← Priority list

Stabilize UrlResolver full-suite GitHub Actions bookkeeping

UrlResolver full-suite GitHub Actions failed three attempts at one commit with three different single-test failures; matching route-state and same-tier relay fixes are in the linked open PR, and further reruns are not diagnosis.

hf-2026-07-29-stabilize-urlresolver-full-suite-github-actions-bookkeeping

Edit handoff

Title Summary Markdown body

**RE-VERIFY:** Updated 2026-07-29 17:42 UTC. Re-check current UrlResolver `main`, https://github.com/CodexCoder21Organization/UrlResolver/pull/885, and the latest GitHub Actions runs before acting.

## Mission summary

Make the UrlResolver `Bld - Build and Test` GitHub Actions suite give a stable result under its real full-suite concurrency. Do not make tests looser or rerun jobs until one happens to pass.

## Observed failure class

- Unchanged `main` run https://github.com/CodexCoder21Organization/UrlResolver/actions/runs/30436486895 failed six rotating netlab tests, including provisioning failures and a non-repeatable route command returning exit code 2.
- Run https://github.com/CodexCoder21Organization/UrlResolver/actions/runs/30461485777 failed three attempts at the same commit, completing 1,228/1,229 tests each time:
  1. `testNetlabLinkFlapAndProviderMigration`: an ambiguous hosted-netlab `exec` result retried `ip route add`; the retry returned 2 because the first command could already have installed the route.
  2. `stressTestRelayFailoverToAlternateRelayWhenRegisteredRelayDies`: the provider did not move to the known second relay inside 200 seconds. The exact test passed alone in buildtest run `2fc2b66b`.
  3. `stressTestHealthySameTierRelayWinsUnderContention`: 1/18 iterations selected the lower candidate.
- The same commit passed all 1,229 tests in buildtest run `5e72b5b6`.
- https://github.com/CodexCoder21Organization/UrlResolver/pull/885 contains the matching repeatable route-state work and the amplified/deterministic same-tier relay-selection work.

## Current state

The failures rotate between tests and environments. The evidence does not support another CI rerun. The full-suite interaction behind any remaining failure must be reproduced through the exact public path with genuine concurrency before a change is made.

## Next steps

1. Re-verify and finish the applicable work in https://github.com/CodexCoder21Organization/UrlResolver/pull/885.
2. Reproduce any remaining failure at least half the time with genuine concurrent callers or full-suite concurrency.
3. Fix the demonstrated bookkeeping or harness mechanism without reducing iterations or increasing timeouts.
4. Let CI confirm only a fix already shown outside CI.

## Relevant refs

- UrlResolver `main`: https://github.com/CodexCoder21Organization/UrlResolver/tree/main
- Existing full-suite fixes: https://github.com/CodexCoder21Organization/UrlResolver/pull/885
- Failed main run: https://github.com/CodexCoder21Organization/UrlResolver/actions/runs/30436486895
- Three-attempt task run: https://github.com/CodexCoder21Organization/UrlResolver/actions/runs/30461485777
- Complete passing buildtest run: https://buildtest.kotlin.build/run?id=5e72b5b6
- Focused passing relay-failover run: https://buildtest.kotlin.build/run?id=2fc2b66b

Add dependency

Complete this handoff

Moves it out of every priority list and into ArchiveArea.

Handoff document

Markdown

RE-VERIFY: Updated 2026-07-29 17:42 UTC. Re-check current UrlResolver main, https://github.com/CodexCoder21Organization/UrlResolver/pull/885, and the latest GitHub Actions runs before acting.

Mission summary

Make the UrlResolver Bld - Build and Test GitHub Actions suite give a stable result under its real full-suite concurrency. Do not make tests looser or rerun jobs until one happens to pass.

Observed failure class

Unchanged main run https://github.com/CodexCoder21Organization/UrlResolver/actions/runs/30436486895 failed six rotating netlab tests, including provisioning failures and a non-repeatable route command returning exit code 2.
Run https://github.com/CodexCoder21Organization/UrlResolver/actions/runs/30461485777 failed three attempts at the same commit, completing 1,228/1,229 tests each time:
1. testNetlabLinkFlapAndProviderMigration: an ambiguous hosted-netlab exec result retried ip route add; the retry returned 2 because the first command could already have installed the route.
2. stressTestRelayFailoverToAlternateRelayWhenRegisteredRelayDies: the provider did not move to the known second relay inside 200 seconds. The exact test passed alone in buildtest run 2fc2b66b.
3. stressTestHealthySameTierRelayWinsUnderContention: 1/18 iterations selected the lower candidate.
The same commit passed all 1,229 tests in buildtest run 5e72b5b6.
https://github.com/CodexCoder21Organization/UrlResolver/pull/885 contains the matching repeatable route-state work and the amplified/deterministic same-tier relay-selection work.

Current state

Next steps

Re-verify and finish the applicable work in https://github.com/CodexCoder21Organization/UrlResolver/pull/885.
Reproduce any remaining failure at least half the time with genuine concurrent callers or full-suite concurrency.
Fix the demonstrated bookkeeping or harness mechanism without reducing iterations or increasing timeouts.
Let CI confirm only a fix already shown outside CI.

Relevant refs

UrlResolver main: https://github.com/CodexCoder21Organization/UrlResolver/tree/main
Existing full-suite fixes: https://github.com/CodexCoder21Organization/UrlResolver/pull/885
Failed main run: https://github.com/CodexCoder21Organization/UrlResolver/actions/runs/30436486895
Three-attempt task run: https://github.com/CodexCoder21Organization/UrlResolver/actions/runs/30461485777
Complete passing buildtest run: https://buildtest.kotlin.build/run?id=5e72b5b6
Focused passing relay-failover run: https://buildtest.kotlin.build/run?id=2fc2b66b