Timeout on proxy connect kills script

[jthomerson]

I'll apologize in advance, because I hate submitting this kind of an issue - I can't find where the issue is, and it's totally possible that the issue is not with this library, but I'm not sure where to go next and hoping your expertise can help.

Some details:

• I'm using proxy-agent for the AWS SDK as documented here: http://docs.aws.amazon.com/sdk-for-javascript/v2/developer-guide/node-configuring-proxies.html
• Traffic is being proxied through a Squid proxy.
• Occasionally the proxy will contain a log line like the following, which appears to be a timeout trying to connect: 1503602453.019 59884 10.7.1.114 TCP_MISS/503 0 CONNECT dynamodb.us-east-1.amazonaws.com:443 - HIER_NONE/- -

When the proxy times out trying to connect, my long-running script that's using the AWS SDK just dies. There does not seem to be any error thrown, the SDK doesn't seem to retry, ... nothing. Compare this to if there's an actual error from the AWS API (any API) - I get an error that I can handle and move on. But when one of these proxy timeouts happens, the script dies with no indication of what went wrong.

I've tried the following:

var _ = require('underscore'),
    url = require('url'),
    proxy = require('proxy-agent'),
    awsOpts = {};

// try passing a timeout to the proxy-agent:
// see https://www.npmjs.com/package/agent-base#agentfunction-callback-object-options-→-httpagent
awsOpts.httpOptions = {
   agent: proxy(_.extend(url.parse('http://someproxy:3128'), { timeout: 30000 })),
};

// also try passing timeouts to the AWS options directly:
// see http://docs.aws.amazon.com/AWSJavaScriptSDK/latest/AWS/Config.html#constructor-property
awsOpts.connectTimeout = 30000;
awsOpts.timeout = 45000;

AWS.config.update(awsOpts);

Do you have any ideas on:

1. what might be causing this, or
2. how I could get an event or an error or something that I can log / debug when this happens?, or
3. anything I might be missing?

[lucacorti]

I'm seeing the exact same thing and also have node deployed behind a squid proxy polling S3 for updated files. When internet connectivity is lost the polling fails and I have to restart node to make it work again.

[TooTallNate]

This code in this repository has been moved to the proxy-agents monorepo, so I am closing this pull request. If you feel that this issue still exists as of the latest release, feel free to open a new issue over there.