Ultramonkey vs pound?


#1

I know most people use pound, but has anyone tried using ultramonkey as
a load balancing solution between a cluster of mongrels and lighttpd?
Thanks,
Ray


#2

I would highly recommend checking out nginx.

It’s even lighter and faster than lighty, and has a good balancer
built in.


– Tom M., CTO
– Engine Y., Ruby on Rails Hosting
– Reliability, Ease of Use, Scalability
– (866) 518-YARD (9273)


#3

On Wed, 10 Jan 2007 08:50:15 +0100
Raymond O’connor removed_email_address@domain.invalid wrote:

I know most people use pound, but has anyone tried using ultramonkey as
a load balancing solution between a cluster of mongrels and lighttpd?
Thanks,
Ray

There is also pen, http://siag.nu/pen

Dominic


#4

Is there any benefit to using one of these load balancers over just
using lighttpd’s mod_proxy to distribute load to Mongrel processes?


#5

For our setup, the plan was to have the load balancing sw on two
seperate boxes (one being a failover) that will loadbalance among
several lighty/mongrel boxes. That leads to my next question, does
nginx’s load balancing have any failover/monitoring so the other load
balancer can take over? Under my limited knowledge, it seems nginx’s
load balancing is part of the webserver itself, not something I can
easily pull out of the webserver part and stick on its own server.

My boss seems to like ultra monkey for whatever reason, but I couldn’t
find one mention about anyone using it with an RoR setup, so I think we
may go with pound or pen instead. Versus pound/pen is one better at
doing failover/monitoring than the other?

Thanks again,
Ray


#6

Dominic M. wrote:

There is also pen, http://siag.nu/pen

pen absolutely rocks from a performance and reliability perspective.

Takes a bit of work to set up properly though, especially if you want
unified access logs with original client IP preserved.


#7

On Wed, 10 Jan 2007 16:35:08 -0000
“Sheldon Hearn” removed_email_address@domain.invalid wrote:

Dominic M. wrote:

There is also pen, http://siag.nu/pen

pen absolutely rocks from a performance and reliability perspective.

I found it would crash under certain circumstances. In my case
loading a page with several hundreds images linked into it. Of
course, having such a page is not such a great idea anyway!

Takes a bit of work to set up properly though, especially if you want
unified access logs with original client IP preserved.

I patched it to insert an additional HTTP header with the
client IP in the forwarding request as a work around. I was
logging the client IP at the application level rather than
server log though.

If anyone wants the patch E-Mail me and I’ll see if I can find
it.

Dominic