Emit events for deployment create, update, and delete operations

[devin-ai-integration]

Adds event emission for deployment CRUD operations (prefect.deployment.created, prefect.deployment.updated, prefect.deployment.deleted), following the same patterns used for work pool and work queue events.

Closes #20176

Changes

src/prefect/server/models/events.py

• Extracts _deployment_related_resources helper from deployment_status_event to build related resources (flow, work-queue, work-pool) shared across all deployment events
• Adds deployment_created_event, deployment_updated_event, deployment_deleted_event factory functions

src/prefect/server/models/deployments.py

• create_deployment: Captures a upsert_start timestamp before the upsert, then compares result_deployment.created >= upsert_start to reliably distinguish create vs update (same pattern used by the API layer for HTTP status codes). Snapshots field values into a plain dict before the upsert for change detection on the update path. Calls session.expire() on the pre-loaded ORM object to prevent stale state after the upsert.
• update_deployment: Snapshots deployment field values before and after update; emits updated event with changed field details (from/to values). Calls session.expire() after snapshotting.
• delete_deployment: Reads deployment and builds the delete event before deletion (while the object is still in session); emits deleted event after successful delete.
• delete_deployments: Loads full deployment objects (instead of just IDs) so delete events can be built before the bulk delete, then emits them after.
• Adds DEPLOYMENT_EVENT_FIELDS set (includes version, concurrency_limit_id, concurrency_options alongside the original fields), _detect_deployment_changed_fields helper, and emit_* wrapper functions

tests/server/orchestration/api/test_deployments.py

• 8 integration tests covering create, upsert-update, PATCH-update, no-op update, delete, nonexistent delete, bulk delete, and work-pool related resources
• Updates existing TestGetScheduledFlowRuns assertions to expect new prefect.deployment.created events alongside prefect.deployment.ready events

Updates since last revision

Addressed two review comments:

1. Race condition in concurrent upserts (P1): Replaced the pre-upsert existence check (existing_snapshot is None) with a timestamp comparison (result_deployment.created >= upsert_start) to determine create-vs-update. Under concurrent POSTs for the same (flow_id, name), both transactions could previously observe "not found" and both emit created events. Now the second request correctly detects the row already existed and emits an updated event instead.

2. Expanded tracked fields (P2): Added version, concurrency_limit_id, and concurrency_options to DEPLOYMENT_EVENT_FIELDS so that concurrency limit changes and version updates emit prefect.deployment.updated events.

Key review areas

• Timestamp-based create detection: create_deployment uses result_deployment.created >= upsert_start (mirroring the API layer's HTTP 201 logic). Relies on timestamp precision not being truncated by the DB below the resolution of now("UTC").
• concurrency_limit_id change detection: Concurrency limit changes are applied via _create_or_update_deployment_concurrency_limit, which modifies the ORM relationship. The post-upsert re-read (via populate_existing=True in create, or read_deployment in update) should pick up the new concurrency_limit_id value, but worth verifying this path works end-to-end.
• Extra DB reads per operation: create_deployment adds one pre-upsert query; update_deployment adds a pre-update read and a post-update read. Verify this performance trade-off is acceptable.
• session.expire() after snapshotting: In create_deployment and update_deployment, the existing ORM object is expired after its field values are copied into a plain dict. This prevents stale state when downstream code (e.g. _create_or_update_deployment_concurrency_limit) later accesses the object via session.get(). Verify this doesn't cause unexpected lazy-load behavior.
• delete_deployments now loads full objects instead of just IDs. The .unique() call was added. Confirm this doesn't change existing return behavior or cause issues at scale.

Checklist

• This pull request references any related issue by including "closes <link to issue>"
• If this pull request adds new functionality, it includes unit tests that cover the changes
• If this pull request removes docs files, it includes redirect settings in mint.json.
• If this pull request adds functions or classes, it includes helpful docstrings.

Link to Devin session: https://app.devin.ai/sessions/6e878425faa64628980a64c78a432911
Requested by: @desertaxle

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring

[codspeed-hq]

Merging this PR will not alter performance

✅ 2 untouched benchmarks

---

_{Comparing devin/1773689161-deployment-crud-events (7871929) with main (ba5a447)}

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no potential bugs to report.

View in Devin Review to see 6 additional findings.

[zzstoatzz]

lgtm!

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9de89417cb

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".