# What changes on your server after a CPU, RAM or disk upgrade?

https://www.hostimor.com/en/knowledgebase/article/resource-upgrade-cpu-ram-disk

A resource upgrade takes a few clicks in the panel; the real work is confirming that the new resources actually reach the operating system and your applications. This guide covers the upgrade paths, verification commands for CPU, RAM and disk, growing the disk by hand when it does not expand automatically, and application settings.

## Upgrade paths

- **Package upgrade:** You move to a larger package through your service's upgrade option in the client area; the fee is shown on the order screen. You can compare packages on the [VDS](https://www.hostimor.com/en/turkiye-vds-servers) and [VPS](https://www.hostimor.com/en/turkiye-vps-servers) pages.
- **Extra units on flexible servers:** On Flexible VPS and Flexible VDS, you can increase vCPU, RAM and disk separately with extra units.
- **Additional IPv4:** If your IP allowance is used up, the additional IPv4 service can be ordered on some packages; see [IP addresses and additional IPv4](https://www.hostimor.com/en/knowledgebase/article/ip-addresses-additional-ipv4-hostname-dns) for details.

VPS packages use shared vCPUs. If you are deciding between a VDS and a VPS, see [the differences between VDS and VPS](https://www.hostimor.com/en/differences-between-vds-and-vps-a-guide-to-choosing-a-virtual-server).

## Do you really need an upgrade?

Before upgrading, make sure the bottleneck really is a lack of resources. A misconfigured application, a full disk or a runaway process will cause the same problem on a larger package. An upgrade makes sense if the following signs persist for several days:

| Sign | How to check | What it means |
| --- | --- | --- |
| Load average constantly above the vCPU count | `uptime`, `nproc` | CPU may be short; if `wa` is high, the bottleneck is the disk. |
| Swap in constant use | `vmstat 1 5` (si/so), `free -h` | RAM is short; find the process that consumes memory first. |
| Processes killed for lack of memory (OOM) | `journalctl -k` | RAM is short, or a service has the wrong memory limit. |
| Disk above 85% | `df -h` | Clean up first; if usage keeps growing, upgrade the disk. |

```
uptime && nproc                  # load average and vCPU count
vmstat 1 5                       # si/so: swap traffic, wa: disk wait, st: steal
free -h                          # look at the available column
df -h                            # disk usage
journalctl -k --since '-7 days' | grep -i 'out of memory'
```

To dig deeper, see the [resource monitoring](https://www.hostimor.com/en/knowledgebase/article/linux-server-resource-monitoring-steal-time) guide, the [swap and OOM](https://www.hostimor.com/en/knowledgebase/article/create-swap-space-prevent-oom) guide for memory and the [disk full](https://www.hostimor.com/en/knowledgebase/article/disk-full-freeing-up-space) guide for disk space.

## After a CPU and RAM upgrade

The server may be restarted briefly while new CPU and RAM are applied, so schedule the upgrade for a low-traffic time. When the server is back up, confirm that the operating system sees the new resources:

```
nproc
lscpu | grep -E '^CPU\(s\)|Model name|Socket\(s\)|Core\(s\) per socket'
free -h
grep MemTotal /proc/meminfo
# Any offline CPUs?
cat /sys/devices/system/cpu/online
lscpu -e
```

The total memory in the `free -h` output looks slightly lower than what you bought, because memory reserved by the kernel and hardware is subtracted; this is normal. If the new cores do not appear in `nproc`, or `lscpu -e` shows some of them offline, a planned reboot is usually enough.

On Windows, the Performance tab of Task Manager shows the core count and memory. You can get the same information from PowerShell:

```
Get-CimInstance Win32_ComputerSystem | Format-List NumberOfLogicalProcessors, TotalPhysicalMemory
Get-Volume -DriveLetter C
```

## Disk upgrade: automatic expansion and verification

Before a disk upgrade (a package upgrade or an additional disk), delete all snapshots in the **Snapshots** tool; a disk cannot be expanded while snapshots exist ([snapshot guide](https://www.hostimor.com/en/knowledgebase/article/snapshots-restore-points-before-risky-changes)). When the upgrade is applied while the server is on and VMware Tools is running, the partition and the file system are expanded automatically. On Linux this uses growpart together with resize2fs, xfs_growfs or the btrfs tools, and LVM is extended too if the root file system is on LVM; on Windows the C: drive is extended. In some cases the server may be powered off and on briefly to apply the change.

```
lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINT
df -hT /
# In-server record of the automatic expansion
tail -n 30 /var/log/wisecp-task.log
```

`lsblk` shows the new size of the disk and partition, and `df` shows the usable space of the file system; if the two agree, you are done. A virtual disk cannot be shrunk, so increase the disk size step by step as you need it.

## If the disk did not expand: growing it by hand

If the server was off during the upgrade or VMware Tools was not running, the automatic expansion may not have happened. Get Tools running first; a saved change is applied once the server is on and Tools is running. Check again with `lsblk` after a few minutes; if the expansion still has not taken place, follow the steps below.

> ****Warning****
> 
> Working on the partition table carries a risk of data loss. Have a current backup before you start; once the disk has been enlarged, you can also take a new snapshot before the manual steps.

### Standard partition (no LVM)

First make the kernel see the new disk size, then grow the partition and the file system. In the example the disk is `/dev/sda` and the root file system is partition 2; take your own values from the `lsblk` output.

```
# 1) Rescan the disk and check the new size
echo 1 > /sys/class/block/sda/device/rescan
lsblk

# 2) growpart: cloud-guest-utils on Ubuntu/Debian, cloud-utils-growpart on AlmaLinux
apt install cloud-guest-utils        # or: dnf install cloud-utils-growpart
growpart /dev/sda 2                  # note the space between disk and partition number

# 3) Grow the file system (find its type with df -hT /)
resize2fs /dev/sda2                  # ext4
xfs_growfs /                         # XFS: takes the mount point, not the device
btrfs filesystem resize max /        # Btrfs
```

If `growpart` reports "NOCHANGE: partition 2 is size ... it cannot be grown", there is no free space after the partition. This usually happens when a swap or extended partition sits after the root partition. In that layout you have to disable and remove the swap partition, grow the root partition and recreate swap as a swap file; if you are not sure, ask our support team.

### Installations that use LVM

If the root file system is on LVM, the chain has one more link: partition, physical volume (PV), logical volume (LV) and file system. Find the names with `pvs`, `vgs` and `lvs`; on Ubuntu you will usually see `ubuntu-vg/ubuntu-lv`, and on AlmaLinux `almalinux/root`.

```
pvs; vgs; lvs
growpart /dev/sda 3
pvresize /dev/sda3
lvextend -l +100%FREE /dev/ubuntu-vg/ubuntu-lv
resize2fs /dev/ubuntu-vg/ubuntu-lv          # for XFS: xfs_growfs /

# Shortcut: -r grows the file system in the same step
# lvextend -r -l +100%FREE /dev/almalinux/root
```

Ubuntu installations may have unused space left in the volume group. If the VFree column in the `vgs` output is greater than zero, you can use that space right away with `lvextend`, without any upgrade.

### Windows

In Disk Management (`diskmgmt.msc`), choose Rescan Disks from the Action menu, then right-click the C: drive and use Extend Volume. You can do the same in PowerShell run as administrator with these commands:

```
Update-HostStorageCache
Resize-Partition -DriveLetter C -Size (Get-PartitionSupportedSize -DriveLetter C).SizeMax
Get-Volume -DriveLetter C
```

If Extend Volume is grayed out, there may be a recovery partition right after the C: drive; the new space sits behind that partition, so C: cannot be extended. If you do not know the consequences of removing that partition, ask our support team.

## Tune your applications to the new resources

Even when the operating system sees the new resources, applications keep running with their old limits. After an upgrade, these settings make the biggest difference.

**PHP-FPM:** `pm.max_children` is the number of PHP processes that can run at the same time. First measure the average memory use of one process, then divide the memory you set aside for PHP by that value.

```
# Find the version (for example php8.3-fpm.sock)
ls /run/php/
# Average PHP-FPM process memory (on AlmaLinux the process name is php-fpm)
ps --no-headers -o rss -C php-fpm8.3 | awk '{ s += $1; n++ } END { if (n) printf "%.0f MB\n", s/n/1024 }'
```

For example, on a server with 8 GB of RAM, if you reserve 3 GB for the operating system, the database and caches, and measure 60 MB per process, the calculation is: (8192 − 3072) / 60 ≈ 85. Round the value down a little and write it to the pool file:

```
; Ubuntu/Debian: /etc/php/8.3/fpm/pool.d/www.conf   AlmaLinux: /etc/php-fpm.d/www.conf
pm = dynamic
pm.max_children = 80
pm.start_servers = 14
pm.min_spare_servers = 8
pm.max_spare_servers = 20
```

**MariaDB and MySQL:** `innodb_buffer_pool_size` is the area where frequently used data is kept in memory. On a server that runs only the database, 50–70% of RAM is a good starting point; if the web server and the database share the server, 25–40%. Put the setting in your own file; if you leave the distribution's files untouched, package updates will not overwrite your setting.

```
# Ubuntu/Debian: /etc/mysql/mariadb.conf.d/99-tuning.cnf
# AlmaLinux:     /etc/my.cnf.d/99-tuning.cnf
[mysqld]
innodb_buffer_pool_size = 3G
```

```
systemctl reload php8.3-fpm         # AlmaLinux: systemctl reload php-fpm
systemctl restart mariadb
mariadb -e "SELECT @@innodb_buffer_pool_size/1024/1024/1024 AS gb;"
systemctl restart nginx             # worker_processes auto picks up the new core count
```

**Others:** After a RAM increase, review the swap size and the `vm.swappiness` value ([swap guide](https://www.hostimor.com/en/knowledgebase/article/create-swap-space-prevent-oom)). Nginx's `worker_processes auto` setting picks up the new core count after a restart. Fixed memory limits such as `-Xmx` for Java applications, `--max-old-space-size` for Node.js and SQL Server's maximum server memory on Windows must be updated by hand.

> ****Tip****
> 
> After changing settings, watch `free -h`, `vmstat 1 5` and your application's response times for a few days. A setting that consumes all of the new resources only postpones the problem you had before the upgrade.

## Frequently asked questions

### Does the server restart after an upgrade?

A short restart may be needed while CPU and RAM are applied, and for a disk expansion the server may be powered off and on briefly in some cases. Do the upgrade at a low-traffic time.

### The disk has grown, but df still shows the old size.

First check the new disk size with `lsblk`. If the disk has grown but the partition has not, confirm that the snapshots were deleted and VMware Tools is running, then follow the manual expansion steps above.

### Can I shrink the disk?

No, a virtual disk cannot be shrunk. If you are thinking about moving to a package with a smaller disk, ask our support team to go over the options.

## Related articles

- [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)
- [Disk full: finding what uses space and cleaning logs and caches](https://www.hostimor.com/en/knowledgebase/article/disk-full-freeing-up-space)
- [Take a snapshot before risky changes: creating and restoring restore points](https://www.hostimor.com/en/knowledgebase/article/snapshots-restore-points-before-risky-changes)
- [IP addresses, additional IPv4, hostname and DNS resolver settings](https://www.hostimor.com/en/knowledgebase/article/ip-addresses-additional-ipv4-hostname-dns)
