One of the things EVO:RAIL excels at is automating stuff. As you might know EVO:RAIL automates the deployment of VMware ESXi, vCenter, Log Insight as well as carrying out many countless configuration steps that allow for a VMware High Availability (HA), Distributed Resource Scheduler (DRS) and VSAN Cluster to work – all this and more in around 15 minutes flat. However, these individual steps are not widely talked about and where this gets “interesting” is when various big tasks are carried out. Understanding these steps helps demystify what EVO:RAIL is doing, and also helps explain some of the messages you see. For me there are three main big tasks the EVO:RAIL engine carries out:

  1. Configuration of the very first EVO:RAIL Appliance
  2. Adding an Additional Appliance expand available compute and storage resources (commonly referred to in marketing-speak as auto-discovery and auto scale-out).
  3. Replacing a failed node with a new node – commonly caused by failure of a motherboard or CPU socket.

This blog post and its two companion posts (Parts 2/3) attempt to describe in a bit more detail what’s happening in these processes, and why there are subtle differences between them. So let’s start with the first – the steps taken during the configuration of the very first EVO:RAIL appliance.

You can see these steps listed in the EVO:RAIL UI during the Initialize/Build/Configure/Finalize process. It’s perhaps not so thrilling to sit and watch that. Believe it or not I recorded the 15min process so I could rewind slowly and document each one below. I guess I could have asked someone in my team internally for this information, but I suspect they have better things to do!

Screen Shot 2015-06-26 at 10.53.46

So here’s the list….

  1. Set password on vCenter
  2. Install private DNS on vCenter
  3. Configure private DNS on vCenter
  4. Configure vCenter to use private DNS
  5. Perform mDNS ESXi host discovery
  6. Setup management network on ESXi hosts
  7. Configure NTP on ESXi Hosts
  8. Configure Syslog on ESXi hosts
  9. Configure vCenters FQDN
  10. Configure NTP on the vCenter
  11. Configure Syslog on the vCenter
  12. Restart Loudmouth on the vCenter
  13. Accept EULA on vCenter
  14. Create vCenter Database
  15. Initialize SSO
  16. Start vCenter (vpxd)
  17. Create management account on vCenter
  18. Register ESXi hosts with vCenter
  19. Configure FQDN on ESXi hosts
  20. Rename vCenter Server management network on ESXi hosts
  21. Configure NIC Team
  22. Setup Virtual SAN, vSphere vMotion, VM Networks on ESXi hosts
  23. Setup DNS on ESXi hosts
  24. Restart Loudmouth on ESXi hosts
  25. Enable vSphere HA/DRS
  26. Create a Storage Policy
  27. Configuring Enhanced vMotion Compatibility
  28. Set vCenter Log Insight to auto-start after Power Cycle events
  29. Configure root password on ESXi hosts
  30. Register EVO:RAIL Service with mDNS

I don’t have much to say about these steps except to make a couple of remarks. Firstly, they form the bedrock of Parts 2/3 of this blog post series. Most of the EVO:RAIL “big tasks” will do some (but critically not ALL) of these steps. For example, there is no point in deploying vCenter when adding an additional appliance – if it has already been done building the first. It is for this reason adding a additional appliance only takes about 7mins – whereas building the first appliance takes around 15mins.

Secondly, knowing the process can help in troubleshooting. For example notice how the vCenter ‘root’ (and administrator@vsphere.local) password is changed at Step 1, whereas the ‘root’ password on the ESXi host is not changed until Step 29. If there was a problem during the configuration process between these two steps – it would mean the password to login to vCenter would be different from the password to log into the ESXi host(s). Incidentally, this separation of the password change is deliberate. We don’t change root password of the VMware ESXi until the very end and when we can guarantee the appliance build process has been successful.

Conclusions:

In the best of all possible worlds (to quote Voltaire’s Candide for a moment) you shouldn’t have to know about these steps. But a little knowledge is a dangerous thing (to quote Alexander Pope). I could go on flashing off my literary credentials but if I carry on like this you’ll think I’m just a snooty Brit who thinks he knows it all (incidentally, you’d be right!). And let’s face it, no one likes a clever Richard, do they?

Tune into the next thrilling episode for Parts 2/3 where it gets infinitely more interesting.