Afterwards, I went into the file and manually edited two lines. I then copied the entire infrastructure folder which included the vm’s to each node to save time. I made a copy of Configure-Prereqs.ps1 and called it servicefabricalteprereqscript.ps1 and placed it in the infrastructure fold. Yes, there is a way to fake the guids but I haven’t had enough time to test it out and see if it is completely viable.įifth, change the location of the prereqs install path to match wherever you put the generated VM’s This is due to the way that lifecycle services installs a program which generates a unique guid for each of these types. But an Orchestrator node and a non-Orchestrator node can be on the same machine. No two non-Orchestrator nodes can be on the same machine. Important: in my testing, I’ve found that no two Orchestrator nodes can be on the same machine. Generate your scripts first with this powershell:Īfterwards, copy each of the relevant ones to the respective machines. You don’t need to touch this unless you are changing or adding database account permissions/roles during the install.įourth, Generate the Scripts and copy them on each VM in a folder Third, leave DatabaseTopologyDefinition.xml alone This may not be necessary but for logging purposes, I love to see a dedicated domain user for some of these processes. What I did here was replace the Network Service with my own domain user who has permissions to do the querying. Second, let’s change the NodeTopologyDefinition.xml file What you should see here is that the install is pretty easy if you match your vm name up with the name over here. We are going to change it up a little here for our certificates. Ordinarily, you would have the names match the vm names for the self-signed certificates. Notice how I did reduce the management reporter machines and the AOSNodeType. Notice how I didn’t change the name of the VM. Watch out for the DNS names here as it will become big. You need to open up the configtemplate.xml file which should be extracted after you downloaded it from Lifecycle services in some folder called “Infrastructure”. ![]() So, this gave me an interesting experiment plus an answer to my client’s question.įirst, let’s change the configuration file and minimize it for a sandbox install ![]() I had to use my slow configuration because I have another configuration setup for load testing right now that has solid state drives and bunches of ram and processors. It happened with the disk, which wouldn’t be suitable even for a testing environment. Predictably, the settings ran into some extreme latency but what’s so interesting is that the performance bottleneck didn’t happen with the RAM or processors. ![]() Note had to eventually remove this because a previous install was interfering with it so I just left the node alone.Ģ processors, 8 GB ram, sharing 7000 RPM with another AOS2, virtualized with slow type 2 hypervisorĢ processors, 8 GB ram, sharing 7000 RPM with another AOS1 and has a file server, virtualized with slow type 2 hypervisorĢ processors, 8 GB ram, sharing 7000 RPM with Orchestrator 2 and 3, virtualized with slow type 2 hypervisorĢ processors, 8 GB ram, sharing 7000 RPM with Orchestrator 1 and 3, virtualized with slow type 2 hypervisorĢ processors, 8 GB ram, sharing 7000 RPM with Orchestrator 1 and 2, virtualized with slow type 2 hypervisor I capped the sql ram at 4 gb’s for the secondary. Part A: here I show exactly what I needed for this setup:Ĥ processors, 16GB ram, sharing 7000 RPM with secondary, virtualized with slow type 2 hypervisorĢ processors, 8 GB ram, sharing 7000, RPM with primary, virtualized with slow type 2 hypervisor So, I built a configuration and put it to the test just to see how AX would perform if I went minimal. At their size, they might have 20 or 30 non-prod environments and it was important to figure out a few things. This client wanted to test their installation with the good hardware, but they also wanted to know what was the minimum hardware they could use for non-prod environments. What I like about this post is that I will show you how to troubleshoot an install gone wrong when all the steps are there, but it just didn’t work initially. This is the only one that failed even though the steps were the exact same as the other configurations (except for my 3 machine install hack which I didn’t use for an Enterprise install). I’ve tested out 5 different configurations with the install and I’ve gotten 4 of them to work. For one, the install is a lot more stable, and the instructions have really improved. ![]() Predictably, a lot has changed since my day 1 install. I’ve been finishing up on a huge Enterprise D365 on premise install and upgrade this month for an organization, and thought I would write a special post with a configuration that I tested out for them.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |