vCAC Website + ManagerService: Same server with different ports - is that even possible?

[Henrique_Cicuto]

Hello again everyone,

As per my first post  about a week ago (Building a Distributed vCAC Installation with vCNS - Not quite sure on how to proceed), I started building a distributed vCAC configuration. Since the problems described within that post were "solved", I decided to open another topic (since now I have a complete different question).

The "vCAC 6.0 - Installation and Configuration" guide describes the following distributed infrastructure:

According to this design, if I´d want to build a high availability scenario, I´d need at least:

- 1x (v)PostgreSQL cluster;

- 1x MS SQL cluster;

- 3x VIPs (1 for the appliances; 1 for the Iaas Web; 1 for the Manager Service);

- 2x vCAC appliances;

- 2x Windows machines for the vCAC Web and Model Manager Data;

- 2x Windows machines for the vCAC Manager Service;

- Mixed amount of machines for DEMs and Agents.

I´m asking this because I´m still trying to make sense why it´s possible to install Web and Manager Service together (same machine) in a simple installation and not when in a distributed installation (for example, making each listen on different HTTPS ports).

Since I have almost zero experience with configuring or maintaining web servers I´m probably asking something which the answer is pretty obvious, but if anyone knows why or anyway to do it, I´d be very thankful. 🙂

Best regards

[gsaliensjr]

I am also setting up a distributed intall using a virtual netscaler. Maybe what I am attaching to my post can help.

[stvkpln]

I understand what you're saying, but the way to approach this sort of thing -- if you want to install the Model Manager website and Manager Service on the same box -- would be to install the Model Manager and Manager Service during the same installation. That way, the IIS instance gets properly configured. You should be able to select them together.

[GrantOrchardVMw]

+1 to Steve's response.

Grant

[Henrique_Cicuto]

I think it´s a good idea to explain from the beginning:

I started with the following, a few weeks ago:

- 2x vCAC Appliances and 1x PostgreSQL: Cluster configuration and LB HTTPS 443 worked as a charm using vCNS;

- 1x Windows Server 2008 R2: Web Sites, Model Manager, Manager Service (Active), DEM Worker/Orchestrator, vSphere Proxy Agent;

- 1x Windows Server 2008 R2: Web Sites, Manager Service (Backup), DEM Worker/Orchestrator, vSphere Proxy Agent.

Load balancing using vCNS for the Web Sites was an easy thing. The nightmare began with load balancing the Manager Service. Since Web Sites and Manager Service were installed at the same time, they´re both binded to TCP 443. But the Web Sites are "active/active" while Manager Service are "active/passive". That way, I needed something more than just "monitoring HTTPS 443" that vCNS does as default because every now and then I´d get an HTTP 404 error when something was trying to access the Manager Service (meaning that it was being redirected to the backup - offline - Manager Service):

1 - Okay, what if I try the "HTTP URI" within the "Health Check" at vCNS? The sad thing is, apparently, as the label states, it´s "HTTP URI", not HTTPS. Each time I´d try to fill in anything within that field vCNS Edge would just ignore and erase it. I tried turning HTTP 80 on. The configuration stays, but it seens to not apply to HTTPS (still getting those 404);

2 - Okay, what if I try to install both Web Sites and Manager Services separately BUT at the same Windows machine? That way I can specify a different HTTPS port for Manager Services and then load balancing would work just fine. When I tried doing so, the installer would stop with an error message stating that the "Application Pool vCAC already exists". So it´s install them both or no go;

3 - Okay, then I´ll just create two more virtual machines. Web Sites within the first two. Manager Service (active and backup) within the other two. Created a new VIP, a new pool/virtual server configuration within vCNS using those two new machines. After installing only the Manager Service I noticed that they keep returning HTTP error 404. Re-reading the installation manual I see that I also must install the Web Sites together with the Manager Services, since they provide the basis for the latest.

But procedure number 3 brings me back to the original problem. I´m starting to think that perhaps I´m just too tired to understand what is wrong and perhaps I should just assume a full "active/passive" design for the Iaas components, including setting Microsoft IIS as a manual service within the second server.

[stvkpln]

I'm no vCNS guru, but if I was trying to do what you were doing, what I would do is setup a VIP for the Website service and a VIP for the Manager service. On the VIP for the Website, add all of your servers that support the IaaS website. On the VIP for the Manager service, just add the server that is the active Manager service. When you setup your DEM Orchestrator / Workers and Proxy Agent(s), point to each VIP independently. That should, conceptually, solve the issue.