Security
Choria integrates Puppet Tasks tightly with its AAA (Authentication, Authorization and Auditing) suite, this section will introduce you to the features of each.
Authorization
Any Choria authorization plugin will be supported by the Tasks integration. Of course Choria configures only the Action Policy so the documentation will focus on that.
The agent have a number of actions, you generally will give people access to all of these:
Action | Description |
---|---|
download | Downloads a Puppet Task into a local cache |
run_and_wait | Runs a Puppet Task that was previously downloaded, wait for it to finish |
run_no_wait | Runs a Puppet Task that was previously downloaded do not wait for it to finish |
task_status | Request the status of a previously run task |
If you give someone access to download, run_and_wait and run_no_wait they can initiate and run tasks, you can give someone access to task_status only to view statuses.
Note that the data plugin effectively shows all the task_status action shows and there are no RBAC for those. So basically task_status is always open.
So the example from earlier in this document gives choria=rip.mcollective full access to the Puppet Task feature:
mcollective_agent_bolt_tasks::policies:
- action: "allow"
callers: "choria=rip.mcollective"
actions: "download run_and_wait run_no_wait task_status"
facts: "*"
classes: "*"
Once you have this in place you have to authorize specific tasks:
mcollective_agent_bolt_tasks::policies:
- action: "allow"
callers: "choria=rip.mcollective"
actions: "puppet_conf gcompute::snapshot"
facts: "*"
classes: "*"
Here this user will have the ability to run the puppet_conf task as well as the gcompute::snapshot one but no others.
Every task invocation will call the RBAC system twice - one for the run_and_wait or run_no_wait action and once for an action matching the task name.
Auditing
Auditing is done as per normal Choria Audit Logs and are written in JSON format. For every run there would be 2 actions performed - one of the run variants and one for the actual task name being invoked.
$ sudo tail -n 1000 /var/log/puppetlabs/mcollective-audit.log|jq 'select(.request_id=="45250c07824f5922be68468d08f6b76c")'
{
"timestamp": "2018-03-19T13:50:51.762983+0000",
"request_id": "45250c07824f5922be68468d08f6b76c",
"request_time": 1521467451,
"caller": "choria=rip.mcollective",
"sender": "dev1.devco.net",
"agent": "bolt_task",
"action": "run_and_wait",
"data": {
"task": "puppet_conf",
"files": "[{\"filename\":\"init.rb\",\"sha256\":\"38351bb4b72a29064f5c9a224a81f1abf7042b0bc9d1ffd2a074d1bd63b2f246\",\"size_bytes\":1231,\"uri\":{\"path\":\"/puppet/v3/file_content/tasks/puppet_conf/init.rb\",\"params\":{\"environment\":\"production\"}}}]",
"input": "{\"action\":\"set\",\"section\":\"user\",\"setting\":\"modulepath\",\"value\":\"/tmp/modules\"}",
"process_results": true
}
}
{
"timestamp": "2018-03-19T13:50:51.763933+0000",
"request_id": "45250c07824f5922be68468d08f6b76c",
"request_time": 1521467451,
"caller": "choria=rip.mcollective",
"sender": "dev1.devco.net",
"agent": "bolt_task",
"action": "puppet_conf",
"data": {
"task": "puppet_conf",
"files": "[{\"filename\":\"init.rb\",\"sha256\":\"38351bb4b72a29064f5c9a224a81f1abf7042b0bc9d1ffd2a074d1bd63b2f246\",\"size_bytes\":1231,\"uri\":{\"path\":\"/puppet/v3/file_content/tasks/puppet_conf/init.rb\",\"params\":{\"environment\":\"production\"}}}]",
"input": "{\"action\":\"set\",\"section\":\"user\",\"setting\":\"modulepath\",\"value\":\"/tmp/modules\"}",
"process_results": true
}
}
Above you can see the double RBAC in action - once for action run_and_wait and once for puppet_conf.
You might of course find additional status calls etc too:
$ sudo tail -n 1000 /var/log/puppetlabs/mcollective-audit.log|jq 'select(.data.task_id=="45250c07824f5922be68468d08f6b76c")'
{
"timestamp": "2018-03-19T13:51:41.232263+0000",
"request_id": "3beaa976857f527281ac192602acad83",
"request_time": 1521467501,
"caller": "choria=rip.mcollective",
"sender": "dev1.devco.net",
"agent": "bolt_task",
"action": "task_status",
"data": {
"task_id": "45250c07824f5922be68468d08f6b76c",
"process_results": true
}
}