nx: Problem with NX Daemon after update from 15.9.2 to 16.0.3

I think I have a 3rd related issue w/ the error Watch error: Daemon closed the connection occurring after every git commit since upgrading to NX 16.

I’m not really sure how to debug since --verbose gives me no error stack or additional errors. I don’t know how to access the logs from the process where the error is occurring.

I’m using executor @nx/js:node

I’m having the same issue (after moving to 16.x):

fatal: no path specified

It often (but not 100% of the time) seems to be right after a watcher line:

[WATCHER]: path/to/some/directory was created or restored
fatal: no path specified

With the @nx/js:node executor, this is not super convenient if the root cause is files changing, as it is used for the serve task… 😅 Any time a file changes, the process exits:

>  NX  File change detected. Restarting...

node:internal/errors:867
  const err = new Error(message);
              ^

Error: Command failed: taskkill /pid 16428 /T /F
ERROR: The process "16428" not found.

16.5 fixes this issue for me

I can easily reproduce it with these targets:

{
  "targets": {
    "echo": {
      "inputs": [],
      "outputs": ["{projectRoot}"],
      "dependsOn": ["non-cacheable-echo"],
      "executor": "nx:run-commands",
      "options": {
        "command": "true"
      }
    },
    "non-cacheable-echo": {
      "inputs": [],
      "outputs": ["{projectRoot}"],
      "dependsOn": ["cacheable-echo"],
      "executor": "nx:run-commands",
      "options": {
        "command": "true"
      }
    },
    "cacheable-echo": {
      "inputs": [],
      "outputs": ["{projectRoot}"],
      "executor": "nx:run-commands",
      "options": {
        "command": "true"
      }
    }
  }
}

if you run echo target at least twice - you will see the error:

>  NX   Error: write EPIPE


   Messages from the log:
   [NX Daemon Server] - 2023-09-25T11:42:25.986Z - [REQUEST]: Client Request for Project Graph Received
   [NX Daemon Server] - 2023-09-25T11:42:26.395Z - [REQUEST]: Updated file-hasher based on watched changes, recomputing project graph...
   [NX Daemon Server] - 2023-09-25T11:42:26.395Z - [REQUEST]:
   [NX Daemon Server] - 2023-09-25T11:42:26.395Z - [REQUEST]:
   [NX Daemon Server] - 2023-09-25T11:42:26.471Z - [REQUEST]: Responding to the client. project-graph
   [NX Daemon Server] - 2023-09-25T11:42:26.473Z - Time taken for 'total for creating and serializing project graph' 484.8588333129883ms
   [NX Daemon Server] - 2023-09-25T11:42:26.474Z - Done responding to the client project-graph
   [NX Daemon Server] - 2023-09-25T11:42:26.474Z - Handled REQUEST_PROJECT_GRAPH. Handling time: 485. Response time: 3.
   [NX Daemon Server] - 2023-09-25T11:42:26.522Z - [REQUEST]: Responding to the client. handleHashTasks
   [NX Daemon Server] - 2023-09-25T11:42:26.523Z - Done responding to the client handleHashTasks
   [NX Daemon Server] - 2023-09-25T11:42:26.523Z - Handled HASH_TASKS. Handling time: 34. Response time: 1.
   [NX Daemon Server] - 2023-09-25T11:42:26.617Z - [REQUEST]: Responding to the client. outputsHashesMatch
   [NX Daemon Server] - 2023-09-25T11:42:26.617Z - Done responding to the client outputsHashesMatch
   [NX Daemon Server] - 2023-09-25T11:42:26.617Z - Handled OUTPUTS_HASHES_MATCH. Handling time: 61. Response time: 0.
   [NX Daemon Server] - 2023-09-25T11:42:26.683Z - [WATCHER]: Processing file changes in outputs
   [NX Daemon Server] - 2023-09-25T11:42:26.686Z - [WATCHER]: 0 file(s) created or restored, 0 file(s) modified, 590 file(s) deleted
   [NX Daemon Server] - 2023-09-25T11:42:26.686Z - [WATCHER]: Stopping the watcher for /Users/platform (sources)
   [NX Daemon Server] - 2023-09-25T11:42:26.686Z - [WATCHER]: Stopping the watcher for /Users/platform (outputs)
   [NX Daemon Server] - 2023-09-25T11:42:26.686Z - Server stopped because: "Stopping the daemon the set of ignored files changed (native)"

I suspect it only happens if the project, you are running the target on, itself has a .gitignore file in it’s root. So while it tries to replace .gitignore file with the same one - it triggers some event, which causes the error.

Have the same issue after upgrading nx to 16.4.0 even in a CI, though Nx Daemon should be disabled by default in a CI environment as described in the docs https://nx.dev/more-concepts/nx-daemon#turning-it-off

Not sure but on my side it looks like a combination of the @nx/node executor and the Webstorm “Nx Console” plugin (on windows). When i disable the plugin and run it with the command, it runs correctly, but if i turn the plugin on again it closes the existing serve process started with @nx/node.