request: Request with interval() eventually produces ESOCKETTIMEDOUT and ECONNRESET

I have a persistent process that goes off and crawls urls every X minutes.

node v0.10.33
request 2.44.0
ubuntu 14.04 on EC2
uname -a
Linux ip-myip 3.13.0-37-generic #64-Ubuntu SMP Mon Sep 22 21:28:38 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux

I get into this state periodically where all my requests get ESOCKETTIMEDOUT and ECONNRESET errors. Also when it gets into this state, my tcp connections for node process is stuck at only 13 established (these are my mongodb connections):

lsof -p 9514 | grep TCP | wc -l
13

If I restart node things go back to normal (for a while). The lsof command goes up and down as expected, until issue is hit again.

This is the block of code that is called via interval()

var separateReqPool = {maxSockets: 100},
      reqObj = {
            pool: separateReqPool,
            uri: feed.feedUrl,
            headers: {"User-Agent" : "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_6_8) AppleWebKit/537.13+ (KHTML, like Gecko) Version/5.1.7 Safari/534.57.2"},
            timeout: 5000,
            jar: false,
            maxRedirects: 5,
            strictSSL: false
      };

    request(reqObj, function(error){
        if(error){
            cfg.logger.error("ERR: bad request: "+error, error);
        }
    })
        .on('response', function(response){
            ...
        })
        .on('error', function(error){
            cfg.logger.error("ERR: "+error, error);
        })
        .on('end', function(){
            cfg.logger.info("feed pull complete - "+feed.feedUrl);            
        });

Any ideas? thanks in advance.