test: restore the old test 1609 as 3271

[bagder]

It was wrongly removed as a duplicate but the code remained in tests/unit and was never exercised.

The internal dns cache logic and API changed in the meantime so test is also updated to reflect those.

Follow-up to 4ead428

[Copilot]

Copilot review overview

🟡 Changes recommended

The test recreates its DNS cache each iteration, so it never exercises cache-entry replacement or permanence conversion.

Get a fresh assessment by requesting another Copilot review.

Review effort: Balanced
Findings: 1

Open (1)

• Update helper provenance from unit3217 to unit3271 · New

What changed in this PR

Restores test 1609 as unit test 3271 and adapts it to the current DNS cache key API.

Changes:

• Registers and renames the restored unit test.
• Uses peer-based DNS cache keys.
• Exposes internal key helpers for unit testing.

File	Description
tests/​unit/​unit3271.c	Restores and updates the resolve override test.
tests/​unit/​Makefile.inc	Registers unit3271.
tests/​data/​test3271	Defines the test case metadata.
tests/​data/​Makefile.am	Adds test3271 to the suite.
lib/​vdns/​dnscache.h	Exposes DNS key structures.
lib/​vdns/​dnscache.c	Makes key helpers unit-testable.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[testclutch]

Analysis of PR #23052 at 8bf2dae5:

Test 1041 failed, but it has been 0.7% flaky lately, so it's probably NOT a fault of the PR. Note that this test has failed in 12 different CI jobs (the link just goes to one of them). Note that this CI job has had a number of other flaky tests recently (3, to be specific) so it may be that this failure is rather a systemic issue with this job and not with this specific PR.

Generated by Testclutch

[Copilot]

Copilot review overview

🔵 Needs a closer look

The test recreates its DNS cache between cases and does not verify the intended permanence behavior.

Review effort: Balanced
Findings: None

Resolved since last review (1)

• Update helper provenance from unit3217 to unit3271

Previously missed (2)

In code that hasn't changed since last review

Reuse one multi handle and DNS cache across iterations

tests/​unit/​unit3271.c:102

This still creates and destroys multi inside every iteration (lines 121 and 204). Each multi owns its DNS cache, so the second case never sees the first entry and therefore tests two independent insertions rather than the advertised override. Keep one multi/cache alive across both iterations (removing each easy handle before cleanup), then destroy it after the loop.

Verify CURLOPT_RESOLVE replacement changes cache entry permanence

tests/​unit/​unit3271.c:141

The retrieved cache entry is only checked for its address and port, so the stated expiry/permanence regression can pass even if CURLOPT_RESOLVE leaves the replacement non-permanent. Seed the first shared-cache entry as non-permanent (using the + form), then assert dns->permanent is false before the replacement and true afterward.