Create a documenation how to debug the agent

[AvielSegev]

Hi Everyone,

I’ve created a document called debugging.md as a follow-up to ECOPROJECT-2656.

I'm pretty sure there's plenty of room for new ideas, improvement and changes in the document. So let me know your thoughts!

[app-sre-bot]

Can one of the admins verify this patch?

[machacekondra]

Looks great! I would maybe not use the Agent everywhere, can we use Discovery VM or Discover Machine in most of the places? Unless you are mentioning the planner-agent container itself

[AvielSegev]

Looks great! I would maybe not use the Agent everywhere, can we use Discovery VM or Discover Machine in most of the places? Unless you are mentioning the planner-agent container itself

Sure. It's Done:)

[AvielSegev]

Should we mention additional cases or take a different approach to those already presented?

[machacekondra]

Should we mention additional cases or take a different approach to those already presented?

What do you mean by additional cases?

[AvielSegev]

Should we mention additional cases or take a different approach to those already presented?

What do you mean by additional cases?

The document presents two edge cases from the customer side. Are there any other potential issues, aside from the disconnected environment and container-related problems?

[machacekondra]

The document presents two edge cases from the customer side. Are there any other potential issues, aside from the disconnected environment and container-related problems?

Most likely are network issues, and issues with containers to start. I didn't encounter any other problem, yet. Maybe issues logging into Vcenter.

[AvielSegev]

The document presents two edge cases from the customer side. Are there any other potential issues, aside from the disconnected environment and container-related problems?

Most likely are network issues, and issues with containers to start. I didn't encounter any other problem, yet. Maybe issues logging into Vcenter.

Logging into vCenter is handled smoothly by the UI. If the username or password is incorrect, the user is notified. If the /sdk suffix is missing, it is automatically appended. Can you think about additional specific cases regarding the login?

[machacekondra]

Can you please use systemctl instead of podman. For logs it's fine, but start & stop could be problematic, as we have pre/post script defined in systemd services

[AvielSegev]

Could you please provide the full command using systemctl? It is not working for me on the planner-agent container as a service.
but I view the ignition file and noticed a retry every 5 seconds to restart in case of a failure. Perhaps we can remove it?

[machacekondra]

Should be:

journaltct --user -u planner-agent

systemctl --user status planner-agent
systemctl --user restart planner-agent

but I view the ignition file and noticed a retry every 5 seconds to restart in case of a failure. Perhaps we can remove it?

Why do you think we should remove it? Actually, it's there to make sure if there is something wrong with startup, we retry it, do you think we should not try to retry?

[AvielSegev]

No, the ignition is fine. I meant to remove the troubleshooting step where the user manually restarts the container, as we've already defined a restart policy, making manual involvment redundant.

[AvielSegev]

Thanks Ondra. I changed the commands.

[machacekondra]

Great work! Thanks!