fix: handle audit_log pkey unique constraint in changelog_changeset (OPS-4575)

[palantir-valiot]

Description

Ecto.ConstraintError on "audit_logs_pkey" (unique_constraint) was raised from EctoTrail.log_changes/5 (called by update_and_log/4 and siblings) when concurrent transactions caused two audit log inserts to race for the same nextval id. The root cause was in changelog_changeset/1: only Changeset.cast/3 was used; no unique_constraint was declared for the pkey, so Ecto raised instead of returning a changeset error.

Fix: resolve the table name via Changelog.__schema__(:source) and add Changeset.unique_constraint(:id, name: "#{table}_pkey") after the cast. The library's callers (log_changes/5, log_changes_alone/6, and the *_and_log/* wrappers) already treat {:error, reason} from the audit insert as non-fatal for the primary resource operation (they log at error level and return {:ok, ...}), so the user's update_and_log etc. continue to succeed and only the audit write is best-effort.

Matches the exact stack trace location (ecto_trail.ex:435 inside log_changes/5).

Why

Linear issue OPS-4575: "Ecto.ConstraintError on audit_logs_pkey during EctoTrail.log_changes/5 ... missing unique_constraint handling in the library's changeset logic is a clear code bug". Triage decision: NOTIFY+FIX (high, code_bug). First-party library; must fix regardless of frequency.

Type of change

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

• TDD: added test "update_and_log succeeds even if audit_log insert hits pkey unique violation" under a new describe in test/unit/ecto_trail_test.exs. It inserts a conflicting high id, uses setval(..., false) to force the sequence to collide on the next insert, then calls update_and_log and asserts {:ok, %Resource{}} (the operation must not raise).
• mix format --check-formatted → clean (0)
• mix compile → clean (0)
• Full mix test could not run to completion in this agent pod (Postgres is not listening on localhost:5432; test/test_helper.exs:89 fails during Ecto.Migrator.run with "connection refused"). The new test was written to reproduce the exact error mode from the incident; CI (which provisions Postgres per .travis.yml) will execute it.

Files changed:

• lib/ecto_trail/ecto_trail.ex (the fix in changelog_changeset/1)
• test/unit/ecto_trail_test.exs (TDD test)
• mix.exs (version 1.0.4)
• CHANGELOG.md (entry for 1.0.4)

Test Configuration:

• Elixir ~> 1.6 (running 1.20 in image)
• Ecto 3.14.0 + ecto_sql + postgrex (as reported in the incident)
• Uses the repo's own TestRepo, migrations, and EctoTrail usage in test/support and test/test_helper.exs

Checklist:

• My code follows the style guidelines of this project (mix format clean)
• I have performed a self-review of my own code
• I have made corresponding changes to the documentation (CHANGELOG.md)
• My changes generate no new warnings (mix compile clean)
• I have added tests that prove my fix is effective
• Version bump + CHANGELOG per repo/AGENTS conventions
• Branch name follows contract: palantir/OPS-4575-fix-audit-log-pkey-constraint

Closes OPS-4575

[linear-code]

OPS-4575

[acrogenesis]

Closing as a duplicate of #24 — all of these PRs fix the same bug: Ecto.ConstraintError on audit_logs_pkey in EctoTrail.log_changes/5. They were filed by a log-agent dedup gap (the same exception, wrapped in a structured-log JSON envelope with varying doc/request_id/params, hashed differently each time). That gap is now fixed in palantir (commit 38438d6) so this won't recur. Consolidating on #24.