Cold, Warm and Hard Reboots with Virtual Instances (Vultr Example)

“Please keep in mind restarting your VPS instance through the portal is essential for making necessary adjustments to its definition on the hardware host node. This step becomes imperative when tasks such as unblocking SMTP, adding extra IPs, or engaging in similar activities.
It’s important to note that restarting via the portal initiates a cold reboot, akin to turning off and on a computer. In contrast, restarting via the operating system only reboots the OS itself. To address the current issue, we kindly request that you perform a restart of this VPS instance via your control panel at http://my.vultr.com.

Introduction

Someone was having issues getting SMTP port 25 opened on Vultr, the block was lifted by Vultr after putting in a support ticket and they rebooted as per Vultr’s suggestion. However, the reboot was a warm reboot and not a cold boot or full instance reset.

Response

I mean you can google it. Cold, Warm and Hard boots are what’s available, there are sometimes different names for them.

Cold boot is the process of starting a computer from a powered-down, or off state. Cold boot is also called as hard boot. Warm boot refers to restarting a computer that is already turned on via the operating system. Restarting it returns the computer to its initial state.

So typically when dealing with bare-metal servers, you have these options through what is called IPMI and out-of-band management system. You’ll have multiple options to restart the system, allowing even a fourth option of a complete system reset. For instance, if a piece of hardware is malfunctioning, an OS reboot aka a warm boot might not reset the device. You have to do a power reset, where power is completely turned off to the electronics and discharges them. The IPMI module stays online however and the power supplies are still running throughout all of this, but the system BIOS is reset and system checks occur.

Then there are instances where need to do a complete power down, unplugging the power supplies and pressing the power button to discharge capacitors.

This is the same situation on a Vultr instance, a Vultr instance is a virtual machine instance controlled by a hypervisor (the controller of the VM). A warm reboot is a reboot through the OS, which is what some SaaS control panel providers use (GridPane, Runcloud, Cloudways). However, a cold boot or cold reset is required via the Vultr control panel as it sends a command to the Hypervisor to reset the instance.

I think providers can use the Vultr API to reset the instance and perform a cold boot / cold reset. Perhaps that’s what they’re doing, but based on what happened it’s not what they’re doing. It would require that SaaS control panel providers (GridPane, Runcloud, Cloudways) shut down the instance gracefully and then reset the instance. As you don’t want to do a cold reset on an instance as data corruption can occur.