windows 2025 container exec problems

[mirage51]

Description

Hi,
I want to exec in a running container but when i get inside, i don't have cursor to move or to give a command. This is not reproduced in windows2022.

Steps to reproduce the issue

1. nerdctl run image
2. nerdctl exec -it container powershell or cmd or bash

Describe the results you received and expected

read-only terminal

What version of nerdctl are you using?

2.0.0 and 2.0.3

Are you using a variant of nerdctl? (e.g., Rancher Desktop)

None

Host information

No response

[TinaMor]

@mirage51

1. Are you using the mcr.microsoft.com/windows/nanoserver:ltsc2025 image or the mcr.microsoft.com/windows/servercore:ltsc2025 image?
2. Please add the results from nerdctl info in What version of nerdctl are you using?

[mirage51]

Hi,
mcr.microsoft.com/windows/servercore:ltsc2025 image
The result is a normal terminal in a container but i can't do anything in it. i don't have the control there. Also, i think is important to say: If I do nerdctl exec -it container dir c:\windows or anything else, is working. but i don't have the control on terminal

[TinaMor]

@mirage51 Thank you for the update. Will investigate.

To help with repro-ing the bug in a similar env to yours, kindly add the results from running nerdctl info on your device in the What version of nerdctl are you using? section of the bug description.

[mirage51]

Client:
Namespace: default
Debug Mode: false

Server:
Server Version: v1.7.23
Storage Driver: windows
Logging Driver: json-file
Plugins:
Log: fluentd journald json-file none syslog
Storage: windows-lcow windows
Kernel Version: 10.0 26100 (26100.1.amd64fre.ge_release.240331-1435)
Operating System: Microsoft Windows Server Version 24H2 (OS Build 26100.3194)
OSType: windows
Architecture: x86_64
CPUs: 16
Total Memory: 31.3GiB

Nerdctl version 2.0.0 but 2.0.3 is behaving in the same manner

[TinaMor]

Filed the issue on Windows Containers. Windows-Containers/issues/575. Please upvote it there.

[mirage51]

thanks!

[apostasie]

@TinaMor I recently got #4054 merged in (because of unrelated double CR in linux output linked to the use of OPOST).

This just a wild hunch, but curious if you could test again against main and see if that makes any difference with regard to this specific issue here.

Cheers.