Ability to override API_SANDBOXES_OPERATION_TIMEOUT

[astound-ci]

There was a strange situation yesterday, request for sandbox creation took 41 minutes. And taking to account that API_SANDBOXES_OPERATION_TIMEOUT const is set to 10 minutes w/o ability to override - automated process failed w/o any error. Proposal is to add ability to override this const, additionally would be nice to have a reasonable error message in case default timeout exceeded.

Command used:
sfcc-ci-linux sandbox:create --json --sync --realm ${SANDBOX_REALM_ID} --ttl ${TTL}

[astound-ci]

@tobiaslohr, please make API_SANDBOXES_OPERATION_TIMEOUT at least 15-20 minutes, because now option --sync for sandbox creation is useless as it fails in 80% cases.

[astound-ci]

In case it for any reason not appropriate to increase API_SANDBOXES_OPERATION_TIMEOUT to a greater value just return sandbox id which is still creating/starting so that it can be handled in any CI tool.

[sholsinger]

As a workaround... @astound-ci you should be able to remove the --sync option and get the UUID back. Then you can have a script sit in a tight loop polling for the status to become 'created' or 'up' or whatever it says when a sandbox is created and running.

[astound-ci]

Yeah @sholsinger, as you mentioned this is a workaround and not quite optimal as there will be many API requests sent, polling the status. Moreover we have a process that is bind to a created sandbox so than it can be assigned users and so on. And that process waits for a json file with initial data set.

[tobiaslohr]

@astound-ci we will leave API_SANDBOXES_OPERATION_TIMEOUT at 10 mins and come up with a new env var to overwrite this. With that you can set it to 15 or 20 mins if needed.

[astound-ci]

Thanks @tobiaslohr, that will help us much. When do you plan to release?

[tobiaslohr]

When do you plan to release?

Currently planned for 2.4.x release. We will most likely build and release v2.3.1 today, then start working on 2.4.x. Contributions are welcome :)

In case it for any reason not appropriate to increase API_SANDBOXES_OPERATION_TIMEOUT to a greater value just return sandbox id which is still creating/starting so that it can be handled in any CI tool.

@astound-ci the sandbox id is included in the creation result in any case. In case of --sync mode, and the creation runs into the operation timeout, the message is containing the sandbox id already. See below for an example:

{
  "message": "Creation of new sandbox 001 for realm zzzz triggered and ongoing. Sandbox id is e7e3d222-291d-49a9-97ad-c69a126f527c, status of sandbox is new. You may use `sfcc-ci sandbox:list` to check the status of the sandbox. New sandbox host zzzz-001.sandbox.us01.dx.commercecloud.salesforce.com added to list of instances using alias \"zzzz-001\". New sandbox set as default instance.",
  "sandbox": {
    "id": "e7e3d222-291d-49a9-97ad-c69a126f527c",
    "realm": "zzzz",
    "instance": "001",
    "versions": {
      "app": "20.3.2.579",
      "web": "19.8.4.0"
    },
    "state": "starting",
    "createdAt": "2020-03-24T15:22:16.799Z",
    "createdBy": "[email protected]",
    "eol": "2020-03-24T18:22:16.795Z",
    "links": {
      "bm": "https://zzzz-001.sandbox.us01.dx.commercecloud.salesforce.com/on/demandware.store/Sites-Site",
      "ocapi": "https://zzzz-001.sandbox.us01.dx.commercecloud.salesforce.com/s/-/dw/data/",
      "impex": "https://zzzz-001.sandbox.us01.dx.commercecloud.salesforce.com/on/demandware.servlet/webdav/Sites/Impex",
      "code": "https://zzzz-001.sandbox.us01.dx.commercecloud.salesforce.com/on/demandware.servlet/webdav/Sites/Cartridges",
      "logs": "https://zzzz-001.sandbox.us01.dx.commercecloud.salesforce.com/on/demandware.servlet/webdav/Sites/Logs"
    }
  },
  "instance": {
    "host": "zzzz-001.sandbox.us01.dx.commercecloud.salesforce.com",
    "alias": "zzzz-001",
    "default": true
  },
  "error": "Sandbox creation timeout has been exceeded."
}

What exactly is missing?

[astound-ci]

@tobiaslohr we figured out how to overcome this error by redirecting stderr to stdout and wrapping to try/catch, all needed info is there. But still would be nice to have ENV var for timeout override. Thanks