[BUG] Network connection lost after restart Service Binding target worker

[ka2n]

Describe the bug

When I try to communicate using RRC from Next.js app to other worker, sometimes connection between Next.js worker and other worker are lost.

• Launch Next.js before other worker launch (and ~/.config/.wrangler/registry/<worker-name>-<environment name> are created, then service's port available).
• After other worker reloaded due to code change or restarted by user, and service's port was changed.

Steps to reproduce

1. Start Next.js app and other worker.
2. Kill other worker, and restart.
3. The error raised when hard reloading browser.

Expected behavior

In local development, platform proxy in getCloudflareContext should always see the fresh content of the registry entry.

@opennextjs/cloudflare version

0.5.3

Wrangler version

3.109.2

next info output

Operating System:
  Platform: linux
  Arch: x64
  Version: #1 ZEN SMP PREEMPT_DYNAMIC Sat, 08 Feb 2025 18:54:38 +0000
  Available memory (MB): 64215
  Available CPU cores: 24
Binaries:
  Node: 22.13.1
  npm: 10.9.2
  Yarn: N/A
  pnpm: 9.14.2
Relevant Packages:
  next: 14.2.24 // An outdated version detected (latest is 15.1.7), upgrade is highly recommended!
  eslint-config-next: 14.2.24
  react: 18.3.1
  react-dom: 18.3.1
  typescript: 5.6.3
Next.js Config:
  output: standalone

Additional context

No response

[vicb]

@dario-piotrowicz this seems to be cause by the wrangler API. Do you know if it's a known limitation or a bug?