Windchill 11.0 M030 CPS10 cluster configuration is not working

[gafoorgk]

Hi,

I created a Windchill 11.0 M030 CPS10 lab as a part of upgrading our current Windchill 10.2 production environment. I'm also willing to move from current monolithic setup to cluster. Following is the brief configuration details.

1. SRV01.lab.com - Oracle Db Server
2. SRV02.lab.com - Windchill App Cluster Node 1
3. SRV03.lab.com - Windchill App Cluster Node 2
4. SRV04.lab.com - Business Reporting and Indexing Server

Everything got installed documentation (Deployment Guide as well as Advanced Deployment Guide for Windchill 11.0) and working fine in individual nodes. Following is a brief of configuration.

• wt.cache.master.slaveHosts=SRV01.lab.com <SRV01_IP> SRV02.lab.com <SRV02_IP>
• wt.rmi.server.hostname=PLM.lab.com
• Hosts file are updated in each Nodes with following entry
  127.0.0.1  PLM.lab.com
• NLB is configured using Windows Load Balancer in the lab. Only port 80 and 443 are allowed through WNLB Cluster group created. (Not sure if this is what causing the issue)

Following are observed as issue. Need help identifying the root cause.

• When Windchill is running in both nodes, browsing works fine from a client computer using http://plm.lab.com
• When Windchill is stopped in SRV02.lab.com, client pc is not able to browse Windchill anymore. Even after closing and opening a new session.
• When HTTP Server is stopped as well in SRV02.lab.com, Windchill couldn't show any page though http://plm.lab.com/ shows "It Works!" page (which means traffic reaches SRV03.lab.com).
• But when Windchill is browsed directly from SRV03.lab.com with url http://srv03.lab.com/Windchill/, it works. All pages are shown as required.

I believe it has something to do with Apache configuration, which is not in documentation or I am missing. Please help

Thank you,

[gafoorgk]

I did R&D on this and came to following conclusion.

• Windows NLB service is not robust enough to check application availability within a URL and accordingly mark the node active/inactive. Windows NLB seems to be checking only heart beat from servers.
• Third party NLB solutions like F5 BIG IP can be configured with much better health monitors to check availability of services in a node, thus directing traffic. This has been successfully configured and mentioned in my community question https://community.ptc.com/t5/Windchill/Configure-F5-BIG-IP-for-Windchill-application-cluster/m-p/562860#M57545