satellite-peers
Use when investigating, debugging or monitoring a specific production Storj satellite peer (api, repair, core, ranged-loop, auditor, gc-bf, gc-sender, admin, console, change-stream, jobq). Gives the peer card - what runs inside it, the verified PromQL health signals, silent-failure paths in the code, key config flags, and known traps.
ソース情報
- リポジトリ
- storj/claude-plugins
- ソースの最終更新活動
- 2026年10月9日 13:02
- 検出された SKILL.md の言語
- 英語
- スター
- 3
- フォーク
- 0
インストール方法
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
ソースファイルを確認
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
ファイルエクスプローラー
5 ファイルSKILL.md を表示中
SKILL.md
ソースの指示 · 読み取り専用プレビュー- name
- satellite-peers
- description
- Use when investigating, debugging or monitoring a specific production Storj satellite peer (api, repair, core, ranged-loop, auditor, gc-bf, gc-sender, admin, console, change-stream, jobq). Gives the peer card - what runs inside it, the verified PromQL health signals, silent-failure paths in the code, key config flags, and known traps.
# Satellite peer cards
Each prod satellite process is one `satellite <sub>` subcommand of the modular ("mud") build:
code entry `satellite/run/<sub>.go` in storj/storj, k8s `app="satellite-<sub>"`.
Read **only** the card(s) you need:
| Peer | app label | Card |
|---|---|---|
| api | `satellite-api` | [peers/api.md](peers/api.md) |
| repair | `satellite-repair`, `*-repair` | [peers/repair.md](peers/repair.md) |
| ranged-loop | `satellite-ranged-loop` | [peers/ranged-loop.md](peers/ranged-loop.md) |
| gc-bf (+ gc-sender handoff) | `satellite-gc-bf` | [peers/gc-bf.md](peers/gc-bf.md) |
| core, auditor, gc-sender, admin, console, change-stream, jobq | `satellite-<name>` | not written yet — build one (below) |
Always also load `satellite-observability` (datasource, selectors, traps) and, for config, `satellite-infra`.
## When there is no card
Build the facts yourself, in this order, and keep notes in the card format below so the user can
turn them into a card:
1. `satellite/run/<sub>.go` → `GetSelector` → resolve each `mud.Select[*T]` to its package (`*/mud.go`).
2. `rg -n 'mon\.(Meter|Counter|IntVal|FloatVal|Event|Task)' <packages>` for metric names; check which exist:
`count by (__name__) ({app="satellite-<sub>",kubernetes_namespace="satellite",environment_name="storj-prod-satellite-us1",scope=~"storj_io_storj_satellite_<pkg>.*"})`.
3. Look for `return nil` after `log.Error`/`log.Warn` in the main loops — those are silent failures.
4. Config: the `Config` struct `help:` tags + `rg` in infra values.
## Card format
1. Identity — subcommand, app labels, regions, replicas.
2. What runs inside — components, loop shape.
3. Key signals — table of question → verified PromQL; a dated baseline.
4. Silent failures — where errors are swallowed or not errors at all.
5. Distinctive logs — logger name + message (from code).
6. Config that matters — flags, defaults, where prod values live.
7. Traps / where to look next.
Every query in a card must have returned data in prod at least once. Put the verification date at the top.
GitHubで見る