[dejanzele]

The Lookout API returns the pod name of each run as run.podName, and the Query API returns it as JobRunDetails.pod_name, after pod_name joins the column lists of the Query API queries. The name is empty while the run has no pod.

Lookout does not back-fill the runs that it stored before it stored pod names, because a backfill of job_run takes many hours on a large deployment. Every pod of those runs has the name armada-<jobId>-0. So both APIs return that name for a run with no stored name and a pending time, a start time or an exit code. The ingester sets these only from the events of a pod, so a run without any of them gets no name. The executor can report a fast pod first as running or finished, so the pending time alone is not enough. The fallback is in one function, lookout.RunPodName, which both APIs call. It can go when the Lookout pruner has removed the runs from before #5279.

[mergify]

Merge Protections

🔴 1 of 1 protections blocking · waiting on ⛓️ dependency

	Protection	Waiting on
🔴	⛓️ Depends-On Requirements	⛓️ dependency

🔴 ⛓️ Depends-On Requirements

Waiting for

• depends-on=armadaproject/armada#5279 [⛓️ Add a column for the pod name of each run in Lookout #5279 [stack]]

This rule is failing.

Requirement based on the presence of `Depends-On` in the body of the pull request

• depends-on=armadaproject/armada#5279 [⛓️ Add a column for the pod name of each run in Lookout #5279 [stack]]

[greptile-apps]

Confidence Score: 5/5

[Medium impact] This PR appears safe to merge; the previous missing-name issue is fixed.

Summary

This PR returns each run’s pod name through the Lookout API and Query API.

• Lookout and the Query API return each run’s pod name.

_{Reviews (5) · Last reviewed commit: "Return the pod name of each run from the..." · Reviewed by Greptile}

[mergify]

⚠️ The sha of the head commit of this PR conflicts with #5279. Mergify cannot evaluate rules on this PR. Once #5279 is merged or closed, Mergify will resume processing this PR. ⚠️