fix: sync event refresh blocking updates when Spock node is not configured

[tsivaprasad]

Summary

This PR fixes hard errors returned by SyncEventResource and WaitForSyncEventResource refresh methods when the database is not configured as a Spock node, which previously blocked follow-up update operations.

Changes

• Return resource.ErrNotFound from SyncEventResource.Refresh when spock.sync_event() fails
• Return resource.ErrNotFound from WaitForSyncEventResource.Refresh when spock.sub_show_status() or spock.wait_for_sync_event() fails
• Treat an empty SyncEventLsn as ErrNotFound instead of a hard error
• Add postgres.IsSpockNodeNotConfigured helper to detect object_not_in_prerequisite_state
• Update NOTICE.txt for the new github.com/jackc/pgerrcode dependency

Testing

Verification:

1. Created a 3 node cluster
2. Created a database with single node
3. Connected to host and db and dropped spock node

select spock.node_drop('n1');
 node_drop 
-----------
 t
(1 row)

select * from spock.node;
 node_id | node_name | location | country | info 
---------+-----------+----------+---------+------
(0 rows)

4. Updated a database with new node n2

"nodes": [
      {"name": "n1", "host_ids": ["host-1"]},
      {"name": "n2", "host_ids": ["host-2"]}
    ]

5. Verified that node n2 is added to the database & spock.nodes table has both n1 and n2

 select * from spock.node;
 node_id | node_name | location | country | info 
---------+-----------+----------+---------+------
   49708 | n1        |          |         | 
   26863 | n2        |          |         | 
(2 rows)

6. List databases and verify that n2 is added to the database

{
  databases: [
    {
      created_at: "2026-02-19T16:39:17Z"
      id: "storefront"
      instances: [
        {
          host_id: "host-1"
          id: "storefront-n1-689qacsi"
          node_name: "n1"
          state: "available"
        }
        {
          host_id: "host-2"
          id: "storefront-n2-9ptayhma"
          node_name: "n2"
          state: "available"
        }
      ]
      state: "available"
      updated_at: "2026-02-20T01:59:41Z"
    }
  ]
}

Checklist

• Tests added or updated
• Changelog

PLAT-420

[coderabbitai]

📝 Walkthrough

Walkthrough

Adds third-party license notices, introduces a postgres error helper to detect unconfigured Spock nodes with tests, updates database resource error handling to return standardized not-found errors, adds a changelog entry, and updates go.mod with an indirect dependency.

Changes

Cohort / File(s)	Summary
License Notices NOTICE.txt	Appends extensive third-party license and attribution texts for multiple dependencies (MIT, Apache-2.0, BSD-3-Clause, MPL-2.0, etc.).
Dependencies go.mod	Adds indirect dependency github.com/jackc/pgerrcode v0.0.0-20250907135507-afb5586c32a6 and consolidates require lines into a grouped block.
Changelog changes/unreleased/Fixed-20260220-191533.yaml	Adds an unreleased entry documenting a fix for sync_event and wait_for_sync_event blocking updates when the DB is not a Spock node.
Postgres Error Handling server/internal/postgres/errors.go, server/internal/postgres/errors_test.go	Adds exported IsSpockNodeNotConfigured(error) bool to detect SQLSTATE 55000 and a table-driven test covering wrapped and non-wrapped PgError cases. Attention: new exported function and test coverage.
Database Resources server/internal/database/sync_event_resource.go, server/internal/database/wait_for_sync_event_resource.go	Modifies error handling to return resource.ErrNotFound when Spock node is not configured; removes LSN-specific error formatting in one path. Review call sites for behavior expectations.

Poem

🐰 I dug through logs and license trails,
Found Spock nodes hiding in postgres vales,
I wrapped errors neat and chased the fright,
Pulled licenses close and set them right,
Hop — the code now sleeps soundly through the night.

🚥 Pre-merge checks | ✅ 2 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Title check	✅ Passed	The PR title clearly and concisely summarizes the main change: fixing sync event refresh behavior when a Spock node is not configured, which is directly supported by the changeset.
Description check	✅ Passed	The PR description comprehensively covers all required sections: Summary, Changes, Testing with detailed verification steps, Checklist with completed items, and issue linking.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

• 📝 Generate docstrings (stacked PR)
• 📝 Generate docstrings (commit on current branch)

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch PLAT-420-refresh-lifecycle-methods-in-sync-event-wait-for-sync-event-resources-can-block-updates

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 3

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@go.mod`:
- Line 49: The go.mod entry for github.com/jackc/pgerrcode is marked "//
indirect" but it's actually imported directly in
server/internal/postgres/errors.go (the import of github.com/jackc/pgerrcode),
so remove the "// indirect" annotation and place that require line into the
direct-dependencies section of go.mod; after editing go.mod run "go mod tidy" to
update the module graph and ensure the dependency is recorded as direct.

In `@NOTICE.txt`:
- Around line 3976-3986: The second github.com/jackc/pgerrcode entry erroneously
labels the block as “[PostgreSQL]” while the content is the MIT license; locate
the duplicated github.com/jackc/pgerrcode block and either correct the SPDX
label to reflect MIT (change “[PostgreSQL]” to “[MIT]”) or, if you cannot alter
the tool output, insert a single-line explanatory HTML comment immediately above
the second block (e.g., <!-- The single LICENSE file covers both licenses; full
text repeated per tool output -->) so auditors understand the repetition is
intentional.

In `@server/internal/postgres/errors_test.go`:
- Around line 44-51: The test case titled "returns true when wrapped with
fmt.Errorf" doesn't actually wrap the PgError; update the err function to return
a wrapped error using fmt.Errorf with %w (e.g. create pgErr :=
&pgconn.PgError{Code: pgerrcode.ObjectNotInPrerequisiteState} then return
fmt.Errorf("wrapped: %w", pgErr)) so the test exercises errors.As chain
traversal used by IsSpockNodeNotConfigured; keep expected: true.

[coderabbitai]

⚠️ Potential issue | 🟡 Minor

Second pgerrcode entry is labeled [PostgreSQL] but its content block opens with MIT license text.

Lines 3976–4030 duplicate the first entry (lines 3920–3974) byte-for-byte — the content block starts with "MIT License" / "Copyright (c) 2019 Jack Christensen" under a header that advertises the PostgreSQL license. Anyone auditing the [PostgreSQL] section for compliance would immediately encounter the wrong license header.

This looks like a tool artifact where the generator repeats the full LICENSE file content under every detected SPDX identifier. If the tooling cannot be changed, a brief explanatory comment above the second block (e.g., <!-- The single LICENSE file covers both licenses; full text repeated per tool output -->) would reduce confusion during legal review.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@NOTICE.txt` around lines 3976 - 3986, The second github.com/jackc/pgerrcode
entry erroneously labels the block as “[PostgreSQL]” while the content is the
MIT license; locate the duplicated github.com/jackc/pgerrcode block and either
correct the SPDX label to reflect MIT (change “[PostgreSQL]” to “[MIT]”) or, if
you cannot alter the tool output, insert a single-line explanatory HTML comment
immediately above the second block (e.g., <!-- The single LICENSE file covers
both licenses; full text repeated per tool output -->) so auditors understand
the repetition is intentional.

[jason-lynch]

Nice! I'd ignore the NOTICE.txt comment, but the other two code rabbit comments in errors_test.go and go.mod are valid. They're definitely minor issues, but could you please fix them up before you merge?

[coderabbitai]

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Duplicate comments:
In `@go.mod`:
- Around line 47-50: The go.mod currently has github.com/gohugoio/hashstructure
listed as indirect while github.com/jackc/pgerrcode remains in the wrong block;
run `go mod tidy` to let Go reorder dependencies so pgerrcode moves into the
primary require block and hashstructure is marked as indirect, or manually move
github.com/jackc/pgerrcode into the main direct-deps require block and restore
github.com/gohugoio/hashstructure to an indirect entry (// indirect) so the file
follows standard direct/indirect grouping.