json.results|
zip(titles)|
map('combine')|
items2dict(key_name='title', value_name='_links') }}"
Now, you can use the dictionary. For example, given the inventory
shell> cat hosts
10.1.0.11 ansible_hostname=machine1
10.1.0.12 ansible_hostname=machine2
10.1.0.13 ansible_hostname=machine3
the playbook
- hosts: all
gather_facts: false
vars:
get_children:
json:
results:
- _links: {tinyui: /x/1234}
title: MACHINE1
- _links: {tinyui: /x/5678}
title: MACHINE2
- _links: {tinyui: /x/9876}
title: MACHINE3
tasks:
- debug:
msg: "My links: {{ title_links[ansible_hostname] }}"
gives (abridged)
TASK [debug] *********************************************************
ok: [10.1.0.11] =>
msg: 'My links: {''tinyui'': ''/x/1234''}'
ok: [10.1.0.12] =>
msg: 'My links: {''tinyui'': ''/x/5678''}'
ok: [10.1.0.13] =>
msg: 'My links: {''tinyui'': ''/x/9876''}'
I have an ansible output from which I want to remove all empty (none) elements for a specific attribute. Here is the example list:
"resources": [
{
"id": "9c40900909",
"name": "some_name1"
},
{
"id": "pc4b09090"
},
{
"id": "8lknkkn45"
},
{
"id": "9df40900909",
"name": "some_name2"
}
]
Here is how I reduced the list to be just "resources" with attribute "name":
- set_fact:
resources_names: "{{ output.resources | map(attribute='name') }}"
Problem with that is that I get the elements that does not have any value for the name. These are "AnsibleUndefined" in the following output:
- debug:
msg: resources_names list is "{{ resources_names }}"
ok: [localhost] => {
"msg": "resources_names list is \"['some_name1', AnsibleUndefined, AnsibleUndefined, 'some_name2']\""
}
I tried to remove it with reject and regexp but that's not working.
- set_fact:
list2: "{{ resources_names | reject('match', '^$') | list }}"
Same with this one:
- set_fact:
resources_names: "{{ output.resources | map(attribute='name') | rejectattr('name', 'none') }}"
Any idea?
Thanks.
In a nutshell:
---
- hosts: localhost
gather_facts: false
vars:
resources: [
{
"id": "9c40900909",
"name": "some_name1"
},
{
"id": "pc4b09090"
},
{
"id": "8lknkkn45"
},
{
"id": "9df40900909",
"name": "some_name2"
}
]
tasks:
- debug:
msg: "{{ resources | selectattr('name', 'defined') | map(attribute='name') }}"
which gives:
$ ansible-playbook /tmp/play.yml
[WARNING]: No inventory was parsed, only implicit localhost is available
PLAY [localhost] *******************************************************************************************************************************************************************************
TASK [debug] ***********************************************************************************************************************************************************************************
ok: [localhost] => {
"msg": [
"some_name1",
"some_name2"
]
}
PLAY RECAP *************************************************************************************************************************************************************************************
localhost : ok=1 changed=0 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
Note for your next question: please make sure you double check your examples and paste valid json/yaml data in your question. Thanks
Use the filter json_query. This filter ignores missing attributes
resources_names: "{{ output.resources|json_query('[].name') }}"
gives
resources_names:
- some_name1
- some_name2
I need to simultaneously traverse a data structure to align the values, I get this structure with the code below :
- name: Set instances
set_fact:
instance_db:
- 'db2inst1'
- 'db2inst2'
- name: Get Dialog Path
shell: db2 get dbm cfg | grep -i "Current member resolved DIAGPATH" | awk {'print $6'}
become: true
become_method: sudo
become_flags: -i
become_user: "{{ item }}"
loop: "{{ instance_db }}"
register: kud_path
- name: set_fact
set_fact:
db2_store: "[{{ instance_db | list }}] + [{{ kud_path.results|map(attribute='stdout')|list }}]"
vars:
db2_store: []
Result
{
"changed": false,
"ansible_facts": {
"db2_store": [
"db2inst1",
"db2inst2",
[
"/db2home/db2inst1/sqllib/db2dump/DIAG0000/",
"/home/db2inst2/sqllib/db2dump/DIAG0000/"
]
]
},
"_ansible_no_log": false
}
Now I need to automatically traverse these indexes where I put [*]. Because as it is, I can only access the data like this item[0][1]
- name: Creating silent config
template:
src: template.txt.j2
dest: '/tmp/template{{ item[0][*] | lower }}.txt'
mode: '0775'
loop:
- "{{ db2_store }}"
This workaround was necessary to be able to pass both values to the template
Template
################## Database connection config ##################
INSTANCE={{ item[0][*] }}
DIAGLOG_PATH={{ item[1][*] }}db2diag.log
Any suggestions on how to do this or a more elegant way to get the same result?
Assuming the following:
instance_db = ["db2inst1", "db2inst2"]
kud_path.results|map(attribute='stdout')|list = ["/db2home/db2inst1/sqllib/db2dump/DIAG0000/", "/home/db2inst2/sqllib/db2dump/DIAG0000/"]
Then you can use the zip filter which will pair the Nth element of the first list with the Nth element of the second list.
- debug:
msg: "{{ instance_db | zip(kud_path.results|map(attribute='stdout')|list) }}"
Outputs:
TASK [debug] ********************************************************************
ok: [localhost] => {
"msg": [
[
"db2inst1",
"/db2home/db2inst1/sqllib/db2dump/DIAG0000/"
],
[
"db2inst2",
"/home/db2inst2/sqllib/db2dump/DIAG0000/"
]
]
}
This makes it easy for you to loop over:
- name: Creating silent config
template:
src: template.txt.j2
dest: '/tmp/template{{ item[0] | lower }}.txt'
mode: '0775'
loop: "{{ instance_db | zip(kud_path.results|map(attribute='stdout')|list) }}"
################## Database connection config ##################
INSTANCE={{ item[0] }}
DIAGLOG_PATH={{ item[1] }}db2diag.log
Here is my output.
"result.containers": [
{
"Image": "cna.docker.dev/webproxy:1.0.0",
},
{
"Image": "cna-docker-dev-local.docker.dev/lega-customer:1.0.1",
}
]
Here is my code
- name: adding it to groups using images
add_host:
name: "{{ inventory_hostname }}"
groups: '{{ (item.Image.split("/")[1:] | regex_replace("^/", "")).split(":")[0] }}'
with_items: "{{ result.containers }}"
I'm getting the below output:
"add_host": {
"groups": [
"['webproxy"
],
"host_name": "vm4.nodekite.com",
"host_vars": {
}
I would like to filter webproxy from cna.docker.dev/webproxy:1.0.0.
But I'm getting "['webproxy" instead of "webproxy.
Could someone please help!! I even tried this code which fails
(item.Image.split("/")[1:]).split(":")[0]
- debug:
msg: "{{ item.Image | regex_replace('.*?/(.*?):.*', '\\1') }}"
loop: "{{ result.containers | flatten(1) }}"
Example regex is non greedy, adjust as needed.
Gives
ok: [localhost] => (item={'Image': 'cna.docker.dev/webproxy:1.0.0'}) => {
"msg": "webproxy"
}
ok: [localhost] => (item={'Image': 'cna-docker-dev-local.docker.dev/lega-customer:1.0.1'}) => {
"msg": "lega-customer"
}
I want to create multiple aws accounts at once and to save different output list into variables.
I tried with debug msg but maybe is not the proper option.
The idea is the following:
-
name: Create accounts.
hosts: localhost
vars_prompt:
- name: "tag_start"
prompt: "Please set the starting number for the account"
private: no
- name: "tag_end"
prompt: "Please set the ending number for the account"
private: no
tasks:
- name: "Emails"
debug:
msg: "test+{{item}}#gmail.com"
with_sequence: start={{ tag_start }} end={{ tag_end }}
register: email
- name: "Account name"
debug:
msg: "account{{item}}"
with_sequence: start={{ tag_start }} end={{ tag_end }}
register: account_name
- name: "Emails list"
debug:
msg: "{{email}}"
- name: "Account names"
debug:
msg: "{{account_name}}"
- name: Create AWS account
shell: >
aws organizations create-account --email "{{ item[0] }}" \
--account-name "{{ item[1] }}" \
--role-name admin \
--iam-user-access-to-billing ALLOW \
--profile default
with_together:
- "{{ email }}"
- "{{ account_name }}"
The point is that first and second task seems to work as expected showing only what I need like:
TASK [Emails] ********************************************************************************************************************************************************************************************************************************************************************************************************
task path: /Users/me/repos/create_aws_account.yaml:15
ok: [localhost] => (item=46) => {
"msg": "test+46#gmail.com"
}
ok: [localhost] => (item=47) => {
"msg": "test+47#gmail.com"
}
TASK [Account name] **************************************************************************************************************************************************************************************************************************************************************************************************
task path: /Users/me/create_aws_account.yaml:21
ok: [localhost] => (item=46) => {
"msg": "account46"
}
ok: [localhost] => (item=47) => {
"msg": "account47"
}
But when I check the output with another debug msg of the variables saved I get the following:
TASK [Emails] ********************************************************************************************************************************************************************************************************************************************************************************************************
task path: /Users/me/create_aws_account.yaml:27
ok: [localhost] => {
"msg": {
"changed": false,
"msg": "All items completed",
"results": [
{
"_ansible_ignore_errors": null,
"_ansible_item_label": "46",
"_ansible_item_result": true,
"_ansible_no_log": false,
"_ansible_verbose_always": true,
"changed": false,
"failed": false,
"item": "46",
"msg": "test+46#gmail.com"
},
{
"_ansible_ignore_errors": null,
"_ansible_item_label": "47",
"_ansible_item_result": true,
"_ansible_no_log": false,
"_ansible_verbose_always": true,
"changed": false,
"failed": false,
"item": "47",
"msg": "test+47#gmail.com"
}
]
}
}
TASK [Account names] *************************************************************************************************************************************************************************************************************************************************************************************************
task path: /Users/me/create_aws_account.yaml:31
ok: [localhost] => {
"msg": {
"changed": false,
"msg": "All items completed",
"results": [
{
"_ansible_ignore_errors": null,
"_ansible_item_label": "46",
"_ansible_item_result": true,
"_ansible_no_log": false,
"_ansible_verbose_always": true,
"changed": false,
"failed": false,
"item": "46",
"msg": "account46"
},
{
"_ansible_ignore_errors": null,
"_ansible_item_label": "47",
"_ansible_item_result": true,
"_ansible_no_log": false,
"_ansible_verbose_always": true,
"changed": false,
"failed": false,
"item": "47",
"msg": "account47"
}
]
}
}
So what can I do to save only the msg or stdout?
I tried also with set_fact but it seems to does not work either.
An option would be to create lists with set_fact
vars:
email: []
account_name: []
tasks:
- set_fact:
email: "{{ email }} + [ 'test{{ item }}#gmail.com' ]"
account_name: "{{ account_name }} + [ 'account{{ item }}' ]"
with_sequence: start="{{ tag_start }}" end="{{ tag_end }}"
Finally found what I was exactly looking for, and found it here, it is a reply from the user KaffeeKiffer. (The second reply).
Link
At the end is the following:
- set_fact:
emails: "{{ email.results | map(attribute='msg') | list }}"
- set_fact:
account_names: "{{ account_name.results | map(attribute='msg') | list }}"
For completeness sake: | list is required, because map will always return an iterator.