source-command-payload-job-deaggregate

# source-command-payload-job-deaggregate Use this skill when the user asks to run the migrated source command `payload-job-deaggregate`. ## Command Template # Deaggregate Release Payload Job You are helping the user convert an aggregated release payload job back to a regular single-run job. ## Command Arguments This command accepts optional arguments: `/payload-job-deaggregate [release] [stream] [arch]` - If arguments are provided (e.g., `/payload-job-deaggregate 4.21 nightly` or `/payload-job-deaggregate 4.21 nightly arm64`), use them directly - If no arguments are provided, prompt the user for the information ## Steps to follow: 1. **Get release, stream, and architecture**: - If ARGUMENTS are provided, parse them to extract release version, stream type, and optional architecture - If no ARGUMENTS are provided, ask the user for: - Release version (e.g., "4.21", "4.20") - Stream type (e.g., "nightly", "ci", "stable") - Architecture (optional, defaults to amd64 if not provided) - Supported architectures: `arm64`, `s390x`, `ppc64le`, `multi`, or leave blank for `amd64` (default) Construct the file path based on stream and architecture: - For `nightly` stream: - Default amd64: `core-services/release-controller/_releases/release-ocp-{version}.json` - Specific arch: `core-services/release-controller/_releases/release-ocp-{version}-{arch}.json` - For other streams (ci, stable, etc.): - Default amd64: `core-services/release-controller/_releases/release-ocp-{version}-{stream}.json` - Specific arch: `core-services/release-controller/_releases/release-ocp-{version}-{stream}-{arch}.json` Note: When architecture is specified and is not amd64, append `-{arch}` to the filename (after stream if present) 2. **Read the release file**: Load the JSON file to find all jobs in the `verify` section. 3. **Find aggregated jobs**: Identify all jobs that have `"aggregatedProwJob"` in their configuration. These are aggregated jobs. 4. **Present job selection**: Display a numbered list of all aggregated jobs found in the release file, showing both the job key, the Prow job name, and the current `analysisJobCount`. Organize jobs into logical categories (no more than 7 groups) based on platform or distinguishing characteristics. Maintain continuous numbering across all categories. Ask the user to provide the number of the job they want to deaggregate. 5. **Deaggregate the job**: Edit the selected job's configuration to: - Remove the `"aggregated-"` prefix from the job key name (e.g., `"aggregated-gcp-ovn-upgrade"` becomes `"gcp-ovn-upgrade"`) - Remove the entire `"aggregatedProwJob": { "analysisJobCount": N },` line (be careful with trailing commas) - Ensure blocking jobs (those without `"optional": true`) have `"maxRetries"` set (typically 2) - Keep all other existing fields like `"optional"`, `"upgrade"`, `"upgradeFromRelease"`, etc. - Maintain proper JSON formatting 6. **Run make release-controllers**: Create a Python virtual environment if needed and execute `make release-controllers` to regenerate the release controller configurations. First check if `venv` directory exists - if not, create it with `python3 -m venv venv`. Then activate it with `source venv/bin/activate`, and finally run `make release-controllers`. 7. **Confirm completion**: Let the user know the job has been converted back to a regular single-run job and the configurations have been regenerated. ## Important notes: - **Only modify files in `core-services/release-controller/_releases/` - DO NOT modify files in the `priv` directory as those are auto-generated** - Deaggregating requires removing the `"aggregated-"` prefix from the job key name - Deaggregating removes the `aggregatedProwJob` configuration, returning the job to single-run mode - Blocking jobs (without `"optional": true`) must have `"maxRetries"` configured (typically 2) - All other fields (blocking/informing status, upgrade configuration) are preserved - The job will now run once per payload instead of multiple times with statistical analysis - The JSON formatting must remain valid - be careful with commas - Handle different stream types correctly (some releases don't have a stream suffix in the filename)