Quick installation guide
This is a quick installation guide. For a complete installation, refer to the following pages:
In the following examples, the deployment includes 8 identical servers. To simplify the installation commands, Mellanox or Intel E810 NICs are used. 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.
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.
To run the commands on all containers in parallel, we use
pdsh as an example only.- 1.Install WEKA software on all servers:
pdsh -R ssh -w "weka0-[0-7]" "curl https://[GET.WEKA.IO-TOKEN]@get.weka.io/dist/v1/install/4.1.0/4.1.0 | sudo sh"
To get the download link with the token, see the Obtain the Weka software installation package topic.
- 1.Remove the single default container created on each server in the cluster:
pdsh -R ssh -w "weka0-[0-7]" 'weka local stop default && weka local rm -f default'
- 1.Get the resource generator to your local server:
wget
https://raw.githubusercontent.com/weka/tools/master/install/resources_generator.py
- 1.Copy the resource generator from your local server to all servers in the cluster:
for i in {0..7}; do scp resources_generator.py weka0-$i:/tmp/resources_generator.py; done
3. To enable execution, change the mode of the resource generator on all servers in the cluster:
pdsh -R ssh -w "weka0-[0-7]" 'chmod +x /tmp/resources_generator.py'
4. Run resource generator on all servers in the cluster:
pdsh -R ssh -w "weka0-[0-7]" '/tmp/resources_generator.py --path /tmp --net ens{5..7}'
On each server, the resource generator generates three resource files in the
/tmp directory: drives0.json, compute0.json, and frontend0.json.For more details about the resource generator, see the 3. Generate the resource files procedure in the WEKA multi-container backend installation topic.
Create the drive containers from the resource generator output file
drives0.json. Run the following command on all servers in the cluster:pdsh -R ssh -w "weka0-[0-7]" 'weka local setup container --resources-path /tmp/drives0.json'
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 full 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 the weka cluster create command.ssh weka0-1
weka cluster create weka0-{0..7}
weka cluster update --cluster-name=WekaProd
Add a drive to each server in the cluster. Run the following command from one of the servers:
for i in {0..7} ; do weka cluster drive add $i /dev/nvme0n1 /dev/nvme1n1 /dev/nvme2n1 /dev/nvme3n1 /dev/nvme4n1 /dev/nvme5n1 ; done
Create the compute containers from the resource generator output file
compute0.json. Run the following command on all servers in the cluster:pdsh -R ssh -w "weka0-[0-7]" 'weka local setup container --resources-path /tmp/compute0.json --join-ips $(hostname -i)'
Enable event notifications to the cloud for support purposes. From one of the servers, run the following command:
weka cluster update --cluster-name=WekaProd
weka cloud enable
From one of the servers, run the following command:
weka cluster license set LICENSE_TEXT_OBTAINED_FROM_GET_WEKA_IO
Start the cluster IO service. From one of the servers, run the following command:
weka cluster start-io
Create the frontend containers from the resource generator output file
frontend0.json. This step is required for mounting from the server or setting one of the additional protocols on the server, which requires a frontend process. Run the following command on all servers in the cluster: pdsh -R ssh -w "weka0-[0-7]" 'weka local setup container --resources-path /tmp/frontend0.json --join-ips $(hostname -i)'
- 1.Check the resources per server (such as NICs and cores), drives, and configuration status:
do
weka cluster container resources $i
done
weka cluster drive
weka status
Output example for a multiple container architecture:
WekaIO v4.1.0 (CLI build 4.1.0)
cluster: WekaProd (00569cef-5679-4e1d-afe5-7e82748887de)
status: OK (8 backends UP, 6 drives UP)
protection: 6+2
hot spare: 1 failure domains
drive storage: 82.94 TiB total, 82.94 TiB unprovisioned
cloud: connected
license: Unlicensed
io status: STARTED 7 seconds ago (96 io-nodes UP, 750 buckets UP)
link layer: Ethernet
clients: 0 connected
reads: 0 B/s (0 IO/s)
writes: 0 B/s (0 IO/s)
operations: 0 ops/s
alerts: 1 active alert, use `weka alerts` to list it
2. Change the default admin password and ensure no other alerts exist.
Related topics