10-13-2009 09:05 AM
I do have the same problem from what it looks like because I am seeing the same log message even if I do not have any max conn setting configured. The real problem is that the real servers do not seem to recover correctly - they are hitting the max conn value of 10 all the time - they do not get above 10 connections. Am I doing anything wrong? Do I have to configure my own slowstart settings to solve this?
10-13-2009 09:21 AM
It is a pretty simple configuration - doing L4 load balancing only for HTTP and SSL traffic - we are using HTTP and SSL health checks.
I doubt this is important but I thought I should mention it.
10-13-2009 09:24 AM
Sounds like the real server port is not leaving the slow-start mode - I do remember a bug fix in that area. What is the release you are using right now?
Out of the release notes:
Module: WSM6, WSM7, SI-4G and SI-4G-SSL
Symptom: A ServerIron may cause a real server port getting stuck in slow-start mode in certain cases when the real server port's state is changing from the SUSPECT state to the ACTIVE state. One of the cases where a real server port changes from the SUSPECT state to the ACTIVE state is where the master port is in a track group.
The fix is part of 9.5.02r, 10.2.01h and 11.0.00b