Dell PowerEdge Installation

!! PLEASE NOTE !! These server machines ONLY have VGA adapters. You will need a VGA compatible monitor and cable.

Introduction

This article describes how to install the ArdexaLinux image on Dell PowerEdge Servers, these devices: https://www.dell.com/en-us/shop/storage-servers-and-networking-for-business/sf/poweredge. This article assumes you have extensive experience installing Linux images. If you don't, then do not attempt to undertake the installation.

Hard Drive Setup

These servers may include more than one drive - eg. 1TB SATA HDD + < 512GB SSD. The ArdexaLinux image needs to be installed on the smaller, faster drive. The default ArdexaLinux installer will choose a drive automatically so its best to physically remove all drives, except the one that is being installed. These can be physically re-installed once ArdexaLinux has been installed on the primary drive.

Note that the drive identifiers below may be changed during installation. For example; /dev/sda <- -> /dev/sdb depending on what was booted - ie. booting from USB stick vs a regular system boot post-install - so always check which drive is which via lsblk or fdisk -l prior to using any commands referencing these identifiers.

BIOS configuration

• F2 to enter the bios

  • system BIOS

    • Integrated Devices

      • enable OS watchdog timer

    • System security

      • AC power recovery ON

    • Misc settings

      • disable Prompt on error

  • save and quit

Install ArdexaLinux and agent

• F11 to enter the boot menu or just let the PXE boot fail and then select the option to open the boot menu

• Boot from the Ardexa USB drive (using amd64 Linux image), select "graphical install"

• wait for the install to finish and the machine to reboot

• install the agent

  • Note 1: use Linux 64 bit (x86_64) agent

  • Note 2: the USB may show up as sdb1 (not sda1) when running lsblk

• [OPTIONAL] configure the second drive

  • figure out the drive name using lsblk

  • sudo parted /dev/sdb

  • mklabel gpt

  • mkpart primary ext4 0% 100%

  • quit

  • sudo mkfs -t ext4 /dev/sdb1

  • mkdir /mnt/storage

  • echo '/dev/sdb1 /mnt/storage ext4 defaults 0 0' >> /etc/fstab

    • ^ note: if you have seen any inconsistency in drive identifiers it is probably better to make this configuration use the partition's UUID as /dev/sdb1 may change when the system is rebooted

    • blkid /dev/sdb1 should give you the UUID of the partition

    • echo 'UUID=<partition-uuid> /mnt/storage ext4 defaults 0 0' >> /etc/fstab

    • This method is more reliable, provided you can correctly enter the UUID. If this fails it will be unable to mount the drive on reboot. If this happens, use a live USB to recover the system by mounting the primary filesystem to /mnt and edit /mnt/etc/fstab to the correct UUID

  • reboot

WARNING: When this procedure was tested for the second drive, the machine did not cleanly reboot. Please make sure a manual reboot of the server before starting testing to ensure that the drive is correctly mounting and the machine fully boots without error

AC Power test: from off

• run "poweroff" to shutdown the machine

• wait for the machine to power off and then unplug the power

• wait for about 10s and then plug the power back in

• the machine should start to boot

• NB: it may take up to a minute for the machine to start booting, but there will be a message on the screen explaining that it's running a security/integrity check

AC Power test: from on

• once the machine has booted, unplug the machine while it is still powered on

  • this is relatively safe because we have not written new data to the machine since the last boot

• wait for about 10s and the plug the power back in

• the machine should boot back up

Watchdog test

• Once the machine has booted, using the remote shell, run the following command:

echo c > /proc/sysrq-trigger

This will simulate a kernel panic which will freeze the machine, cut off the watchdog and the system should reboot. There will be a message on the screen saying that the system rebooted due to a watchdog failure, but the system should continue to boot without any interaction. If you need to press a key, the test has failed and we will need to investigate.