# Stop Server, Force Stop or Hard Reset: which power action to use and when

https://www.hostimor.com/en/knowledgebase/article/server-power-actions-stop-force-stop-hard-reset

The client area offers five power actions, and each one treats the operating system differently. This guide explains when each action is safe, how to diagnose a server that does not respond and what to check after an unexpected shutdown.

## The five power actions at a glance

The rule of thumb is simple: first try the action that lets the operating system shut itself down cleanly, and use the hard actions only when the system does not respond.

| Action | What happens to the operating system? | Data risk | When to use it |
| --- | --- | --- | --- |
| **Start Server** | A powered-off server is switched on and boots normally. | None | To turn a server you shut down back on. |
| **Stop Server** | Graceful shutdown through VMware Tools: services stop, data is written to disk and file systems are unmounted cleanly. | Very low | Planned maintenance or a server you will not use for a while. |
| **Restart Server** | If VMware Tools is running, the operating system restarts cleanly; if it is not, the panel performs a hard reset. | Low if the system responds | After updates or configuration changes, while the system is running. |
| **Force Stop** | Like pulling the plug: power is cut without shutting down the operating system. | Unsaved data may be lost | When Stop Server has no effect or Tools is not running. |
| **Hard Reset** | The machine restarts instantly without shutting down the operating system, like pressing the reset button. | Unsaved data may be lost | A kernel panic or a completely frozen system. |

## Why does Stop Server need VMware Tools?

**Stop Server** does not cut the virtual machine's power directly; it sends a shutdown request to VMware Tools inside the server, which starts the operating system's own shutdown process. Services are stopped in order, databases write in-memory data to disk and file systems are unmounted cleanly. If Tools is not running, nothing is there to receive the request; in that case the panel warns you and suggests **Force Stop**.

A shutdown can take anywhere from a few seconds to a few minutes; a large database or a service that is slow to stop makes it longer. Wait a few minutes before you switch to Force Stop, and watch the shutdown progress on the VNC Console. You can confirm from inside the server that Tools is running:

```
# Ubuntu / Debian
systemctl status open-vm-tools --no-pager
# AlmaLinux 9
systemctl status vmtoolsd --no-pager
# Any distribution: Tools version
vmware-toolbox-cmd -v
```

If the service has stopped, start it; if the package was removed, reinstall it:

```
# Ubuntu / Debian
apt install open-vm-tools
systemctl enable --now open-vm-tools

# AlmaLinux 9
dnf install open-vm-tools
systemctl enable --now vmtoolsd
```

On Windows, `Get-Service VMTools` in PowerShell should show **Running**; if it has stopped, start it with `Start-Service VMTools`.

> ****Warning****
> 
> Do not remove or stop open-vm-tools. Besides graceful shutdown, panel actions such as password reset, network settings, additional IPs and disk expansion also depend on VMware Tools.

## Shutting down and rebooting from inside the server

For planned work, the most predictable route is to give the command to the operating system directly. That way you watch the shutdown in your own terminal and can warn logged-in users in advance.

```
# Power off now / reboot now
shutdown -h now
systemctl reboot

# Reboot in 5 minutes and warn logged-in users
shutdown -r +5 "Scheduled maintenance: rebooting in 5 minutes"
# Cancel the scheduled action
shutdown -c
```

The same actions on a Windows server:

```
# Restart / shut down now
shutdown /r /t 0
shutdown /s /t 0
# PowerShell equivalents
Restart-Computer
Stop-Computer
```

> ****Note****
> 
> A server you power off from inside does not come back on by itself; you have to start it from the panel with **Start Server**. Billing continues while a server is powered off. If you no longer need the server, export your data first and then submit your cancellation request from the client area.

## If the server does not respond: check the VNC Console first

Do not reach for **Hard Reset** just because you cannot connect over SSH or RDP. The problem may be limited to the network or a single service, and seeing what is on the screen leads to the right decision. Open the [VNC Console](https://www.hostimor.com/en/knowledgebase/article/access-your-server-with-the-vnc-console) from the panel (it opens while the server is running) and see which case you are in:

- **A login prompt is shown:** The operating system is running; the problem is in the network, the firewall or the SSH service. Log in on the console and fix it; there is no need for a reset.
- **A file system check (fsck) or boot messages are scrolling:** The system is working; let it finish. Resetting now restarts the check and interrupts the repair.
- **The Windows update screen** ("Working on updates"): Updates are being applied; wait, even if it takes a while. An interrupted update can leave the system unable to boot.
- **A kernel panic or a frozen screen that ignores every key:** The system will not recover on its own; use Hard Reset.
- **Repeated "Out of memory" messages:** Memory has run out and the kernel is killing processes. Give it a few minutes to recover and reset it if it does not; then find the root cause with the [swap and OOM prevention](https://www.hostimor.com/en/knowledgebase/article/create-swap-space-prevent-oom) and [resource monitoring](https://www.hostimor.com/en/knowledgebase/article/linux-server-resource-monitoring-steal-time) guides.

## How a forced shutdown affects databases

Force Stop and Hard Reset destroy data that the operating system holds in memory and has not yet written to disk. Journaling file systems such as ext4 and XFS replay their journal at boot and bring the file system structure back to a consistent state. That does not mean the last data written by your application is preserved; it only means the file system itself is not corrupted.

The InnoDB engine of MariaDB and MySQL, as well as PostgreSQL, write transactions to their own logs first (redo log, WAL), so at startup they perform crash recovery and restore committed transactions. Recovery can take minutes on large databases; do not try to restart the service while it runs. Legacy MyISAM tables, on the other hand, can be corrupted and may need repair. Stopping the database before planned work removes these risks.

```
# Stop the database cleanly before planned work
systemctl stop mariadb

# Traces of crash recovery after an unclean shutdown
journalctl -u mariadb -b --no-pager | grep -iE 'crash|recovery|corrupt'

# Check the tables (puts load on large databases)
mariadb-check --all-databases --check
```

If you use MySQL, the service is usually called `mysql` and the check command is `mysqlcheck`.

## A checklist after an unexpected shutdown

After a Force Stop, a Hard Reset or an unexpected shutdown, once the system is back up, look at what happened during the previous boot and what is not working now:

```
# Recorded boots; -1 is the previous boot
journalctl --list-boots

# Last messages and errors of the previous boot
journalctl -b -1 -n 100 --no-pager
journalctl -b -1 -p err --no-pager

# Kernel warnings and errors in the current boot (file system, disk)
dmesg -T --level=err,warn

# Services that failed to start
systemctl --failed
```

If `journalctl -b -1` says it cannot find the previous boot, logs are not kept persistently and the records of that boot are gone. Turn on persistent logging so that you have records next time; with its default setting (Storage=auto), journald writes logs to disk when this directory exists:

```
mkdir -p /var/log/journal
systemctl restart systemd-journald
journalctl --disk-usage
```

On Windows, unexpected shutdowns appear in the System log as events 41 (Kernel-Power) and 6008:

```
Get-WinEvent -FilterHashtable @{LogName='System'; Id=41,6008} -MaxEvents 10 |
  Format-Table TimeCreated, Id, Message -Wrap
```

Once the services are up, check your application too: does the website respond, are queues and scheduled jobs running, and is there a new error in the database log?

## Track actions with the Activity Log

The panel's **Activity Log** lists actions taken from the panel. When you investigate why a server went down, first check here for a Stop Server, Force Stop or restart entry, then compare its time with the output of `journalctl --list-boots`. If no action of yours appears in the panel, look for the cause inside the server: a command, a reboot after automatic updates or a crash.

An unpaid renewal invoice also leads to the server being shut down; see [what happens if a server invoice is overdue](https://www.hostimor.com/en/knowledgebase/article/what-happens-if-a-server-invoice-is-overdue) for the timeline. For a shutdown you cannot explain, open a support ticket with the date and time; the [support ticket guide](https://www.hostimor.com/en/knowledgebase/article/what-information-should-you-include-in-a-support-ticket) lists what to include.

## Frequently asked questions

### I stopped the server from the panel. Does billing stop too?

No, billing continues while the server is powered off. If you no longer need the server, export your data and submit your cancellation request from the client area.

### Stop Server warns that VMware Tools is not running. What should I do?

If the server responds, connect over SSH, run `shutdown -h now` inside the server and repair the Tools service on the next boot. If you cannot connect, check the situation on the VNC Console; if the system is frozen, use Force Stop.

### What is the difference between Restart Server and Hard Reset?

Restart Server is for restarting a running system: when VMware Tools is running it asks the operating system for a clean reboot, and when it is not, it falls back to a reset. Hard Reset restarts the machine instantly without asking the operating system and can cause unsaved data to be lost; use it only when the system does not respond. For planned reboots, `systemctl reboot` inside the server is the most predictable route.

## Related articles

- [Reach your server without SSH or RDP using the VNC Console](https://www.hostimor.com/en/knowledgebase/article/access-your-server-with-the-vnc-console)
- [Take a snapshot before risky changes: creating and restoring restore points](https://www.hostimor.com/en/knowledgebase/article/snapshots-restore-points-before-risky-changes)
- [Monitoring CPU, RAM, disk and network on a Linux server (including steal time)](https://www.hostimor.com/en/knowledgebase/article/linux-server-resource-monitoring-steal-time)
- [Creating swap space and preventing out-of-memory (OOM) errors](https://www.hostimor.com/en/knowledgebase/article/create-swap-space-prevent-oom)
- [What happens if a server invoice is overdue? The stop, cancel and deletion timeline](https://www.hostimor.com/en/knowledgebase/article/what-happens-if-a-server-invoice-is-overdue)
