[Bug]: Terminal closes immediately after opening it since v4.16.1

[Rucadi]

Is there an existing issue for this?

• I have searched the existing issues

OS/Web Information

• Web Browser: X
• Local OS:
• Remote OS: Centos7 & Ubuntu
• Remote Architecture: x86-64
• code-server --version: anything > 4.16.1

Steps to Reproduce

1. open code-server on the web
2. Open a folder
3. Open a folder again in a new tab seems to help
4. Begin using the terminals, run processses etc...
5. Suddently, when you try to open a terminal, it'll automatically close. For some reason, trying to open 20 terminals or so at the same time makes it so 2-3 survive.
6. killing code-server and opening it again restores functionality

The error that most probably is the culprit:
Error [CodeExpectedError]: Could not find pty on pty host

It also occurs with openvscode-server, and I have the same issue with different computes and people, That's the reason I keep using the older v4.16.1 which does not have this issue.

Expected

The terminal opens.

Actual

The terminal is killed

Logs

Error [CodeExpectedError]: Could not find pty on pty host

Screenshot/Video

No response

Does this issue happen in VS Code or GitHub Codespaces?

• I tested this in native VS Code.
• This does not happen in native VS Code.
• I tested this in GitHub Codespaces.
• This does not happen in GitHub Codespaces.

Are you accessing code-server over a secure context?

• I am using a secure context.

Notes

No response

[code-asher]

In 4.17.0 the minimum glibc version was raised and code-server no longer works in CentOS 7, so that could be why. If you check code-server's logs you might see something about glibc.

[code-asher]

Although, you say it does work if you open enough or restart code-server so that must not be it. I am not sure, in that case. Do you have some way for me to reproduce? A Dockerfile maybe? If not, check code-server's output and see if anything shows up there.

[Rucadi]

Although, you say it does work if you open enough or restart code-server so that must not be it. I am not sure, in that case. Do you have some way for me to reproduce? A Dockerfile maybe? If not, check code-server's output and see if anything shows up there.

The glibc problem was already taken care of by compiling node + some dependencies and no errors related to that are present.

I was wondering if node-pty had a regression, since the error log more plausible is the one I put on the first message... I'll try to create a minimum example