cdk deploy: fails with cryptic error if notices request times out

[shahinism]

Describe the bug

When I run cdk deploy, it fails with the following error and exit code 1:

Subprocess exited with error null

No extra information is printed out. However, when I use --verbose it raises the following info:

[13:31:21] outdir: cdk.out
[13:31:21] env: {
  CDK_DEFAULT_REGION: 'us-east-1',
  CDK_DEFAULT_ACCOUNT: 'XXXXXXXXXXX',
  CDK_OUTDIR: 'cdk.out',
  CDK_CLI_ASM_VERSION: '32.0.0',
  CDK_CLI_VERSION: '2.87.0'
}
[13:31:24] Could not refresh notices: Error: Request timed out
Subprocess exited with error null
[13:32:38] Could not refresh notices: Error: Request timed out
[13:32:38] Error: Subprocess exited with error null
    at ChildProcess.<anonymous> (/usr/local/lib/node_modules/aws-cdk/lib/index.js:414:50537)
    at ChildProcess.emit (node:events:513:28)
    at ChildProcess.emit (node:domain:489:12)

The same stack is able to deployed to other accounts, but for some unknown reason it fails with some others.

Expected Behavior

Deploy normally, or show a more helpful error message.

Current Behavior

Fails with a not helpful message.

Reproduction Steps

I'm not sure how you can reproduce this, without having this stack.

Possible Solution

No response

Additional Information/Context

No response

CDK CLI Version

2.87.0

Framework Version

No response

Node.js Version

18.16.1

OS

Amazon Linux 2

Language

Python

Language Version

3.9.0

Other information

No response

[peterwoodworth]

Does --notices false unblock?

[shahinism]

No, the error persists, also when --notices false is enabled:

[05:45:47] outdir: cdk.out
[05:45:47] env: {
  CDK_DEFAULT_REGION: 'us-east-1',
  CDK_DEFAULT_ACCOUNT: 'XXXXXXXXXXX',
  CDK_OUTDIR: 'cdk.out',
  CDK_CLI_ASM_VERSION: '32.0.0',
  CDK_CLI_VERSION: '2.87.0'
}
Subprocess exited with error null
[05:46:11] Error: Subprocess exited with error null
    at ChildProcess.<anonymous> (/usr/local/lib/node_modules/aws-cdk/lib/index.js:414:50537)
    at ChildProcess.emit (node:events:513:28)
    at ChildProcess.emit (node:domain:489:12)
    at Process.ChildProcess._handle.onexit (node:internal/child_process:293:12)

The following is the full command I run now:

cdk deploy --verbose --notices false --require-approval never --all

[shahinism]

Well, we managed to resolve this issue by increasing the memory capacity of our CI/CD worker instance from an AWS t3-medium to t3-large. It seems like it was an out-of-memory issue; however, I'm wondering:

1. Why CDK needs more than 4GB of RAM to function.
2. Why is such an error not caught before the process is killed?

[nl-brett-stime]

Try adding an entry in your package.json for the latest version of aws-crt .