Deploy the ARTESCA Operating System¶

This page outlines the step-by-step process to install and configure the ARTESCA Operating System (OS) on your servers.

The ARTESCA OS is based on the Rocky Linux distribution and is required for ARTESCA installation. It is maintained as part of ARTESCA deployment and upgrade.

Important

Using a Rocky Linux, or any other OS distribution is not supported, and will fail.
Do not tune, or reconfigure ARTESCA OS except as recommended by Scality.
Do not install any software except as recommended by Scality.

Tip

If you are based in the Asia-Pacific region, you may want to use this link to download the ARTESCA OS package.

To guarantee the best performance and compatibility, install the OS according to the following workflow:

Read the Prerequisites to ensure your server is ready for the OS installation.
Proceed with the Operating System Installation.
Perform the Readiness Checks to ensure the new server is set up correctly.
Continue the ARTESCA deployment with the Next Step.

Prerequisites¶

Hostname¶

Important

To avoid DNS issues you must use flat hostnames such as storage001 and avoid compound hostnames such s001.storage.

Hostnames of servers must conform to the RFC 1123 and RFC 952 requirements. It means they must contain only lowercase alphanumeric characters, '-', and start and end with an alphanumeric character. They must also not be longer than 52 characters.

If your naming scheme does not allow these requirements, you will be able to use alternate names for nodes, when generating the installer configuration.

Additional Drive Formatting¶

During installation, partitioning will run automatically for the target operating system drive. DATA disks, HDDs, and SSDs used for ARTESCA data and metadata are handled directly by the ARTESCA installer at a later stage.

Service disks are set under RAID 1 during the hardware preparation. Refer to ARTESCA Service Drives for more information.

Required Information for Installation¶

Before starting OS installation, gather the following information:

IP address/netmask for control and workload plane interfaces
Default network route
DNS server(s)
Target server hostname
Local timeserver(s)
System timezone
An artesca-os account credentials

Operating System Installation¶

For virtual machines, you need to follow the Virtual Machine Node Preparation section to create the virtual machine with the ARTESCA OS.

Proceed to the standard ARTESCA OS installation using the ARTESCA OS ISO (if you are based in the Asia-Pacific region, you may want to use this link instead).

Boot the server from the installation media, the following screen appears:

Select Install ARTESCA OS.
Press Enter to start the installation.

OS Disk Partitioning¶

Important

If you are using Optimized Boot Drives (BOSS, NS204i, etc.), ensure they are configured using hardware RAID 1 before initiating the installation, if applicable (refer to Hardware Node Preparation / Storage). If not, we recommend letting the installation handle the partitioning as described below.

During the boot, disks are automatically detected and partitioned. If the disk topology does not match the expected one (for example, 1 or 2 disks of 480GB), the installation will ask you which disk(s) to use for the OS installation.

Auto detection

If the disk topology matches the expected one, the installation automatically finds the OS disk(s). You just need to confirm the selection to proceed with the installation.

Manual selection

If the disk topology does not match the expected one, the installation prompts you to select the disk(s) to use for the OS installation.

Select the disk(s) to use for the OS installation and confirm to proceed with the installation.

Warning

In rare cases, the detection fails to find suitable disks for the OS installation or the installation fails due to pre-existing partitions on the disks that cannot be wiped automatically.

In these cases, follow Manual OS Partitioning Recovery.

Installation Summary¶

Once disk partitioning is completed, the installation summary page appears. You need to configure the following options before starting the installation:

Time and date are configured with the right NTP Servers.
Network and host management have bond interfaces for workload plane and control plane.
Keyboard layout should be set according to your site policy.
System users are pre-provisioned.

Note

Partitioning is already done, so you have to leave it as is, even if warnings appear.

Network Interfaces¶

Network interfaces must be configured using the Anaconda GUI following this procedure (depending on your network environment).

Note

In either single or dual network configurations —provided it is supported by your router/switch—, there are no restrictions on the number of VLAN tags on an interface.

Single Network

Bonding (2 ports)

Define the Host Name and click Apply. Ensure it actually updates the Current host name.

Important

To avoid DNS issues you must use flat hostnames such as storage001 and avoid compound hostnames such s001.storage.
Click the + button.
Select the Bond device and click Add.
Set the bond information and click Save.

Note

The 802.3ad (LACP) is the recommended mode but it needs some configuration on your switches. In case you cannot apply those, use the Active backup mode.
Add the bonding slaves by clicking Add, keep the Ethernet connection type, and click Create….
Set the slave information and click Save.
Repeat the two previous steps to add the other bonding slaves.
Ensure you have the list of slaves as expected, then go to IPv4 Settings.
Set the IPv4 address information.

Important

To avoid DNS issues, do not set any search domains.
Set Ignore for IPv6 and click Save.
Ensure the interfaces are properly set up and click Done.

Warning

If you are using 802.3ad (LACP) to have proper aggregation, after the OS installation, you will need to change the bond options using the following commands.

[artesca-os@bootstrap ~]$

sudo nmcli con mod bond-artesca bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca

Bonding + VLAN Tagging (2 ports)

Define the Host Name and click Apply. Ensure it actually updates the Current host name.

Important

To avoid DNS issues you must use flat hostnames such as storage001 and avoid compound hostnames such s001.storage.
Click the + button.
Select the Bond device and click Add.
Set the bond information and click Save.

Note

The 802.3ad (LACP) is the recommended mode but it needs some configuration on your switches. In case you cannot apply those, use the Active backup mode.
Add the bonding slaves by clicking Add, keep the Ethernet connection type, and click Create….
Set the slave information and click Save.
Repeat the two previous steps to add the other bonding slaves.
Ensure you have the list of slaves as expected, then go to IPv4 Settings.
Set Disabled for IPv4 on this bonding interface and click Save.
Set Ignore for IPv6 and click Save.
Click the + button to add the VLAN interface.
Select VLAN and click Add.
Set the VLAN information and click Save..
Set the IPv4 address information and click Save..

Important

To avoid DNS issues, do not set any search domains.
Set Ignore for IPv6 and click Save.
Ensure the interface are properly set up and click Done.

Warning

If you are using 802.3ad (LACP) to have proper aggregation, after the OS installation, you will need to change the bond options using the following commands.

[artesca-os@bootstrap ~]$

sudo nmcli con mod bond-artesca bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca

Dual Networks

Bonding (2 x 2 ports)

Define the Host Name and click Apply. Ensure it actually updates the Current host name.

Important

To avoid DNS issues you must use flat hostnames such as storage001 and avoid compound hostnames such s001.storage.
Click the + button.
Select the Bond device and click Add.
Set the bond information and click Save.

Note

The 802.3ad (LACP) is the recommended mode but it needs some configuration on your switches. In case you cannot apply those, use the Active backup mode.
Add the bonding slaves by clicking Add, keep the Ethernet connection type, and click Create….
Set the slave information and click Save.
Repeat the two previous steps to add the other bonding slaves.
Ensure you have the list of slaves as expected, then go to IPv4 Settings.
Set the IPv4 address information.

Note

You cannot have 2 gateways. Ensure to put the gateway only on the interface that needs it.

Either you set it here or on the Control Plane one.

Important

To avoid DNS issues, do not set any search domains.
Set Ignore for IPv6 and click Save.
Click the + button to add the second bonding interface.
Select the Bond device and click Add.
Set the bond information and click Save.

Note

The 802.3ad (LACP) is the recommended mode but it needs some configuration on your switches. In case you cannot apply those, use the Active backup mode.
Add the bonding slaves by clicking Add, keep the Ethernet connection type, and click Create….
Set the slave information and click Save.
Repeat the two previous steps to add the other bonding slaves.
Ensure you have the list of slaves as expected, then go to IPv4 Settings.
Set the IPv4 address information.

Note

You cannot have 2 gateways. Ensure to put the gateway only on the interface that needs it.

Either you set it here or on the Workload Plane one.

Important

To avoid DNS issues, do not set any search domains.
Set Ignore for IPv6 and click Save.
Ensure the interfaces are properly set up and click Done.

Warning

If you are using 802.3ad (LACP) to have proper aggregation, after the OS installation, you will need to change the bond options using the following commands.

[artesca-os@bootstrap ~]$

sudo nmcli con mod bond-artesca-wp bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca-wp

sudo nmcli con mod bond-artesca-cp bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca-cp

Bonding + VLAN Tagging (2 ports)

Define the Host Name and click Apply. Ensure it actually updates the Current host name.

Important

To avoid DNS issues you must use flat hostnames such as storage001 and avoid compound hostnames such s001.storage.
Click the + button.
Select the Bond device and click Add.
Set the bond information and click Save.

Note

The 802.3ad (LACP) is the recommended mode but it needs some configuration on your switches. In case you cannot apply those, use the Active backup mode.
Add the bonding slaves by clicking Add, keep the Ethernet connection type, and click Create….
Set the slave information and click Save.
Repeat the two previous steps to add the other bonding slaves.
Ensure you have the list of slaves as expected, then go to IPv4 Settings.
Set Disabled for IPv4 on this bonding interface and click Save.
Set Ignore for IPv6 and click Save.
Click the + button to add the VLAN interface.
Select VLAN and click Add.
Set the VLAN information and click Save..
Set the IPv4 address information and click Save..

Note

You cannot have 2 gateways. Ensure to put the gateway only on the interface that needs it.

Either you set it here or on the Control Plane one.

Important

To avoid DNS issues, do not set any search domains.
Set Ignore for IPv6 and click Save.
Click the + button to add the second VLAN interface.
Select VLAN and click Add.
Set the VLAN information and click Save..
Set the IPv4 address information and click Save..

Note

You cannot have 2 gateways. Ensure to put the gateway only on the interface that needs it.

Either you set it here or on the Workload Plane one.

Important

To avoid DNS issues, do not set any search domains.
Set Ignore for IPv6 and click Save.
Ensure the interface are properly set up and click Done.

Warning

If you are using 802.3ad (LACP) to have proper aggregation, after the OS installation, you will need to change the bond options using the following commands.

[artesca-os@bootstrap ~]$

sudo nmcli con mod bond-artesca bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca

Time and Date - NTP Settings¶

Important

Best practice is to have at least three NTP servers (3 lines beginning with server). Or be synchronized with a pool of multiple NTP servers. Refer to Change NTP Configuration for more detailed information.

By clicking TIME & DATE, you are redirected to the following page:

Configure the settings as follows:

Using the Region and City fields, adapt the timezone configuration according to your site policy.

Note

If case of doubt, ask your IT department for the proper timezone.
Click the settings button next to the Network Time switch to access the NTP servers list.
1. Add the NTP servers you want to use by writing their addresses and clicking the + button.
2. Remove the servers you do not want to use by unchecking the box in the Use column.
3. Ensure to have at least one NTP server with the green flag in the Working column.
  
  Note
  
  If none of the servers are working, check the network configuration or contact your IT department.
4. Click OK to save the settings.
Toggle the Network Time switch to ON to start synchronizing the time with the NTP servers.

Note

If the Network Time switch returns to the off position, check the network configuration.
Click Done to leave the TIME & DATE page.

Launch OS Installation¶

Click Begin Installation to start the installation of the Operating System.

Note

The installation is only available once all the previous required options have been set up successfully.
Once the Installation is completed, reboot the server by selecting Reboot System.

Secure ARTESCA OS and Root Credentials¶

The ARTESCA OS provides two users with a respective default and insecure password:

An artesca-os user with the default password scality.
A root user with the default password scalityadmin.

Warning

To ensure the security of your ARTESCA OS, you must log in with these users at least once to update the password.

Refer to Change Password for more information on how to change your passwords.

Once ARTESCA is installed, you cannot SSH as the root user. To log in as the root user, connect directly to the server (physically or using an IPMI console). You must use the root user only on Scality support requests.

Readiness Checks¶

Disk Partitions¶

Check for current detected disk partitions and volumes using the lsblk command:

Partitioning with hardware RAID 1 setup

Partitioning with software RAID 1 setup

This command lists OS partitions and volumes. Wipe existing partitions on any additional data drives before proceeding.

Network Bondings¶

Once the network interface is configured, use the following command to list the bonding interfaces for the workload and control planes, as well as their sub-interfaces.

Check that the bonding interfaces are up and running. The output of the following command must show the bonding mode, the status of the bonding and slaves interfaces, and the matching speed and duplex of the slaves interfaces. You can also check the status of bond-artesca-cp.

Warning

If you are using 802.3ad (LACP) for proper aggregation, after the OS installation you must change the bond options. Select the tab that aligns with your network configuration, and execute the commands.

Single Network

Bonding (2 ports)

[artesca-os@bootstrap ~]$

sudo nmcli con mod bond-artesca bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca

Bonding + VLAN Tagging (2 ports)

[artesca-os@bootstrap ~]$

sudo nmcli con mod bond-artesca bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca

Dual Networks

Bonding (2 x 2 ports)

[artesca-os@bootstrap ~]$

sudo nmcli con mod bond-artesca-wp bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca-wp

sudo nmcli con mod bond-artesca-cp bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca-cp

Bonding + VLAN Tagging (2 ports)

[artesca-os@bootstrap ~]$

sudo nmcli con mod bond-artesca bond.options mode=802.3ad,miimon=100,lacp_rate=1,xmit_hash_policy=layer3+4
sudo nmcli con up bond-artesca

Service Disk Preparation¶

Identify the device path for each wanted disk.

Some ARTESCA components require persistent storage. This storage is provided in a single LVM physical volume, which must be pointing at a RAID 1 array for durability.

Important

The service disk must be configured as a SW RAID1 array following the instructions below.

Software RAID Array Creation¶

Important

You must follow these steps on each ARTESCA server.

Raid Configuration with 2 Devices

Fill the devices to use for the RAID array. Replace with your actual device.
$
MDDISK1=/dev/nvme0n1 MDDISK2=/dev/nvme1n1

Ensure to stop the /dev/md/artesca RAID array.

sudo test -e /dev/md/artesca && sudo mdadm --stop /dev/md/artesca

Wipe the devices to remove any existing partition table.

In general, the command below is sufficient. If your devices have different paths, ensure to replace them in the command.
$
sudo wipefs -fa $MDDISK1 $MDDISK2

Initialize the RAID 1 array.

sudo mdadm --create --verbose /dev/md/artesca \
    --level=1 --raid-devices=2 --metadata=1.2 \
    $MDDISK1 $MDDISK2

mdadm: size set to 104791040K
mdadm: array /dev/md/artesca started.

Check the array creation and progress.

cat /proc/mdstat

Personalities : [raid1]
md127 : active raid1 nvme1n1[1] nvme0n1[0]
      104791040 blocks super 1.2 [2/2] [UU]
[=====>...............]  resync = 26.8% (28111552/104791040) finish=27.1min speed=46998K/sec

unused devices: <none>

Raid Configuration with 4 Devices

Fill the devices to use for the RAID array. Replace with your actual device.

MDDISK1=/dev/nvme0n1
MDDISK2=/dev/nvme1n1
MDDISK3=/dev/nvme2n1
MDDISK4=/dev/nvme3n1

Ensure to stop the /dev/md/artesca RAID array.

sudo test -e /dev/md/artesca && sudo mdadm --stop /dev/md/artesca

Wipe the devices to remove any existing partition table.

In general, the command below is sufficient. If your devices have different paths, ensure to replace them in the command.
$
sudo wipefs -fa $MDDISK1 $MDDISK2 $MDDISK3 $MDDISK4

Initialize the RAID 10 array.

sudo mdadm --create --verbose /dev/md/artesca \
    --level=10 --raid-devices=4 --metadata=1.2 \
    $MDDISK1 $MDDISK2 $MDDISK3 $MDDISK4

mdadm: size set to 104791040K
mdadm: array /dev/md/artesca started.

Check the array creation and progress.

cat /proc/mdstat

Personalities : [raid10]
md127 : active raid10 nvme3n1[3] nvme2n1[2] nvme1n1[1] nvme0n1[0]
      104791040 blocks super 1.2 [4/4] [UUUU]
[=====>...............]  resync = 26.8% (28111552/104791040) finish=27.1min speed=46998K/sec

unused devices: <none>

SSH Configuration¶

When deploying a multi-node cluster, each machine must be accessible using SSH from the Bootstrap node (the node used to run the Installer). SSH access must rely on an authorized passwordless SSH key as password-based SSH access is not supported.

When generating the configuration file, the Installer needs a valid SSH key, which needs to be authorized on other machines.

Tip

You can also manually configure SSH access using an already existing SSH key, but ensure the private key used is stored under the /etc/metalk8s/pki/ directory.

Follow this procedure to generate a new SSH key that will be used by the ARTESCA Installer.

Prepare the MetalK8s PKI directory.

[artesca-os@bootstrap ~]$

sudo mkdir -p /etc/metalk8s/pki

Generate the passwordless SSH key.
[artesca-os@bootstrap ~]$
sudo ssh-keygen -t rsa -b 4096 -N '' -f /etc/metalk8s/pki/salt-bootstrap
Warning

Although the key name is not critical, it is the default one used during the installer configuration generation.
Upload the public key to other machines.
[artesca-os@bootstrap ~]$
ssh-copy-id -f -i /etc/metalk8s/pki/salt-bootstrap.pub artesca-os@<node_ip>
This command adds the contents of the local /etc/metalk8s/pki/salt-bootstrap.pub public key to the /home/artesca-os/.ssh/authorized_keys file on the remote machine. Repeat this step for each machine in the cluster.

Next Step¶

You have deployed the ARTESCA Operating System. You can now proceed with Deploy the ARTESCA Software Components.