Shepherd and churn efficiency: five fixes from the 2026-11-01 data

Cerrado
  • ¿Algo no está bien?
#7,455 0 comentarios 0 reacciones 0 asignados Ver en GitHub

Los mantenedores suelen responder en 1 día

Nadie ha tomado este issue todavía.

Evaluación

Dificultad
5/5
Tiempo estimado
Más de una semana
Aptitud para principiantes
8/100
Tipo de issue
Refactorización
Claridad
Bastante claro
Estado de actividad
Activo
Stack tecnológico
github-actions, python
Área
ci-cd, devops

Línea de trabajo

The payload names the CI shepherd (reconcile) workflow, the final gate step that makes cancelled runs read as FAILURE, the shepherd:merge approval logic on merge-only updates, and the promoter-bot plan holds for data files; it does not name file paths, so locate them first. Each of the five items should land as its own PR referencing this issue, and the work is done when the cancelled-run rate and the holds data move as the weekly conflict job (#7444) would show. Start with the reconcile trigger and the gate step, since they are the two causes with the most evidence in the payload.

Escrito por el modelo de indexación a partir del texto del issue.

Descripción

enhancement priority: medium scope:infra

Five efficiency fixes, found from one day's data on 2026-11-01. Each lands as its own PR "Part of" this issue.

  1. Calm the shepherd's reconcile runs. 319 of the 500 most recent cancelled runs were CI shepherd (reconcile), not real CI. It fires when a workflow finishes and newer runs cancel older ones, which likely explains most of the 22-28% cancelled rate on the dashboard.
  2. Stop cancelled runs showing as failed. A cancelled run's final gate step exits with an error, so GitHub marks it FAILURE. Workers were sent to fix these: 22 of 41 env fix runs from 10-29 to 10-31 found only cancelled checks. #7432 fixed how ci_digest reads them; this fixes the source.
  3. Keep approval on a merge-only update. A PR that only takes main loses shepherd:merge, goes to draft and re-runs CI, as #7441 did twice. If the patch is identical, skip that. Any real code change still needs re-approval. Owner approved starting this 2026-11-01.
  4. Stop holds on data files. Data files were 54 of the week's 75 file-level plan holds. Regenerated files should not hold at all, and same-estate slices should be scheduled apart.
  5. Break repeat fix loops. #7389 went through ten rounds of one blocked fix. After two rounds with the same cause, hold the PR and name the blocker.

Items 4 and 5 are in churn (promoter-bot). The weekly conflict job (#7444) tracks the effect.

Lenguaje dominante
Python
Estrellas
57
Forks
11
Merge medio
1 h 21 min
PR fusionados (30 d)
971

Preparar el entorno

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de squid-protocol/gitgalaxy

Todos los issues de squid-protocol/gitgalaxy

Issues similares

Más issues de Python

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.