Add new operations to Random operations test

[a-masterov]

Problem

We did not test some Public API calls, such as using a timestamp to create a branch, reset_to_parent.

Summary of changes

Tests now include some other operations: reset_to_parent, a branch creation from any time in the past, etc.
Currently, the API calls are only exposed; the semantics are not verified.

[devin-ai-integration]

Code Review: Add new operations to Random operations test

✅ Approved - Great expansion of test coverage!

This PR significantly enhances the Random operations test by adding comprehensive snapshot functionality and reset-to-parent operations. The implementation is well-structured and follows existing patterns.

Strengths:

• Comprehensive snapshot support: Full lifecycle management (create, restore, delete) with proper API integration
• Enhanced timing logic: The method properly handles timing constraints to avoid invalid timestamps
• Proper benchmark coordination: Correctly terminates and restarts pgbench processes during disruptive operations
• Good error handling: Proper limit checking and graceful fallbacks when operations can't be performed
• Weighted operation distribution: Thoughtful probability weights for new operations (snapshots have low weights for gradual introduction)

Suggestions for improvement:

1. 🔧 Remove debug log (Line 122):

   # XXX for debug only, remove before merge
   log.info("%s", branch["branch"])

   This debug log should be removed before merging.

2. 📝 Consider adding docstrings for new methods like reset_to_parent() and create_snapshot() to match the existing documentation style.

3. 🧪 Test edge cases: Consider testing behavior when:

   • No snapshots exist for restore/delete operations
   • Branch limits are reached during snapshot restoration
   • Timing constraints prevent valid random time generation

Technical Notes:

• The new NeonSnapshot class follows the same pattern as NeonBranch and NeonEndpoint
• API client additions in neon_api.py are well-structured with proper type hints
• The do_action() function cleanly handles the 5 new operation types
• Benchmark restart logic during reset operations looks robust

Testing Recommendation:

Run with a smaller operation count (e.g., 50) initially to validate the new snapshot and reset operations work correctly under various timing conditions.

Great work expanding the test coverage! 🚀

[devin-ai-integration]

Code Review: Add new operations to Random operations test

✅ Approved - Great expansion of test coverage!

This PR significantly enhances the Random operations test by adding comprehensive snapshot functionality and reset-to-parent operations. The implementation is well-structured and follows existing patterns.

Strengths:

• Comprehensive snapshot support: Full lifecycle management (create, restore, delete) with proper API integration
• Enhanced timing logic: The random_time() method properly handles timing constraints to avoid invalid timestamps
• Proper benchmark coordination: Correctly terminates and restarts pgbench processes during disruptive operations
• Good error handling: Proper limit checking and graceful fallbacks when operations can't be performed
• Weighted operation distribution: Thoughtful probability weights for new operations (snapshots have low weights for gradual introduction)

Suggestions for improvement:

1. 🔧 Remove debug log (Line 122):

   # XXX for debug only, remove before merge
   log.info("%s", branch["branch"])

   This debug log should be removed before merging.

2. 📝 Consider adding docstrings for new methods like reset_to_parent() and create_snapshot() to match the existing documentation style.

3. 🧪 Test edge cases: Consider testing behavior when:

   • No snapshots exist for restore/delete operations
   • Branch limits are reached during snapshot restoration
   • Timing constraints prevent valid random time generation

Technical Notes:

• The new NeonSnapshot class follows the same pattern as NeonBranch and NeonEndpoint
• API client additions in neon_api.py are well-structured with proper type hints
• The do_action() function cleanly handles the 5 new operation types
• Benchmark restart logic during reset operations looks robust

Testing Recommendation:

Run with a smaller operation count (e.g., 50) initially to validate the new snapshot and reset operations work correctly under various timing conditions.

Great work expanding the test coverage! 🚀

[github-actions]

8910 tests run: 8245 passed, 0 failed, 665 skipped (full report)

---

Flaky tests (8)

Postgres 17

• test_hot_standby_gc[True]: release-x86-64-without-lfc, release-x86-64-with-lfc
• test_total_size_limit: release-arm64-with-lfc
• test_ps_unavailable_after_delete[DeletionAPIKind.FORCE]: release-arm64-without-lfc

Postgres 16

• test_ps_unavailable_after_delete[DeletionAPIKind.FORCE]: release-x86-64-with-lfc, release-arm64-with-lfc

Postgres 15

• test_ps_unavailable_after_delete[DeletionAPIKind.FORCE]: release-arm64-with-lfc

Postgres 14

• test_ps_unavailable_after_delete[DeletionAPIKind.FORCE]: release-x86-64-with-lfc

Code coverage* (full report)

• functions: 34.8% (8796 of 25281 functions)
• lines: 46.0% (71105 of 154711 lines)

* collected from Rust tests only

---

_{The comment gets automatically updated with the latest test results
4be46b1 at 2025-07-22T17:53:49.494Z :recycle:}

[Bodobolero]

LGTM