# Tuning Mongrel w/ httperf

**URL:** <https://www.ruby-forum.com/t/tuning-mongrel-w-httperf/109787>\
**Category:** Deployment\
**Created:** [July 19, 2007, 2:33am UTC](https://www.ruby-forum.com/t/tuning-mongrel-w-httperf/109787 "2007-07-19T02:33:00Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![ddancer](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/ecb155/32.png) [@ddancer](https://www.ruby-forum.com/u/ddancer)\
**Post date:** [July 19, 2007, 2:33am UTC](https://www.ruby-forum.com/t/tuning-mongrel-w-httperf/109787/1 "2007-07-19T02:33:00Z")

</div>

Hello Friends,

I’ve recently watched the peepcode on httperf and read various blog  
articles about performance testing web servers w/ httperf.

My problem is that my mongrel apps all seem to be running just  
horribly slow. But since I’m the lone railser in my company I have no  
idea if this slowness is correct or I’m doing blatantly wrong with my  
server setups.

My Test Setup:  
I’m testing by running httperf on a linux box with nothing else of  
significance on the box. Running httperf is it’s sole purpose, for  
now. The app is running on a VM with 512mb mem, not sure of the cpu  
but it’s the machine is less than a year old. I’m running Apache 2.2  
proxying to 5 mongrels. I even installed the eventmachine gem but it  
didn’t seem to do anything.

The app is entirely behind ssl. I’ve created a special “hello\_world”  
action that has sessions disabled, all it does is render(:text =\>  
‘hello world’).

Unlike many others I’ve read about my best testing indicates about 10  
replies/sec. Which just seems way too slow.

Any thoughts/ideas would be greatly appreciated.

A sample of my httperf output:

andy@hst-25-134:~$ httperf --server intranet3 --ssl --uri /public/  
hello\_world --hog --num-conns 100 --rate 15  
httperf --hog --client=0/1 --server=intranet3 --port=443 --uri=/public/  
hello\_world --rate=15 --send-buffer=4096 --recv-buffer=16384 --ssl –  
num-conns=100 --num-calls=1  
Maximum connect burst length: 1

Total: connections 100 requests 100 replies 100 test-duration 37.394 s

Connection rate: 2.7 conn/s (373.9 ms/conn, \<=100 concurrent  
connections)  
Connection time [ms]: min 9622.4 avg 24891.7 max 36704.4 median  
27602.5 stddev 8464.1  
Connection time [ms]: connect 2606.8  
Connection length [replies/conn]: 1.000

Request rate: 2.7 req/s (373.9 ms/req)  
Request size [B]: 78.0

Reply rate [replies/s]: min 0.0 avg 2.1 max 6.2 stddev 2.3 (7 samples)  
Reply time [ms]: response 22283.5 transfer 1.4  
Reply size [B]: header 226.0 content 11.0 footer 0.0 (total 237.0)  
Reply status: 1xx=0 2xx=100 3xx=0 4xx=0 5xx=0

CPU time [s]: user 3.95 system 33.44 (user 10.6% system 89.4% total  
100.0%)  
Net I/O: 0.8 KB/s (0.0\*10^6 bps)

Errors: total 0 client-timo 0 socket-timo 0 connrefused 0 connreset 0  
Errors: fd-unavail 0 addrunavail 0 ftab-full 0 other 0

Thanks,

Andy K.

---

<div class="post-metadata">

**Author:** ![ddancer](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/ecb155/32.png) [@ddancer](https://www.ruby-forum.com/u/ddancer)\
**Post date:** [July 20, 2007, 12:17am UTC](https://www.ruby-forum.com/t/tuning-mongrel-w-httperf/109787/2 "2007-07-20T00:17:11Z")

</div>

I haven’t read your full post yet Andy (I’m on my way out the door to  
work), but 5 mongrels, 512mb of ram… you must be pretty close to  
choking the server down completely (swapping).