CSS11506-load balancing problems(no service hits)

[Roble Mumin]

I need some help regarding "least connection" load balancing of 4 web servers.

I configured a rule and 4 services which in generally work fine. According to my understanding of least connection, the servers should be used equally or at least in a manner that every service gets a hit once in a while. My Current problem is that only sevice 1/4 gets regular hits. Service 2/4 gets reduced hits, 3/4 very few hits and 4/4 no hits at all.

Does anyone have an explanation for this behavior?

--------------------------

Have a look @ the summary:

CSS11506-1# sh summary

Global Bypass Counters:

No Rule Bypass Count: 1346787

Acl Bypass Count: 0

Owner Content Rules State Services Service Hits

PP1_Owner PP1-Rule Master

PP2_SAS00454 19643

PP2_SAS00455 8369

PP2_SAS00456 745

PP2_SAS00457 0

----------------------------------------

The Content/Owner Rule :

CSS11506-1# sh running-config owner PP1_Owner

!*************************** OWNER ***************************

owner PP1_Owner

content PP1-Rule

vip address <vipaddress1>

redundant-index 10

advanced-balance sticky-srcip

add service PP2_SAS00454

add service PP2_SAS00455

add service PP2_SAS00456

add service PP2_SAS00457

balance leastconn

active

content PP1-Rule2

vip address <vipaddress1>

redundant-index 11

protocol tcp

port 80

url "/"

redirect "//portal.saspf.svc/SAPPortal"

active

---------------------------

The four HTTP Services:

service PP2_SAS00454

ip address <myip1>

redundant-index 54

keepalive uri "/irj/servlet/prt/prtroot/MemoryGame.Default"

keepalive port 8100

keepalive frequency 15

keepalive retryperiod 15

keepalive type http

active

service PP2_SAS00455

ip address <myip2>

redundant-index 55

keepalive uri "/irj/servlet/prt/prtroot/MemoryGame.Default"

keepalive port 8100

keepalive frequency 15

keepalive retryperiod 15

keepalive type http

active

service PP2_SAS00456

ip address <myip3>

redundant-index 56

keepalive uri "/irj/servlet/prt/prtroot/MemoryGame.Default"

keepalive port 8100

keepalive frequency 15

keepalive retryperiod 15

keepalive type http

active

service PP2_SAS00457

ip address <myip4>

redundant-index 57

keepalive uri "/irj/servlet/prt/prtroot/MemoryGame.Default"

keepalive port 8100

keepalive frequency 15

keepalive retryperiod 15

keepalive type http

active

[Gilles Dufour · Accepted solution]

least conns means that any time t, all the services should have the same amount of ACTIVE connections.

But if you look the number of total connections, you do not see the same amount.

This is perfectly normal.

ie: you open 1 connection, it goes to service 1.

You close the connection.

You open a new connection, it will also go to service 1 since all services have 0 active connections.

The total connections for service 1 is therefore 2 and 0 for all the others.

With this simple example you can understand why service 1 will always have much more connections in total than service 2, which will more than service 3,...

If you want the total connections to be same the same, you should use round-robin.

Regards,

Gilles.

[Roble Mumin]

Thanks for the info!

I understand the difference now. I also changed the LB Method to round-robin in the meantime and could see the difference first hand.