Through this new articles series, I’ll share with you some field experience, things that were a little bit blurry or for which I had to adapt in order to match with the current situation. Today, let’s have a look to VxRAIL Manager Connectivity and the associated configuration for initial setup.
At initial setup, VxRAIL nodes come with a default configuration and it is the same for VxRAIL Manager VM. When the node election process is terminated, the VxM VM runs on the winner node (host having the lowest PSNT number) and has the default IP address 192.168.10.200/24.
Following DellEMC SolVe procedure, to start the initial setup of a VxRAIL node you must connect a “laptop” to the Top-of-Rack switch and set your TCP/IP configuration in the same subnet as the VxM, no big deal. But, how do you do this if the customer opted for ToR switches with Fiber connection? Or if they prevent you to connect an untrusted external host in the network?
The answer is, use a VM or a workstation already in the appropriate VLAN to go through the wizard. That’s possible, but requires to have something available in an untagged network or want to change the current IP layout of this particular segment, I’d be surprised if the production subnet matches the VxM one, in summary not that much handy…
Indeed, in this case, we must adapt VxM TCP/IP configuration to match customer’s network prior to start with the initial setup.
Let’s first have a look to nodes’ factory network configuration:
As you can see above, with version 4.7.x Private Network has been split. These private networks are actually where the discovery process occurs, it is mandatory for the discovery process to work to make sure the VxM and the hosts can communicate on these VLANs, by default 3939.
If, for any reason, your customer setup does require to modify these VLANs, you can do it before the initial discovery, on each ESXi node, using the following:
[root@node-29822:~] esxcli network vswitch standard portgroup set -p "Private Management Network" -v 5000 [root@node-29822:~] esxcli network vswitch standard portgroup set -p "Private VM Network" -v 5000
Note: ESXi shell must be enabled
At this time, loudmouth can do its job and the node hosting the system VMs will be designated. An easy way to see where it is running if needed:
[root@node-29822:~] esxcli vm process list
The VxRAIL Manager VM should run on a single host only, the master.
We have now our VxM VM running and we must connect to its web interface, to start with the initial setup. This web interface is reachable through the IP address 192.168.10.200, by default, but remember we wanted to change it because our environment couldn’t make it.
The VxRAIL Manager VM comes with two network interfaces, connected to different port groups:
VM Network is actually the External Management network, the one which is connected to customer environment.
If this must to be modified:
[root@node-29822:~] esxcli network vswitch standard portgroup set -p "VM Network" -v 300
Private VM Network, is where the Internal communication with the nodes will occur.
In order to communicate with the VxM on its VM Network interface, we must set an IP address that matches with customer’s network. This can be done, from any of the node of the un-configured cluster. However, the initial election must have been done for the VxM be started on one ESXi.
[root@node-29822:~] vxrail-primary --setup --vxrail-address 10.1.10.200 --vxrail-netmask 255.255.255.0 --vxrail-gateway 10.1.10.1
It can take one or two minutes for the VxM to be reachable on the newly configured IP. When VxM answers to ping, the wizard can initiated. If for some reasons, the web server doesn’t respond, you may have to restart marvin service on the VxM:
Login: mystic Password: vxrmgr:/home/mystic # su - root Password: [root@vxrmgr:~] systemctl restart vmware-marvin
Note: the default password are part of SolVe documentation.
After a couple seconds, the web interface can ben opened:
Very useful resources for connectivity troubleshooting:
Check-out our last posts: