CSS Critical Issue

Unanswered Question
Jan 19th, 2009

Hi,

We have two content and i am faicng the problem with the content gorse. Each content has ocm.And the lb is use is Round Robin.

The problem is that only the second service i.e gorse2 gets the 100 percent traffic for few hours and there is nothing much on the gorse 1 service but when we see both the services will be alive.Kindly let me know why is this happining. Why only the service 2 is prefered by the CSS.

service OCM_gorse1

ip address 192.168.1.12

keepalive port 5070

keepalive type tcp

keepalive frequency 10

redundant-index 23

active

service OCM_gorse2

ip address 192.168.1.13

keepalive port 5070

keepalive type tcp

keepalive frequency 10

redundant-index 36

active

content OCM_gorse

protocol udp

port 5060

vip address 10.1.0.2

application sip

add service OCM_gorse1

add service OCM_gorse2

active

thanks in advance.

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Gilles Dufour Mon, 01/19/2009 - 03:56

There is nothing to explain the difference in the config.

However, since this is udp (no connection close messages) if you started one service first from few clients than activated the other one, it is possible that the traffic keeps using the same connections over and over (sip traffic tends to always the src port).

Try to clear all the connections and see what happens.

Gilles.

nehakulsum Mon, 01/19/2009 - 05:09

Hi Giles,

AT some point the gorse2 may get Load problems in peak time, the other point is our HA concept. What happens if gorse2 goes down. Will the css decide to send traffic to gorse1?

Gilles Dufour Tue, 01/20/2009 - 00:06

When a server goes down, active connections are not switched over to the other server.

Only new connections are sent to the alive server.

ACE and CSM comes with the 'failaction purge' command which kills connections to those dead servers so that a new one can be opened to another server.

The CSS has something similar.

CSCek56915 - Enhancement for failaction purge on server failure

But I don't think it would help for udp traffic.

Gilles.

Actions

This Discussion