Retry for deploy

[peterpeterparker]

I often get following issue which can probably (maybe) be solved by a retry strategy.

In this case your connection is closed so you won't get any status code to retry on. You should dive into an error (Exception?) and figure out if it was caused by this error (SocketError: other side closed or code: 'UND_ERR_SOCKET') and retry if it is. If you get 429 then for sure it's safe to retry on it.

An unexpected error happened 😫.
c3 [AgentCallError]: Error while making call: terminated
    at e.call (file:///usr/local/lib/node_modules/@junobuild/cli/dist/index.js:31:53799)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async i (file:///usr/local/lib/node_modules/@junobuild/cli/dist/index.js:57:4993)
    at async s5t (file:///usr/local/lib/node_modules/@junobuild/cli/dist/index.js:635:128506)
    at async c5t (file:///usr/local/lib/node_modules/@junobuild/cli/dist/index.js:635:128965)
    at async f5t (file:///usr/local/lib/node_modules/@junobuild/cli/dist/index.js:635:129718)
    at async a (file:///usr/local/lib/node_modules/@junobuild/cli/dist/index.js:643:611)
    at async i (file:///usr/local/lib/node_modules/@junobuild/cli/dist/index.js:621:5080)
    at async l.task (file:///usr/local/lib/node_modules/@junobuild/cli/dist/index.js:621:5424) {
  response: TypeError: terminated
      at Fetch.onAborted (node:internal/deps/undici/undici:10473:53)
      at Fetch.emit (node:events:519:28)
      at Fetch.terminate (node:internal/deps/undici/undici:9641:14)
      at Object.onError (node:internal/deps/undici/undici:10587:38)
      at Request.onError (node:internal/deps/undici/undici:6381:31)
      at Object.errorRequest (node:internal/deps/undici/undici:1063:17)
      at TLSSocket.<anonymous> (node:internal/deps/undici/undici:7639:16)
      at TLSSocket.emit (node:events:531:35)
      at node:net:338:12
      at TCP.done (node:_tls_wrap:657:7) {
    [cause]: SocketError: other side closed
        at TLSSocket.<anonymous> (node:internal/deps/undici/undici:7614:28)
        at TLSSocket.emit (node:events:531:35)
        at endReadableNT (node:internal/streams/readable:1696:12)
        at process.processTicksAndRejections (node:internal/process/task_queues:82:21) {
      code: 'UND_ERR_SOCKET',
      socket: [Object]
    }
  },
  requestId: '300472d6fd9dc0b372a649d14432688e6882b6a58819a5740ca7646a7985d42e',
  senderPubkey: '302a300506032b65700321006e1daef97ef3adbb9b319ec9f7473b929ed41a24db0e22a8d1a43609b8306b34',
  senderSig: 'd3f9ae59897d52ea5bc6191ecb3a5f9e7aa30c072364538ec5296ca72d59b1616c71909a53e2e7392035f35ec91c870d579ac6a9d823b68562bf3b78741e9f0e',
  ingressExpiry: '1748612700000000000',
  context: [Object: null prototype] {}
}

Example: https://github.com/junobuild/docs/actions/runs/15348196400/job/43189397544

[peterpeterparker]

Actually not sure we need a retry. We probably rather need the fix developed in Agent-js dfinity/icp-js-core#1031 to resolve the issue.

[ilbertt]

Yes, I think dfinity/icp-js-core#1031 will fix this issue, following the default retry logic of the agent or the retry logic that you provide. I've tested it locally and the agent now retries the request when facing a terminated error.

I've also added tests to cover the case: https://github.com/dfinity/agent-js/blob/66be1aa249003358dbe9b6a196ccb7dd5fa2fc8f/packages/agent/src/agent/http/http.test.ts#L820-L893

[peterpeterparker]

Discussed today, we still noticed (other) issues after upgrade to AgentJS v3, so still worth implementing some "retry" logic.

[FAILED: Request timed out after 300000 msec
  Request ID: cf184383e5fa2947a474d185ac02f507851655d6ffb85ae6c0db371099728550
  Request status: received
]
An unexpected error happened 😫.

Request timed out after 300000 msec
  Request ID: cf184383e5fa2947a474d185ac02f507851655d6ffb85ae6c0db371099728550
  Request status: received