Async "Fire and Forget" Issues - Ansible 1.7.1

Archives Ansible Project
1

Hello,

I am trying to use the Async capabilities of ansible to “fire and forget” a job and then check it later and wait till its end.
I have followed the example mentioned here : http://docs.ansible.com/playbooks_async.html

But I am facing an issue and I don’t really understand what’s happening.

The playbook looks like

`

  • name: ‘MyPlay | Fire and Forget’
    command: “/opt/PROGRAM/bin/gremlin -e /opt/PROGRAM/scripts/load_graph.grm …”
    async: 1000
    poll: 0
    register: load_graph_async

  • name: ‘MyPlay | Check Job’
    async_status: jid={{ load_graph_async.ansible_job_id }}
    register: job_result
    until: job_result.finished
    retries: 30
    `

The playbook logs look like :

`
TASK: [MyPlay | Fire and forget] *************************
<job 759951815367.27671> finished on Host1

TASK: [MyPlay | Check Job] *************************************
fatal: [Host1] => error while evaluating conditional: job_result.finished
`

What I do understand is :
job 759951815367.27671

  • 759951815367 is the JobID
  • 27671 is the PID of the wrapper process

But when I go on the Host after the failed run, I saw

root 27674 0.0 0.0 136548 3672 ? S 21:15 0:00 /usr/bin/python /home/USER/.ansible/tmp/ansible-tmp-1419020105.57-13577669747697/async_wrapper 759951815367 1000 /home/USER/.ansibl root 27675 0.0 0.0 136548 3900 ? S 21:15 0:00 \_ /usr/bin/python /home/USER/.ansible/tmp/ansible-tmp-1419020105.57-13577669747697/async_wrapper 759951815367 1000 /home/USER/.an root 27676 0.1 0.0 138008 7320 ? S 21:15 0:00 \_ /usr/bin/python /home/USER/.ansible/tmp/ansible-tmp-1419020105.57-13577669747697/command /home/USER/.ansible/tmp/ansible-tm root 27677 0.0 0.0 115504 1544 ? S 21:15 0:00 \_ /bin/bash /opt/PROGRAM/bin/gremlin -e /opt/PROGRAM/scripts/load_graph.grm ... root 27727 84.0 14.4 4875920 1187256 ? Sl 21:15 0:21 \_ /usr/lib/jvm/java-7-oracle/bin/java -server -Xms4g -Xmx4g -cp /opt/PROGRAM/conf:/opt/PROGRAM

So any ideas ?

Thanks a lot for your help every one :slight_smile:

2

Hello,

I am trying to use the Async capabilities of ansible to "fire and forget" a
job and then check it later and wait till its end.
I have followed the example mentioned here :
http://docs.ansible.com/playbooks_async.html

Async "fire and forget" was a new feature in Ansible 1.8 so you'll
need to update to get that functionality.

-AdamM

3

Well, it wasn’t a new feature in 1.8, though there was a bug fixed in the fire and forget operation when you don’t specify a lifetime of the process, IIRC - I remember you filing it, all the same :slight_smile:

Let us know if you have the same problem in 1.8, regardless.

4

Ok, thanks all for your responses.

I will upgrade to Ansible 1.8. and be back to you.

5

No luck … it still does not work

$ pip install Ansible $ pip list ansible (1.8.2) ecdsa (0.11) Jinja2 (2.7.3) MarkupSafe (0.23) paramiko (1.15.2) pip (1.5.6) pycrypto (2.6.1) PyYAML (3.11) setuptools (3.6) wsgiref (0.1.2)

`
$ ansible-playbook -i inventories/demo playbooks/run/load-graph.yml

PLAY [VNG-Bootstrap | create-graph] *******************************************

GATHERING FACTS ***************************************************************
ok: [Host1]
Host1: importing …/…/…/vars/demo.yml

TASK: [Creating Titan graph schema] ***************************
changed: [Host1]

TASK: [Loading data into Titan graph] *************************
<job 356642401411.12344> finished on Host1

TASK: [Check Loading data Status] *****************************
fatal: [Host1] => error while evaluating conditional: load_graph_result.finished

FATAL: all hosts have already failed – aborting

PLAY RECAP ********************************************************************
to retry, use: --limit @/Users/USER/create-load-graph.retry

Host1 : ok=10 changed=2 unreachable=1 failed=0
`

Playbook changes

`
diff --git a/deployment/ansible/playbooks/run/create-load-graph.yml b/deployment/ansible/‘playbooks/run/create-load-graph.yml’
index 86a4fac…246f820 100644
— a/deployment/ansible/playbooks/run/create-load-graph.yml
+++ b/deployment/ansible/playbooks/run/create-load-graph.yml
@@ -38,8 +38,15 @@

  • async: 14400
  • poll: 60
  • async: 1000
  • poll: 0
  • register: load_graph
6

Hello,

Gist is here it some body wants to give a look : https://gist.github.com/xakraz/2ec7086d04630b833d23

Thanks again

7

Hello,

Gist is here it some body wants to give a look :
https://gist.github.com/xakraz/2ec7086d04630b833d23

I think the problem is here:
    https://gist.github.com/xakraz/2ec7086d04630b833d23#file-create-load-graph-yml-L46

Where the variable is 'jiid':

    async_status: jiid={{ load_graph.ansible_job_id }}

But should be 'jid'

    async_status: jid={{ load_graph.ansible_job_id }}

Hopefully that works.

-AdamM

8

Shame on me … I was spitting against ansible whereas I made a dummy typo in the YAML playbook file.

Thanks Adam.

But by the way, what is the cleanest manner to

  • launch a long running task with Ansible
  • check it is still running
  • display progress

Do I have to use the async / poll statement and adjust their values to have a limit and fail if it was too long to run ?

Thanks again for your help !