run webserver on all available cores, resolves #6

[missinglink]

@heffergm I can see the webserver bottlenecking at 100% CPU on a single core under load, this PR spreads the load across all available cores, which should result in better parallel processing of HTTP requests.

resolves long standing issue: #6 using a module ebay wrote to get parallelism when using the express webserver module.

$ siege -c500 -r10 "http://localhost:3100/search?input=foo"

# single core
# Response time:                0.54 secs
# Response time:                0.52 secs

# multi core
# Response time:                0.16 secs
# Response time:                0.13 secs

[dianashk]

Good call! +100000

[missinglink]

note this may not be compatible with node 0.12, linking pelias/pelias#35

[heffergm]

We run 0.12.2.

[riordan]

Bravo
On Tue, Jul 14, 2015 at 9:39 AM Grant Heffernan [email protected]
wrote:

We run 0.12.2.

—
Reply to this email directly or view it on GitHub
#158 (comment).

[hkrishna]

👍

[orangejulius]

nice

[missinglink]

upstream PR opened with ebay: cubejs/cluster2#86

[hkrishna]

👍

[missinglink]

💥 now passing against 0.10 and 0.12 @see: https://travis-ci.org/pelias/api/builds/73865581