Evening everyone,
Its been around 3 days that I´ve been building this distributed vCAC installation using vCNS as a load balancer. Now that I finally managed to install every component, I´m facing some strange behavior when configuring vCAC throught the Web portal. Here are the specs:
- 1x PostgreSQL 9.3.4 running on a Suse Linux Enteprise 11 64-bits;
- 2x vCAC Appliances 6.0.1;
- 1x vCNS 5.5.2;
- 2x Edge Load Balancers (HA);
- 1x vCenter Single Sign-On (vCenter Server 5.5 Update 1);
- 1x SQL Server 2008 R2 (Iaas DB);
- 2x Windows Server 2008 R2 (2x Websites; 1x ModelManager; 2x ManagerService; 2x DEM Orchestrator; 2x DEM Workers; 2x vSphere Agents);
- Everything was built following the Installation and Configuration Guide (vCloud Automation Center Installation and Configuration)
And here´s how things were done:
1 - Downloading, installing, compiling, configuring SLES 11 and PostgreSQL - OK
2 - Deployed vShield Manager from vCNS - OK
3 - Deployed Edge Appliance with HA: The Appliance has 3 addresses: 1 for itself; 1 for the vCAC Portal; 1 for the Iaas - OK
4 - Configured Edge Pools and Virtual Servers: Now that is one which I´m not 100% with. I didn´t exactly find anything with instructions on how to load balance vCAC components, but here´s what I´ve done:
- Created a Pool for the Web Portal: Services - HTTP (80; LEAST_CONN) and HTTPS (443; LEAST_CONN); Health Check - HTTP (80; HTTP; default values) and HTTPS (443; SSL; default values). Did not specify a URI (since I´ve no idea if there is one); Members - Added both vCAC Appliances (HTTP - 80; and HTTPS - 443);
- Created a Pool for the Iaas machines: Copied the configuration from above, changing the address to both Windows vCAC Iaas Servers;
- Created a Virtual Server for the Web Portal: HTTP (80; COOKIE; JSESSIONID; INSERT) and HTTPS (443; SSL_SESSION_ID). Set Portal external IP and Web Portal Pool;
- Created a Virtual Server for the Iaas: Copied the configuration from above. Set Iaas external IP and Iaas Pool;
5 - Deployed 2 vCAC Appliances. Configured for distributed installation. Disabled internal vPostgre and Orchestrator. All services up - OK
6 - Deployed 2 Windows 2008 machines - OK
7 - Installed 2 Websites (1 per machine - both running), 1 ModelManager (first machine), 2 ManagerServices (1 active, 1 standby), 2 DEM Orchestrators (1 per machine - both running), 2 DEM Workers (1 per machine - both running) and 2 vSphere Agents (1 per machine - both running).
A few notes:
For every time the installer requested the vCAC appliance, ModelManager or ManagerService hostname, I´d always input the external FQDN (that resolve to the Edge IP addresses), as written in the manual. The "Test" button would always "Pass" each time.
Now the issues I´m facing 😞
1 - I successfully accessed the vCAC web portal by pointing to the Edge address. I successfully managed to log in. But, during the initial tenant configuration (branding, identity store, mail servers) the interface would reload or simply "stuck";
2 - Now I´ve logged with an Infrastructure admin and the interface states I don´t have a DEM Orchestrator (although I have two running). Exact message:
"There is no Distributed Execution Manager (DEM) running as an Orchestrator. Many actions such as provisioning a machine will not complete until a DEM-Orchestrator is started."
I´ll start looking over some more, playing a little with the Edge balancing methods and do some more logging but I´d like to know if anyone has ever tried a configuration similar to this one and maybe could share some notes or ideas.
Thank you very much.
