====== System requirements ======

  * 1 GHz or faster 32 bits (x86) or 64 bits (x64) processor
  * 1 Gio memory (minimum) - For heavily loaded servers or high flow is recommended at least 8 Gio
  * 1 Gio of available hard disk space for i-MSCP and managed services, excluding user data
  * Internet access (at least 100 Mbits/s recommended)
  * A Linux kernel >= 2.6.26
  * A file system supporting extended attributes such as ext2, ext3, ext4 or reiserfs*****
  * Appropriate privileges to create devices (**CAP_MKNOD** capability)
  * Appropriate privileges to mount, unmount and remount filesystems (**CAP_SYS_ADMIN** capability)

===== Reiserfs users =====

In order, to use the reiserfs file system with i-MSCP, you must follow these steps:

Edit your **/etc/fstab** file to add the **attrs** option for your device (e.g. device containing the /var partition). For
instance:
    UUID=74699091-3ab8-43f2-bdd5-d1d898ab50fd /     reiserfs notail          0    1
should be updated to:
    UUID=74699091-3ab8-43f2-bdd5-d1d898ab50fd /     reiserfs notail,attrs    0    1
Once you did that, you can remount your device. For instance:
    mount -o remount /dev/disk/by-uuid/74699091-3ab8-43f2-bdd5-d1d898ab50fd
If needed, you can find the uuid of your device, with the following command:
    blkid <device>
where **<device>** must be replaced by your device path such as **/dev/sda1**

===== LXC containers =====

If you want install i-MSCP inside a LXC container, the following conditions have to be met:

  * You must have the **CAP_MKNOD** capability inside the container. Thus, you must ensure that **mknod** is not in the list of dropped capabilities
  * You must have the **CAP_SYS_ADMIN** capability inside the container (required to mount filesystems). Thus, you must ensure that `sys_admin` is not in the list of dropped capabilities.
  * You must allow the creation of devices inside the container by white-listing them. Easy solution is to add **lxc.cgroup.devices.allow = a *:* rwm** in LXC container configuration file.
  * If you use **Apparmor**, you must allow **mount**,**umount** and **remount** operations inside your container by modifying the default apparmor profile **/etc/apparmor.d/lxc/lxc-default** or by creating a specific apparmor profile for the container.

Note that these operations must be done on the host, not in the container.

**See also:**
  * https://i-mscp.net/index.php/Thread/14039-i-MSCP-inside-a-LXC-container-Managed-by-Proxmox-4-x
  * https://linuxcontainers.org/fr/lxc/manpages/man5/lxc.container.conf.5.html
  * https://help.ubuntu.com/lts/serverguide/lxc.html#lxc-apparmor
  * http://wiki.apparmor.net/index.php/AppArmor_Core_Policy_Reference#Mount_rules_.28AppArmor_2.8_and_later.29

===== OpenVZ containers (Proxmox and Virtuozzo) =====

You could have to increase the **fs.ve-mount-nr** limit, else, an error such as **mount: Cannot allocate memory** could be threw by CageFS. To avoid this problem you must:

  - Increase the limit by adding an entry such as **fs.ve-mount-nr = 4096** to your **/etc/sysctl.conf** file
  - Make the new limit effective by executing the **sysctl -p** command

Note that these operations must be done on the host, not in the container.