Request failed with status code 400

[tdmalone]

I consistently get the following error a couple of seconds after attempting to start a session:

• My region is set to ap-southeast-2 (I tried with us-east-1 too just in case)
• I have set a profile that is authenticated via MFA, and has a session token present (the profile has admin privileges and works with the AWS CLI)
• I have the Session Manager plugin installed, and successfully create Session Manager sessions regularly

Where would you suggest looking next to work out how to resolve this error?

[iann0036]

I saw this a couple of times but couldn't pin it down. Could you try restarting the environment from the console?

[ex-rookie]

From the "Console" tab of "Help -> Toggle Developer Tools" window in VSCode:

...
{status: 400, statusText: "Bad Request", headers: {…}, config: {…}, request:
...
data:
code: "InvalidStatusException"
message: "Expected the environment status to be in SUSPENDED."
...

Restarted VSCode and tried again, the Session went to RUNNING state this time(from CONNECTING state during the 400 Error above), but it was failed with the below error and got removed from the session list:

...
The terminal process failed to launch: A native exception occurred during launch (File not found: ).
...

[iann0036]

Hey @ex-rookie,

Thanks for helping out!

Are you able to open your Terminal/CMD and simply run session-manager-plugin --version? That error implies the Session Manager plugin is not available in your $PATH.

[iann0036]

I've also published 0.1.5 to the extension marketplace that adds a bunch of debugging.

[ex-rookie]

Thanks Ian, the session-manager-plugin does seem to show up correctly for both cmd and the wsl terminals.

$ session-manager-plugin --version
1.2.30.0

I am thinking this is more related to the terminal launch itself (when done by the extension, and could be a config issue with my environment).

Side note, also had this error when trying to upgrade to 0.1.5:
"ERR Corrupt ZIP: end of central directory record signature not found: validating: Corrupt ZIP: end of central directory record signature not found"

Will read more in the VSCode docs to see which of those hundreds of settings can be tuned better, Windows version (conpty, winpty etc etc), probably not related to this 400 issue, or any issue with the plugin, will keep posted if I find anything new.

[iann0036]

Thanks for that. I'm on a Mac so can't really replicate what you're seeing.

Let me know if you find anything.

[tblakex01]

Hi Ian,

I'm experiencing the same issue, although after updating to 0.1.5 and rebooting I'm not getting the 404 error anymore. The cloudshell shows "connected" but I receive the following error; "The terminal process failed to launch: A native exception occurred during launch (File not found: )". I have a few shells setup already, such as git bash, wsl2, powershell, running on Windows 10 latest version of vscode and the extension. Does the AWS CloudShell get created automatically or do I need to adjust or check any of the terminal.integrated.* settings?

Thanks

[iann0036]

Hey @tblakex01,

The extension attempts to directly create a terminal using session-manager-plugin as the main process, as per this line.

I'm wondering if this is not a supported action on Windows? My assumption is that the "File not found:" relates to the location of session-manager-plugin? i.e. changing that to C:\\%PROGRAMFILES%\\Amazon\\SessionManagerPlugin\\bin\\session-manager-plugin.exe (or similar) might help

[ex-rookie]

yes something here apparently, but will need some trial-and-error:
https://code.visualstudio.com/docs/supporting/troubleshoot-terminal-launch

[tblakex01]

Thanks for the quick reply. I was able to make some progress. If I use the insiders edition of vscode setup with remote wsl and install the extension in wsl, essentially ubuntu, then it works perfectly. This would probably also work in the regular version of vscode, I just needed two versions up to do some testing. Using the tracing on vscode it looks like it tries to grab my local windows env variables, instead of the cloud shell's "cwd" when it loads then after that the error hits that the terminal failed to load and file not found. I'll do some more testing tomorrow and see if I can figure out a way for it to spawn in Windows natively, but I'm perfectly content with it working with WSL. Thanks for creating this extension and former2/console recorder, they are awesome!

[ex-rookie]

Thanks @tblakex01,

I do have WSL set as well as the default shell (but how do we install the extension in WSL?),

With the environment, there is a flag "terminal.integrated.inheritEnv" that is not support in Windows, so setting it to False has helped some users solve their terminal launch issues in Windows (but did not work for me).

This is the latest I got:

Since you have similar platform, probably you have the correct flow to see if that "undefined" is really expected.

[iann0036]

Yep, the undefined is expected (logged the wrong thing), so I believe these are the same issue.

For @tblakex01 , if we can figure out a consistent way of spawning the shell regardless of cwd/shell, then we can test for Windows OS and use whatever execution command is best.

[ex-rookie]

Yes thanks @tblakex01, works with "Remote WSL connection to Local copy of WSL(or with Remote SSH connection to any Linux machine)" with the regular VSCode version as well. A standalone testing of the "createTerminal" API may help to further troubleshoot the Windows part.