In this post we are going to go over a common issue with VMware vCenter Server Appliance (VCSA) where when VCSA with High Availability (HA) enabled powers up after being powered off in the incorrect order none of the services will start on any of the VCSA instances.

---

1. SECTION I – The Problem
   1. FIGURE 1 – VCSA Console Screen
2. SECTION II – Solution
3. SECTION III – Footnotes

---

SECTION I – The Problem

The problem that you will commonly see is that when a HA enabled VCSA site powers up after being powered off in the incorrect order none of the services on any of the VCSA instances will start.

With VCSA in a HA configuration there will be 3 VCSA instances:

1. Main VCSA node
2. Passive VCSA node
3. Witness VCSA node

The proper VCSA HA power down procedure requires powering down the nodes in the following order:

1. Passive VCSA node
2. Main (Active) VCSA node
3. Witness VCSA node

This issue occurs when they are not powered down in the correct order which could be because of, for example, an Uninterruptible Power Supply¹ (UPS) failure causing the servers to power down in an unplanned manor.

This happens because VCSA HA is designed in a way that whichever VCSA node isn’t currently active won’t take control and become the active node unless it is specifically told to do so. This means that the issue you are seeing is none of the VCSA instances have been told to take control so they are all sat there waiting to be told what to do.

You can easily identify this issue by connecting to the console screen of each of the VCSA node instances and if they are all displaying the Secure Shell (SSH) login prompt rather than the VCSA console screen like Figure 1 then this may the cause of your issue.

FIGURE 1 – VCSA Console Screen

You can further confirm if the services have started or not by logging into the main active VCSA instance via SSH. Then running the following command:

service-control --status

SECTION II – Solution

A solution for this issue involves making sure that the Passive and Witness nodes are powered off and making sure that the main VCSA node is powered on.

Log into the main VCSA node via SSH using the root credentials and enter the following command to manually remove the HA configuration from the main VCSA node:

vcha-destroy -f

With the HA configuration removed you can run the following command to manually start all the services:

service-control --start --all

NOTE: the services may take a while to fully start up.

Once you can connect to your VCSA instance via its web Graphical User Interface (GUI) you can then remove the two old HA nodes (Passive and Witness). After they are removed you can re-enable the HA configuration for the VCSA instance.

SECTION III – Footnotes

1. UPS – this is a power supply that is fitted with a battery designed to keep a server or any item it is supplying power to powered on in a power failure state. Depending on the load on the UPS will depend on how long the battery will last. It is typically able to hold the items until a backup power source kicks in. ↩︎