ci-retry

重试失败的 CI 管道。兼容 GitLab CI 以及 Konflux/Tekton。适用于用户输入“重试管道”、“重试 CI”、“重新运行管道”或“重试 MR !1234”时使用。

SKILL.md

--- frontmatter

name: ci-retry
description: Retry a failed CI pipeline. Works with GitLab CI and Konflux/Tekton. Use when user says "retry pipeline", "retry CI", "rerun pipeline", or "retry MR !1234".

CI Retry

Retry failed GitLab CI or Konflux/Tekton pipelines.

Inputs

Input	Type	Default	Purpose
`mr_id`	int	-	GitLab MR ID — retries its pipeline
`pipeline_id`	int	-	GitLab CI pipeline ID
`tekton_run`	string	-	Tekton PipelineRun name
`project`	string	automation-analytics/automation-analytics-backend	GitLab project
`namespace`	string	aap-aa-tenant	Konflux/Tekton namespace
`wait`	bool	false	Wait for pipeline to complete (up to 15 min)

Workflow

1. Determine CI System

•If tekton_run → Tekton
•If pipeline_id → GitLab
•If mr_id → GitLab (get pipeline from MR)

2. GitLab Path

•persona_load("developer")
•If mr_id: gitlab_ci_status(project="{{ project }}", mr_id="{{ mr_id }}") — extract pipeline_id
•gitlab_ci_status(project="{{ project }}", pipeline_id="{{ pipeline_id }}") — get failed jobs
•gitlab_ci_retry(project="{{ project }}", job_id="{{ failed_job_id }}")
•If wait: poll gitlab_ci_status until complete

3. Tekton Path

•persona_load("release")
•tkn_pipelinerun_describe(run_name="{{ tekton_run }}", namespace="{{ namespace }}")
•tkn_pipelinerun_logs(run_name="{{ tekton_run }}", namespace="{{ namespace }}", all_tasks=true) — if failed
•Re-trigger: tkn_pipeline_start or equivalent (depends on tool availability)

4. Error Recovery

•GitLab "no such host" → vpn_connect()
•GitLab "unauthorized" → check GitLab token
•Tekton "unauthorized" → kube_login("konflux")
•Tekton "pipelinerun not found" → list with tkn_pipelinerun_list() for correct name

5. Memory

•memory_session_log("Retried CI pipeline", "target={{ target }}, project={{ project }}")

Output

Report: system (gitlab/tekton), target, status, retry status, and monitor commands.