ApiDeploy fails with NetworkingError: write EPROTO

[jancona]

My stack failed with the following error and the rollback failed with the same messages. The rollback succeeded on the second retry.

This stack has worked in the past and other stacks using the same custom resource lambda are working, so I assume the problem is intermittent and probably similar to https://github.com/aws/aws-sdk-js/issues/862. If so, then I suspect that using nodejs instead of nodejs4.3 is a workaround.

2016-12-21T21:53:01.163Z	d7905b33-c7c7-11e6-8f0a-afab7bcb4d7b	Loading module ApiDeploy
2016-12-21T21:53:01.638Z	d7905b33-c7c7-11e6-8f0a-afab7bcb4d7b	Error ApiDeployService::createDeployment { error: { [NetworkingError: write EPROTO] message: 'write EPROTO', code: 'NetworkingError', errno: 'EPROTO', syscall: 'write', address: undefined, region: 'us-east-1', hostname: 'apigateway.us-east-1.amazonaws.com', retryable: true, time: Wed Dec 21 2016 21:53:01 GMT+0000 (UTC) }, params: { stageName: 'stage', restApiId: 'bXXXXXXa', description: 'Version 1.0.211' } }
2016-12-21T21:53:01.719Z	d7905b33-c7c7-11e6-8f0a-afab7bcb4d7b	Response body:
{
    "Status": "FAILED",
    "Reason": "NetworkingError: write EPROTO. See the details in CloudWatch Log Stream: 2016/12/21/[$LATEST]d4a2938abe1c42ed9ea26d77433a3f1f",
    "PhysicalResourceId": "bXXXXXXXa/stage",
    "StackId": "arn:aws:cloudformation:us-east-1:5XXXXXXXXXX4:stack/appconnect-restapi-stack-stage-stage/c9cb0ad0-c627-11e6-8e04-50fae9826c35",
    "RequestId": "06a6256d-acc5-4cdb-bcfa-49e726bb3889",
    "LogicalResourceId": "ApiDeploy",
    "Data": {
        "error": "NetworkingError: write EPROTO"
    }
}

[carlnordenfelt]

Sounds like you are correct. Looking through that thread, one possible solution for nodejs4.3 would be to add the keepAlive setting on the sdk client though.

Let me know if this is problem persists. I won't take any action on it now as it doesn't appear to be directly connected to this project and, as you mentioned, looks like an intermittent issue.

/ Carl