Quick installation guide
This is a quick installation guide. For a complete installation, refer to the following pages:
Weka installation on bare metal
Prerequisites
For a complete prerequisite list, refer to the Prerequisites for Installation section.
In the following examples, the deployment includes 8 identical servers (named weka01
to weka08
).
This quick installation exemplifies two architecture types:
Multiple containers architecture: an architecture with 8 servers. Each container is configured with only one process type: drive, compute, or frontend. See Weka containers architecture overview.
Single container architecture: an architecture with 8 identical servers. Each container is configured with the drive, compute, and frontend processes.
In the following examples, we use Mellanox or Intel E810 NICs to simplify the installation commands. For example, only a single IP for data is required. No need to expose VFs, identification of the interface netmask, and default routing gateway.
If the architecture is different, the installation commands are slightly different.
Once the WEKA system is installed, learn how to view, manage and operate it using either the GUI or the CLI, and perform the first IO to a WEKA filesystem.
The WEKA system supports a RESTful API for automating the interaction with the Weka system, integrating it into your workflows, and monitoring systems.
Quick installation for a multi-container backend architecture
It is assumed that the servers are ready for the WEKA software installation. In this example, there are 8 servers. Each server has over 20 cores, 6 NVME drives, and a single Mellanox NIC.
To run the commands on all containers in parallel, we use pdsh
as an example only.
Install the WEKA software
Install WEKA software on all servers:
To get the download link with the token, see the Obtain the Weka software installation package topic.
Remove the default container
Remove the single default container created on each server in the cluster:
Generate resource files
Get the resource generator to your local server:
Copy the resource generator from your local server to all servers in the cluster:
3. To enable execution, change the mode of the resource generator on all servers in the cluster:
4. Run resource generator on all servers in the cluster:
The resource generator generates three resource files on each server in the /tmp
directory: drives0.json
, compute0.json
, and frontend0.json
.
Configuration
Create the drive containers
Create the drive containers from the resource generator output file
drives0.json
. Run the following command on all servers in the cluster:
Create a cluster
Connect to one of the servers, create the cluster, and set the cluster name. Using bash, you can provide a compact list of containers cluster-name{n..m}. Otherwise, specify the complete list of all containers. For example:
weka cluster create weka0-0 weka0-1 weka0-2 weka0-3 weka0-7
. It is assumed that the DNS is set up. Otherwise, specify the explicit IPs in theweka cluster create
command.
Add a drive to the cluster
Add a drive to each server in the cluster. Run the following command from one of the servers:
Create compute containers
Create the compute containers from the resource generator output file
compute0.json
. Run the following command on all servers in the cluster:
Name the cluster and enable event notifications to the cloud
Enable event notifications to the cloud for support purposes. From one of the servers, run the following command:
Set the license
Obtain a classic or PAYG license from get.weka.io.
Set the license. From one of the servers, run the following command:
Start the cluster IO service
Start the cluster IO service. From one of the servers, run the following command:
Create frontend containers
Create the frontend containers from the resource generator output file
frontend0.json
. This step is required to mount from the server or set one of the additional protocols on the server, which requires a frontend process. Run the following command on all servers in the cluster:
Post configuration
Check the cluster configuration
Check the resources per server (such as NICs and cores), drives, and configuration status:
Output example for a multiple container architecture:
2. Change the default admin password and ensure no other alerts exist.
Quick installation for a single container architecture
It is assumed that the servers are ready for the Weka software installation. In the following example, there are 8 servers. Each server has over 20 cores, 6 NVME drives, and a single Mellanox NIC.
Install the Weka software
Install Weka software on all containers:
To get the download link with the token, see the Obtain the Weka software installation package topic.
Configuration
Connect to one of the servers and run the commands below to:
Create the cluster and set the cluster name, stripe width, and protection scheme. Using bash, you can provide a compact list of containers
cluster-name{n..m}
. Otherwise, specify the full list of all containers. For example:weka cluster create weka01 weka02 weka03 weka04 weka05 weka06 weka07 weka08
. It is assumed that the DNS is set up. Otherwise, specify the explicit IPs in theweka cluster create
command.Enable event notifications to the cloud for support purposes.
Configure the network, drives, and CPUs per container:
2. Configure the network NICs, drives, and cores per container. Replace network, drives, and cores configuration with your actual data.
Post configuration
Check the resources per container (such as NICs and cores), drives, and configuration status:
2. If the configuration status is according to your needs, apply the configuration:
3. Verify that the apply
operation is successful. Check the alerts and verify that the
ResourcesNotAppliedalert
alert does not show.
4. Obtain a classic or PAYG license from get.weka.io and set the license.
5. Start the cluster:
6. Check the cluster information and status:
Output example:
7. Change the default admin password and ensure no other alerts exist.
Related topics
Weka installation on bare metal
Manage the system using the Weka CLI
Last updated