Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Setting callback for on 'data' event of stderr causes error on v10.0.0 #21203

Closed
jesobreira opened this issue Jun 8, 2018 · 13 comments
Closed
Labels
confirmed-bug Issues with confirmed bugs. net Issues and PRs related to the net subsystem. process Issues and PRs related to the process subsystem. stream Issues and PRs related to the stream subsystem.

Comments

@jesobreira
Copy link

  • Version: 10.0.0
  • Platform: OS X 10.13.2
  • Subsystem: Process

I am unable to set a callback on the 'data' event (since process.stderr is a net.Socket stream, it should emit the data event).

The following snippet works on v8.0.0 but throws an error on v10.0.0:

process.stderr.on('data', function(data) {
    // not relevant, error is threw anyway
});

Error threw:

Error: read ENOTCONN
    at WriteStream.Socket._read (net.js:530:20)
    at WriteStream.Readable.read (_stream_readable.js:458:10)
    at resume_ (_stream_readable.js:897:12)
    at process._tickCallback (internal/process/next_tick.js:174:19)
    at Function.Module.runMain (internal/modules/cjs/loader.js:721:11)
    at startup (internal/bootstrap/node.js:228:19)
    at bootstrapNodeJSCore (internal/bootstrap/node.js:575:3)
Emitted 'error' event at:
    at emitErrorNT (internal/streams/destroy.js:92:8)
    at emitErrorAndCloseNT (internal/streams/destroy.js:59:3)
    at process._tickCallback (internal/process/next_tick.js:174:19)
    [... lines matching original stack trace ...]
    at bootstrapNodeJSCore (internal/bootstrap/node.js:575:3)
@ryzokuken
Copy link
Contributor

Okay, it doesn't throw on Node 8 and throws on 10.

/cc @nodejs/process @nodejs/streams

@ryzokuken ryzokuken added process Issues and PRs related to the process subsystem. confirmed-bug Issues with confirmed bugs. stream Issues and PRs related to the stream subsystem. labels Jun 8, 2018
@ryzokuken
Copy link
Contributor

Even if it's a valid error, we should try throwing something descriptive.

@mcollina mcollina added the net Issues and PRs related to the net subsystem. label Jun 8, 2018
@mcollina
Copy link
Member

mcollina commented Jun 8, 2018

This is likely a bug on stdout/stderr and net, and not on stream.

@jrasanen
Copy link
Contributor

jrasanen commented Jun 9, 2018

I git bisected when the error was introduced, if that helps (this commit):

ae2b5bcb7c17a2d2a488f234c736201eed8200db is the first bad commit
commit ae2b5bcb7c17a2d2a488f234c736201eed8200db
Author: cjihrig <[email protected]>
Date:   Mon Apr 2 13:33:48 2018 -0400

    deps: upgrade libuv to 1.20.0

    Notable changes:
    - uv_fs_copyfile() adds support for copy-on-write behavior.
    - uv_relative_path() now uses the long directory name
      for handle->dirw.
    - File operations on files > 2 GB on 32-bit platforms are
      working again.
    - uv_fs_fchmod() on Windows works on files with the
      Archive flag cleared.

    Fixes: https://github.com/nodejs/node/issues/19170
    Fixes: https://github.com/nodejs/node/issues/19455
    Fixes: https://github.com/nodejs/node/issues/12803
    PR-URL: https://github.com/nodejs/node/pull/19758
    Reviewed-By: Ben Noordhuis <[email protected]>
    Reviewed-By: Santiago Gimeno <[email protected]>
    Reviewed-By: James M Snell <[email protected]>

:040000 040000 b5f3cc6178caf6ee1a46ee21fc89c83b2b32a69c c093efd2a818669a8aa2d3641905d6cf2e432c30 M	deps

@benjamingr benjamingr added libuv Issues and PRs related to the libuv dependency or the uv binding. and removed libuv Issues and PRs related to the libuv dependency or the uv binding. labels Jun 9, 2018
@cjihrig
Copy link
Contributor

cjihrig commented Jun 10, 2018

The specific commit in libuv responsible for this is libuv/libuv@8f9ba2a. Specifically, the return -ENOTCONN; when uv_read_start() is called on a stream that is not readable.

This breakage was not intended, but this is arguably a bug in Node (or the application code) as the socket's readable property is false.

@thatshailesh
Copy link
Contributor

thatshailesh commented Jun 21, 2018

Hi All, would like to work on this need some guidance, as per comments i think this hasn't been fixed yet please let me know if this still need to fixed for me to proceed :)

@AyushG3112
Copy link
Contributor

AyushG3112 commented Jun 24, 2018

Same issue exists with v10.5.0 and stdout

process.stdout.on('data', function(data) {
    
});

@AyushG3112
Copy link
Contributor

AyushG3112 commented Jun 24, 2018

Resolving issue is more complicated than it seemed at first glance(atleast to me), but the gist of the issue is is the stdout or stderrs fd handle type corresponds to TTY, PIPE or TCP, the stream is created as a WriteStream instead of a DuplexStream.

Relevant code is at:

function createWritableStdioStream(fd) {
var stream;
const tty_wrap = process.binding('tty_wrap');
// Note stream._type is used for test-module-load-list.js
switch (tty_wrap.guessHandleType(fd)) {
case 'TTY':
var tty = require('tty');
stream = new tty.WriteStream(fd);
stream._type = 'tty';
break;
case 'FILE':
const SyncWriteStream = require('internal/fs/sync_write_stream');
stream = new SyncWriteStream(fd, { autoClose: false });
stream._type = 'fs';
break;
case 'PIPE':
case 'TCP':
var net = require('net');
stream = new net.Socket({
fd: fd,
readable: false,
writable: true
});
stream._type = 'pipe';
break;
default:
// Probably an error on in uv_guess_handle()
throw new ERR_UNKNOWN_STREAM_TYPE();
}

@IamManchanda
Copy link

IamManchanda commented Jun 24, 2018 via email

@gireeshpunathil
Copy link
Member

I guess this is the relevant code, for a top level Node process (stdout / err is a TTY):

node/lib/tty.js

Lines 82 to 86 in 831821b

net.Socket.call(this, {
handle: tty,
readable: false,
writable: true
});

While I think this is a design decision rather than a bug, the ask in the first comment looks like valid requirement to me, at least allowing stdio to be readable - absorbing stdio logs without allowing the write onto the actual device (don't know if it is possible though) can be useful for custom redirections without code modifications.

@mcollina
Copy link
Member

mcollina commented Jul 4, 2018

I'm proposing #21654 as a fix for this to be landed in master and Node 10. Then we might want to change this behavior in Node 11 to produce an explicative error.

mcollina added a commit to mcollina/node that referenced this issue Jul 20, 2018
This change avoid an 'read ENOTCONN' error introduced by libuv 1.20.0
when trying to read from a TTY WriteStream. Instead, we are throwing
a more actionable ERR_TTY_WRITABLE_NOT_READABLE.

Fixes: nodejs#21203
@addaleax
Copy link
Member

I’m reopening this since #22997 should be a good step towards making this work again

@addaleax addaleax reopened this Sep 21, 2018
addaleax added a commit to addaleax/node that referenced this issue Sep 24, 2018
addaleax added a commit to addaleax/node that referenced this issue Sep 24, 2018
Allow reading from stdio streams that are conventionally
associated with process output, since this is only convention.

This involves disabling the oddness around closing stdio
streams. Its purpose is to prevent the file descriptors
0 through 2 from being closed, since doing so can lead
to information leaks when new file descriptors are being
opened; instead, not doing anything seems like a more
reasonable choice.

Fixes: nodejs#21203
addaleax added a commit that referenced this issue Oct 4, 2018
This reverts commit 91eec00.

Refs: #21654
Refs: #21203

PR-URL: #23053
Reviewed-By: James M Snell <[email protected]>
targos pushed a commit that referenced this issue Oct 5, 2018
This reverts commit 91eec00.

Refs: #21654
Refs: #21203

PR-URL: #23053
Reviewed-By: James M Snell <[email protected]>
targos pushed a commit that referenced this issue Oct 5, 2018
Allow reading from stdio streams that are conventionally
associated with process output, since this is only convention.

This involves disabling the oddness around closing stdio
streams. Its purpose is to prevent the file descriptors
0 through 2 from being closed, since doing so can lead
to information leaks when new file descriptors are being
opened; instead, not doing anything seems like a more
reasonable choice.

Fixes: #21203

PR-URL: #23053
Reviewed-By: James M Snell <[email protected]>
targos pushed a commit that referenced this issue Oct 7, 2018
Allow reading from stdio streams that are conventionally
associated with process output, since this is only convention.

This involves disabling the oddness around closing stdio
streams. Its purpose is to prevent the file descriptors
0 through 2 from being closed, since doing so can lead
to information leaks when new file descriptors are being
opened; instead, not doing anything seems like a more
reasonable choice.

Fixes: #21203

PR-URL: #23053
Reviewed-By: James M Snell <[email protected]>
jasnell pushed a commit that referenced this issue Oct 17, 2018
This reverts commit 91eec00.

Refs: #21654
Refs: #21203

PR-URL: #23053
Reviewed-By: James M Snell <[email protected]>
jasnell pushed a commit that referenced this issue Oct 17, 2018
Allow reading from stdio streams that are conventionally
associated with process output, since this is only convention.

This involves disabling the oddness around closing stdio
streams. Its purpose is to prevent the file descriptors
0 through 2 from being closed, since doing so can lead
to information leaks when new file descriptors are being
opened; instead, not doing anything seems like a more
reasonable choice.

Fixes: #21203

PR-URL: #23053
Reviewed-By: James M Snell <[email protected]>
BethGriggs pushed a commit that referenced this issue Mar 19, 2019
Allow reading from stdio streams that are conventionally
associated with process output, since this is only convention.

This involves disabling the oddness around closing stdio
streams. Its purpose is to prevent the file descriptors
0 through 2 from being closed, since doing so can lead
to information leaks when new file descriptors are being
opened; instead, not doing anything seems like a more
reasonable choice.

Fixes: #21203

Backport-PR-URL: #25351
PR-URL: #23053
Reviewed-By: James M Snell <[email protected]>
shaun554 added a commit to shaun554/vuepress that referenced this issue Mar 21, 2019
devs-cloud pushed a commit to devs-cloud/vue-dev that referenced this issue Dec 27, 2019
@ZissisT
Copy link

ZissisT commented Apr 3, 2020

I am a bit confused on what was the solution for this one (fixed or it still throws error but in another way?).
I notice a very similar behavior in v12.16.1
No error was thrown in v8.11.4 but when I try to upgrade to v12.16.1 I get same error.

In my project these lines of code exist for redirecting stderr to syslogger

        process.addListener('uncaughtException', err => {
            // Log this in syslog
            .
            .
        });

        process.stderr.on('data', function(data) {
            // Redirect data to syslogger
        });

This triggers uncaughtException listener always with

Error: read ENOTCONN
at tryReadStart (net.js:567:20)
at Socket._read (net.js:578:5)
at Socket.Readable.read (_stream_readable.js:478:10)
at Socket.read (net.js:618:39)
at resume_ (_stream_readable.js:965:12)
at processTicksAndRejections (internal/process/task_queues.js:84:21)

Whatever the callback is, exception is thrown.
Is this expected? If yes, what causes this to be thrown now and not before (so I know how to change my code)
Thank you

abhishekumar-tyagi pushed a commit to abhishekumar-tyagi/node that referenced this issue May 5, 2024
Allow reading from stdio streams that are conventionally
associated with process output, since this is only convention.

This involves disabling the oddness around closing stdio
streams. Its purpose is to prevent the file descriptors
0 through 2 from being closed, since doing so can lead
to information leaks when new file descriptors are being
opened; instead, not doing anything seems like a more
reasonable choice.

Fixes: nodejs/node#21203

PR-URL: nodejs/node#23053
Reviewed-By: James M Snell <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
confirmed-bug Issues with confirmed bugs. net Issues and PRs related to the net subsystem. process Issues and PRs related to the process subsystem. stream Issues and PRs related to the stream subsystem.
Projects
None yet
Development

Successfully merging a pull request may close this issue.