TestServicePostgres hangs intermittently in E2E tests

[dpup]

Summary

TestServicePostgres in internal/e2e/services_test.go hangs intermittently during E2E test runs.

Details

The test has been skipped with t.Skip() to unblock CI. The hang likely involves the custom network + proxy setup interaction, possibly related to synthetic hostname resolution on custom Docker networks.

Steps to reproduce

Run the E2E test suite repeatedly:

go test -tags=e2e -v -run TestServicePostgres ./internal/e2e/

Expected behavior

The test should complete within the test timeout.

Actual behavior

The test hangs indefinitely (or until the test timeout is hit).

Investigation notes

• May be related to the synthetic hostname changes in the proxy bypass fix (fix/host-traffic-proxy-bypass branch)
• Custom network gateway rewriting and --add-host entries may not resolve correctly in all network configurations
• The proxy's forwarding of allowed host-gateway traffic now rewrites moat-host → actual IP, which may interact with service container networking

[dpup]

Also affecting TestServiceRedis with the same hang signature — WaitContainer blocks for the full test timeout. Goroutine dump from a Mac e2e run shows it stuck at internal/container/docker.go:338 (DockerRuntime.WaitContainer) for 8+ minutes. Skipping TestServiceRedis as well until root cause is identified.

[dpup]

Also affecting TestServiceMultiple — same hang signature as TestServicePostgres and TestServiceRedis. Skipping for now; root cause likely common to all service-dependency tests using container WaitContainer.