[Da Huang (git286)]

Each ateapi replica reports the whole fleet on ate.workerpool.workers, and the default install runs two, so a sum across replicas counts every worker twice. The registry told readers they could add the counts together. Say that a sum across replicas double counts, and to take the max across replicas by pool namespace, pool, state and sandbox class before adding.

To mitigate #3055

• Tests pass
• Docs updated

[Jeff Luo (JeffLuoo)]

Thanks for this mitigation. I checked the claim against RegisterWorkerCount and the ate-api-server manifest, and it is correct. I have one comment on the table row in observability.md.

[Jeff Luo (JeffLuoo)]

Please name the grouping here, as the registry note does. "Take the max across replicas before you sum" does not say which labels to keep. A literal sum(max(...)) returns the largest single series, not the fleet. Suggested text: "Every ateapi replica reports the whole fleet. Take the max by pool namespace, pool, state, and sandbox class first, then sum." An example query helps more than prose:

sum by ("ate.workerpool.namespace", "ate.workerpool.name") (
  max by ("ate.workerpool.namespace", "ate.workerpool.name", "ate.worker.state", "ate.sandbox.class") (
    {"ate.workerpool.workers"}
  )
)

[Da Huang (git286)]

Good catch, thanks. Fixed in 278ba7d. The row now names the grouping, and I added your query in a note below the table (a code block cannot go in a table cell). I also added a line to keep cluster in both groupings when one backend gets several clusters, otherwise same-named pools merge. I ran the query on a GKE cluster: it gives 3 for a 3-worker pool, while the plain sum gives 6.