fix(admin): env filter and honest count on DLQ replay (alt to #76)

[ravidsrk]

This PR is an alternative implementation of #56, opened alongside #76 (thanks @soufian3hm).

POST /admin/api/dlq/{job_id}/replay gains an optional ?environment=<slug> filter (404 on an unknown slug before any row moves, mirroring the CLI --env), and the response reports the real moved count instead of a hardcoded 1 — dlq::replay returns the count and the CLI is adapted. Three failing-first HTTP tests over testcontainers cover it (red at baseline, green at head, negative control confirmed).

One note from reviewing #76: its hand-edited OpenAPI YAML adds nullable: true, which utoipa does not emit for Option IntoParams query fields — regenerating dirties the artifacts the PR asks to keep byte-identical. This PR regenerates them via pnpm generate.

[ravidsrk]

@claude review this PR. @codex review this PR.

[greptile-apps]

Greptile Summary

This PR fixes two bugs in POST /admin/api/dlq/{job_id}/replay: it was always returning replayed: 1 regardless of how many DLQ rows were moved, and it had no way to scope a replay to a single environment. Both are corrected cleanly.

• dlq::replay now returns u64 (the real moved-row count) instead of bool; the CLI adapts with a > 0 comparison and the handler casts to i64 for the response, consistent with the pre-existing replay_all pattern.
• An optional ?environment=<slug> query parameter is added; the slug is resolved to a UUID before the DELETE/INSERT CTE executes, so an unknown slug returns 404 without touching any row.
• Three failing-first HTTP integration tests over real Postgres cover the scoped replay, the unscoped multi-row count, and the unknown-slug 404-before-move guarantee.

Confidence Score: 4/5

Safe to merge — the environment-filter and honest-count changes are correct and thoroughly tested against real Postgres.

The core logic is solid: slug resolution happens before the atomic DELETE/INSERT CTE, the replayed count is accurate, and the CLI adapter is a trivial one-character fix. The only gap is a doc-level inconsistency in the utoipa 404 annotation.

The utoipa 404 annotation example field in server/src/api/admin.rs could be updated to cover the new no such environment case.

Important Files Changed

Filename	Overview
server/src/api/admin.rs	Adds optional ?environment query param to replay_dead_letter; resolves slug to UUID before any DB mutation (correct fail-before-move order); returns real row count instead of hardcoded 1.
server/src/dlq.rs	Signature of replay changed from anyhow::Result to anyhow::Result, eliminating the moved > 0 collapse and exposing the real count. SQL CTE logic is unchanged.
server/src/main.rs	CLI updated to compare dlq::replay result against > 0 instead of treating it as a bool; clean one-line adaptation.
server/tests/redteam_dlq_env_scope.rs	Three new HTTP integration tests added over real Postgres covering scoped replay, unscoped multi-row count, and unknown-slug 404-before-move.
docs/openapi/chimely.yaml	Generated via pnpm generate; adds the optional environment query param and updates the 404 description.
packages/client/src/generated/api.d.ts	Generated TypeScript types updated to reflect the optional environment query parameter and revised 404 description.

Sequence Diagram

%%{init: {'theme': 'neutral'}}%%
sequenceDiagram
    participant Client
    participant Handler as replay_dead_letter
    participant DB as Postgres

    Client->>Handler: "POST /admin/api/dlq/{job_id}/replay?environment=slug"
    Handler->>Handler: parse_typeid(job_id) - 404 on bad format
    alt environment param present
        Handler->>DB: "SELECT id FROM environments WHERE slug = $1"
        DB-->>Handler: Some(env_uuid) or None
        alt slug not found
            Handler-->>Client: 404 no such environment
        end
    end
    Handler->>DB: WITH revived AS DELETE FROM dead_letters INSERT INTO jobs
    DB-->>Handler: rows_affected u64
    alt "rows_affected == 0"
        Handler-->>Client: 404 no such parked job
    else
        Handler-->>Client: 200 replayed count
    end

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
sequenceDiagram
    participant Client
    participant Handler as replay_dead_letter
    participant DB as Postgres

    Client->>Handler: "POST /admin/api/dlq/{job_id}/replay?environment=slug"
    Handler->>Handler: parse_typeid(job_id) - 404 on bad format
    alt environment param present
        Handler->>DB: "SELECT id FROM environments WHERE slug = $1"
        DB-->>Handler: Some(env_uuid) or None
        alt slug not found
            Handler-->>Client: 404 no such environment
        end
    end
    Handler->>DB: WITH revived AS DELETE FROM dead_letters INSERT INTO jobs
    DB-->>Handler: rows_affected u64
    alt "rows_affected == 0"
        Handler-->>Client: 404 no such parked job
    else
        Handler-->>Client: 200 replayed count
    end

Loading

Comments Outside Diff (1)

1. server/src/api/admin.rs, line 1298-1303 (link)

   404 example covers only one of the two cases

   The description now says "No such parked job, or no such environment," but the example field still only shows the "no such parked job" message. A caller consulting the generated docs or the TypeScript types has no example of what the "no such environment" 404 looks like, even though the two cases produce distinct messages. Adding a second example (via examples) or updating the single one to something generic would align the annotation with the extended description.

   Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

   

_{Reviews (1): Last reviewed commit: "feat(admin): env filter on DLQ replay en..." | Re-trigger Greptile}

[ravidsrk]

Closing in deference to #76, per the one-PR-per-issue preference expressed in #84. The findings from this implementation are already commented on #76, and the branch stays available if any of the tests or fixes are worth cherry-picking.