[RFE] The Neutron agents should report the sync process status

[Rodolfo Alonso]

The goal of this RFE is to provide information about the sync status of the Neutron agents. Most of the agents during the initialization transient state or during the RPC recovery, execute a full resync process to retrieve all the information needed to start working.

The resync process time could differ depending on the agent load, RPC channel or host load. The Neutron server, installer tools or the system administrator/user should be able to retrieve from the agent if the "full resync" is in progress or finished, in order to know if the agent is already in the normal processing loop (can attend to new events) or not.

[Lajos Katona (lajos-katona)]

Related mail thread for l3-agent:
https://lists.openstack.org/pipermail/openstack-discuss/2023-March/032670.html

And a related patch for l3-agent:
https://review.opendev.org/c/openstack/neutron/+/878248

[Christian Rohmann (christian-rohmann)]

Conceptually all "services" or "agents" (neutron or others) have the same issue - how to report to the control plane that they have fully synced.

Looking at

* compute service list
* network agent list
* volume service list
* ...

Maybe extending the "State" field go to "syncing" or "starting" prior to reporting "up" would be an approach?

In the end it's more about determining the ready state on the machine the agent/service run on, but also to have this info quickly available in the control plane (to coordinate things).

[Rodolfo Alonso (rodolfo-alonso-hernandez)]

That will be discussed (at least for the networking services) during the PTG next week. The goal is to have a flexible way to report the status, also multiple status (starting+syncing, restarted+stale, etc). We should define a set of status "standard" for any agent and also provide the flexibility to allow specific states, defined only per agent type.

[Christian Rohmann (christian-rohmann)]

Rodolfo ... May I ask if there are any (documented) results out of your discussion at the PTG?
Having a simple way to query the status of an agent would be awesome as monitoring source.

[Rodolfo Alonso (rodolfo-alonso-hernandez)]

Hello Christian:

The RFE was approved during the PTG. It needs a spec before starting the coding process. However I didn't have time to start it. If you are willing to push the spec and code it, you are more than welcome to do it.

Regards.

[OpenStack Infra (hudson-openstack)]

Change abandoned by "Slawek Kaplonski <email address hidden>" on branch: master
Review: https://review.opendev.org/c/openstack/neutron/+/878248
Reason: This review is > 4 weeks without comment, and failed Zuul jobs the last time it was checked. We are abandoning this for now. Feel free to reactivate the review by pressing the restore button and leaving a 'recheck' comment to get fresh test results.