[Fleet] Add a new keep_monitoring_alive flag to kibana.yml
#168629
Comments
kpollich
added
the
Team:Fleet
|
Pinging @elastic/fleet (Team:Fleet) |
|
I think we should just change the agent configuration used in cloud instead of doing this in Kibana. See elastic/elastic-agent#2734 (comment) What we really want on the Fleet side of this is to expose all of the |
|
changing agent cloud config is not that simple as it seems as logic there which translate what is defined in a template and what gets to agent does not have a notion of flags we use. the thing @kpollich mentioned will result in what you are requesting. using additional flag they will generate config we expect. we will not see additional flag in our config and our logic stays the same as defined in elastic/elastic-agent#3584 |
|
Thanks, we can't change the default behaviour of elastic/elastic-agent#3584 or elastic/elastic-agent#2509 without the change here to ensure cloud keeps working as expected. I'll mark this as blocking those two issues and we can figure out when to prioritize this. |
|
This + the necessary change in the preconfigured cloud policy would be pretty low effort for the Fleet & Integrations team to implement. We could turn around pretty quickly on this to unblock if needed. |
|
@kpollich just pulled it in the current sprint to make it happened. Happy to have you working on it or ask someone else, just let me know. |
|
I can get this done late this week or early next. |
## Summary Ref #168629 Adds a new `keep_monitoring_alive` flag to agent policies that allows agent environments to keep the monitoring server if they choose. > [!note] > `monitoring_enabled: []` needs to be explicitly set for this flag to be honored. If `monitoring_enabled` is omitted entirely, Fleet defaults to enabling monitoring for logs and metrics. ## To test Create a preconfigured policy where logs/metrics monitoring is explicitly disabled and include this new flag e.g. ```yml xpack.fleet.agentPolicies: - name: Test preconfigured policy id: test-preconfigured-policy package_policies: [] monitoring_enabled: [] keep_monitoring_alive: true ``` Then, run Kibana and verify the `monitoring` block in the full agent policy has monitoring enabled, but logs/metrics as disabled, e.g.  You can also verify the need for `monitoring_enabled: []` by creating another preconfigured policy e.g. ```yml - name: Test preconfigured policy 2 id: test-preconfigured-policy-2 package_policies: [] keep_monitoring_alive: true ``` This will generate a policy e.g.  All other combinations of enabling monitoring for logs/metrics should also generate the policies as expected. If you want to verify every combination manually: ```yml xpack.fleet.agentPolicies: - name: Test preconfigured policy id: test-preconfigured-policy package_policies: [] monitoring_enabled: [] keep_monitoring_alive: true - name: Test preconfigured policy 2 id: test-preconfigured-policy-2 package_policies: [] keep_monitoring_alive: true - name: Test preconfigured policy 3 id: test-preconfigured-policy-3 package_policies: [] monitoring_enabled: ['logs'] - name: Test preconfigured policy 4 id: test-preconfigured-policy-4 package_policies: [] monitoring_enabled: ['metrics'] - name: Test preconfigured policy 5 id: test-preconfigured-policy-5 package_policies: [] monitoring_enabled: ['logs', 'metrics'] ``` --------- Co-authored-by: kibanamachine <[email protected]>
…#168865) ## Summary Ref elastic#168629 Adds a new `keep_monitoring_alive` flag to agent policies that allows agent environments to keep the monitoring server if they choose. > [!note] > `monitoring_enabled: []` needs to be explicitly set for this flag to be honored. If `monitoring_enabled` is omitted entirely, Fleet defaults to enabling monitoring for logs and metrics. ## To test Create a preconfigured policy where logs/metrics monitoring is explicitly disabled and include this new flag e.g. ```yml xpack.fleet.agentPolicies: - name: Test preconfigured policy id: test-preconfigured-policy package_policies: [] monitoring_enabled: [] keep_monitoring_alive: true ``` Then, run Kibana and verify the `monitoring` block in the full agent policy has monitoring enabled, but logs/metrics as disabled, e.g.  You can also verify the need for `monitoring_enabled: []` by creating another preconfigured policy e.g. ```yml - name: Test preconfigured policy 2 id: test-preconfigured-policy-2 package_policies: [] keep_monitoring_alive: true ``` This will generate a policy e.g.  All other combinations of enabling monitoring for logs/metrics should also generate the policies as expected. If you want to verify every combination manually: ```yml xpack.fleet.agentPolicies: - name: Test preconfigured policy id: test-preconfigured-policy package_policies: [] monitoring_enabled: [] keep_monitoring_alive: true - name: Test preconfigured policy 2 id: test-preconfigured-policy-2 package_policies: [] keep_monitoring_alive: true - name: Test preconfigured policy 3 id: test-preconfigured-policy-3 package_policies: [] monitoring_enabled: ['logs'] - name: Test preconfigured policy 4 id: test-preconfigured-policy-4 package_policies: [] monitoring_enabled: ['metrics'] - name: Test preconfigured policy 5 id: test-preconfigured-policy-5 package_policies: [] monitoring_enabled: ['logs', 'metrics'] ``` --------- Co-authored-by: kibanamachine <[email protected]>
|
@kpollich can we close? |
|
Yep thanks! |
Ref elastic/elastic-agent#2734
The agent team is shipping some work soon to automatically shut down the monitoring server if metrics are disabled. However, in cloud we want to enable to keep the monitoring server alive even if nothing's being collected. To support this use case, we need to add a new flag to
kibana.ymlso that Cloud policies can opt into this behavior.keep_monitoring_aliveflag to Fleet's config schema for preconfigured agent policies here{ enabled : true, logs: false, metrics: false }whenkeep_monitoring_aliveis true on the policy objecttrueThe text was updated successfully, but these errors were encountered: