Shutdown more gracefully #633
Comments
|
Part of this is based on some of my work developing an event source for RabbitMQ, where we have work that is placed in a queue, where each message is a piece of work that needs to be done. Ideally, if we do an upgrade or make a change to the system, we need a way to have Obviously there's a chance that I have just not written the event source plugin for RabbitMQ well enough to handle this, but I coded it while following the Kafka and Amazon SQS examples of event sources and it appears that those event sources would suffer from the same issue. |
|
Here's an example log: |
|
Here's a log of the webhook rule matching to run the shutdown action, and then another event being received from RabbitMQ that matches another rule in the ruleset and starts a playbook run after the shutdown action has been called. |
|
Rulebook: ---
- name: Rulebook for webhook
hosts: all
sources:
- ansible.eda.rabbitmq:
host: localhost
port: 5672
queue: openstack-tsf-network-controller
user: guest
password: guest
routing_key: 'platform.openstack.*.tsf.rendering.network.*'
exchange: "platform_requests"
- ansible.eda.webhook:
port: 5000
rules:
- name: "tsf network create rabbitmq event"
condition: event.body.event == 'createOpenStackNetwork'
actions:
- print_event:
pretty: true
- run_playbook:
name: controller/create_tsf_tenant_network.yml
verbosity: 3
- name: "shutdown via webhook"
condition: event.payload.action == "shutdown"
action:
shutdown:
delay: 15
message: "Shutting down" |
sc68cal
added a commit
to sc68cal/ansible-rulebook
that referenced
this issue
Dec 28, 2023
This is an attempt to address ansible#633 but it's very ham-fisted.
sc68cal
added a commit
to sc68cal/ansible-rulebook
that referenced
this issue
Dec 28, 2023
This is an attempt to address ansible#633 but it's very ham-fisted.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Please confirm the following
Feature type
Enhancement to Existing Feature
Feature Summary
Currently, there doesn't seem to be a way to get
ansible-rulebookto stop processing new events from event sources and allow all existing work to complete. Even using theshutdownaction and a webhook, other rules will continue to pull work from the even sources and start kicking off playbook runs. This means that there's a high likelihood thatansible-rulebookwill exit while a playbook is still executing.There needs to be a way to have
ansible-rulebookstop all the event sources, allow all work to complete, and then exit.This would also be useful when handling SIGTERM signals. Allow the work to complete, then exit, instead of just exiting hard with work still in progress.
Steps to reproduce
Create a rulebook that contains an event source that is a queue, like the Kafka event source or Amazon SQS, and also has a webhook source that has a rule that matches the webhook and does a shutdown action.
Put an event into the Kafa or SQS queue, watch the playbook start running. Send an event to the webhook
Current results
Work continues up until the shutdown happens, with no real "graceful" behavior
Sugested feature result
Ability to stop
ansible-rulebookcleanly, where all work that has been pulled from an event source has completed, while stopping pulling from an event source and allow work to queue up in event sources where this construct exists (SQS queue, etc).Additional information
No response
The text was updated successfully, but these errors were encountered: