WEKA legacy system installation process
This page describes the stages in the installation process for the WEKA legacy system (single container architecture) using the CLI.
Workflow
1. Install the WEKA software on each server
Run the untar command and install.sh command, according to the instructions, on each server.
On completion of this stage in the installation process, the WEKA software is installed on all the allocated servers and running in the stem mode. That is, no cluster is attached and the WEKA system is awaiting instructions.
Note: If a failure occurs during this installation stage, an error message detailing the source of the failure is received. If possible, try to recover this error or alternatively, contact the Customer Success Team.
Stage 2: Formation of a cluster from the containers
2. Create a cluster from the containers
Command: weka cluster create
This stage involves the formation of a cluster from containers. It is performed using the following command line:
weka cluster create <hostnames> [--host-ips <ips | ip+ip+ip+ip>]
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| Space-separated strings | Hostnames or IP addresses | Need at least 6 strings, as this is the minimal cluster size | Yes | |
| Comma-separated IP addresses | IP addresses of the management interfaces. Use a list of | The same number of values as in | No | IP of the first network device of the container |
Note: It is possible to use either a hostname or an IP address; this string serves as the identifier of the container in subsequent commands.
Note: If a hostname is used, make sure that the hostname to IP resolution mechanism is reliable since a failure of this mechanism will cause a loss of service in the cluster. It is recommended to add the hostnames to /etc/hosts.
Note: Once the command is completed successfully, the cluster is in the initialization phase, and some commands can only run in this phase.
Note: For configuring HA, at least two cards must be defined for each container.
On successful completion of the formation of the cluster, every container receives an ID. To display a list of the containers and their IDs, run the command line weka cluster container.
Note: In IB installations the --hosts-ips parameter must specify the IP addresses of the IPoIB interfaces.
3. Set a name for the cluster (optional)
Command: weka cluster update
This command is used to give the cluster a name. Although this is optional, it is highly recommended, because the name enables cloud event notification and increases the ability of the Weka Support Team to resolve any issues that may occur. To perform this operation, use the following command line:
weka cluster update --cluster-name=<cluster-name>
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| String | Identifier of the cluster name | Must be a valid identifier | No |
4. Enable event notifications to the cloud (optional)
Enable support through Weka Home
Command: weka cloud enable
This command enables cloud event notification (via Weka Home), which increases the ability of the Weka Support Team to resolve any issues that may occur.
To learn more about this and how to enable cloud event notification, refer to Weka Support Cloud.
Enable support through a private instance of Weka Home
In closed environments, such as dark sites and private VPCs, it is possible to install a private instance of Weka Home.
Command: weka cloud enable --cloud-url=http://<weka-home-ip>:<weka-home-port>
This command enables the use of a private instance of Weka Home.
For more information, refer to Private Instance of Weka Home and contact the Customer Success Team.
Stage 5: Set containers dedicated to the cluster (optional)
5. Set containers dedicated to the cluster (optional)
Command: weka cluster container dedicate
It is possible to set the container as dedicated to the Weka cluster. By setting the container to dedicated, no other application is expected to run on it, and the Weka system optimizes it for performance and stability. For example, the container can be rebooted by the system at need, and all the container's memory is allocatable by the Weka processes.
6. Configure the networking
Command: weka cluster container net add
When PKEYs are used, the device name for InfiniBand should follow the name.PKEY convention.
Note: Although devices can generally be renamed arbitrarily, Weka only functions correctly if the .PKEY naming convention is followed.
The networking type can be either Ethernet (direct over DPDK) or InfiniBand (IB), and can be mixed in the same container (by running multiple cluster container net add commands for the same container). A physical network device must be specified for both types. This can be a device dedicated to the Weka system or a device also being used for other purposes in parallel. For IP over DPDK, the standard routing parameters can be specified for routed networks.
To perform this operation, the cluster container net add command must be run for each container. The commands can run from one container configuring another and run on a single container. The IP addresses specified using this command are the data plane IPs allocated in the planning stage. To perform this operation, use the following command line:
weka cluster container net add <container-id> <device> [--ips-type=<POOL|USER>] [--ips=<ips>]... [--gateway=<gateway>] [--netmask=<netmask>] [--label=<label>]
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| String | Identifier of the container to which a network interface will be added | Must be a valid container identifier | Yes | |
| String | A device, or bond-device e.g., | Must be a valid Unix network device name | Yes | |
| String | POOL or USER | Must be one of the two options | No | POOL |
| Comma-separated IP address | The data plane IP addresses for internal Weka system traffic. In IB, use the IPoIB address | Must be part of the data plane IP pool defined in the planning phase. See Weka Networking and Networking Prerequisites. | No | From Pool |
| Number | Number of bits in the netmask | Describes the number of bits that identify a network ID (also known as CIDR). Not relevant for IB / L2 non-routable networks, and must be supplied for the ethernet NICs if the cluster is set to use both ethernet and IB interfaces. | No | |
| IP address | The IP address of the default routing gateway | The gateway must reside within the same IP network of Not relevant for IB / L2 non-routable networks. | No | |
| String | A label to describe the network device connectivity. | The Weka system will prefer to use paths with the same labels to send data. This is useful when the system is configured with HA networking, to hint the system to send between containers through the same switch rather than using the ISL. | No |
The number of IP addresses should be according to Weka Networking and Networking Prerequisites.
Note: Additional IP addresses may be assigned for each container if IP per core is needed. In this case, unused IP addresses are reserved for future expansions and can be automatically assigned if the number of cores assigned to the Weka system on that container is increased.
Note: For HA configurations, this command has to be run separately for each interface.
Optional: Configure default data networking
Command: weka cluster default-net set
Instead of explicit IP address configuration per each network device, dynamic IP address allocation is supported. Weka supports adding a range of IP addresses to a dynamic pool, from which the IP addresses can be automatically allocated on demand.
For Ethernet networking only, a mixed approach is supported: for certain network devices the IP addresses are assigned explicitly by the administrator, while the other devices in the cluster get an automatic allocation from the IP range. Such an approach could be useful in an environment where clients are being spawned automatically.
weka cluster default-net set --range <range> [--gateway=<gateway>] [--netmask-bits=<netmask-bits>]
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| IP address range | A range of IP addresses that can be used for dynamic allocation across the whole cluster | Format: A.B.C.D-E E.g., 10.10.0.1-100 | Yes | |
| Number | Number of bits in the netmask | Describes the number of bits that identify a network ID (also known as CIDR). | Yes | |
| IP address | The IP address of the default routing gateway | The gateway must reside within the same IP network of IPs in Not relevant for IB / L2 non-routable networks. | No |
To view the current default data networking settings use the command weka cluster default-net.
If a default data networking was previously configured on a cluster and is no longer needed, it is possible to remove it using the command weka cluster default-net reset.
7. Configure the SSDs
Command: weka cluster drive add
This stage in the installation process is used to add a local SSD to be used by a Weka filesystem. The same command can be used for adding multiple drive paths. To perform this operation, use the following command line:
weka cluster drive add <container-id> <device-paths>
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| String | Identifier of the container to which a local SSD will be added | Must be a valid container identifier | Yes | |
| Space-separated list of strings | List of block devices that identify local SSDs, e.g., | Must be a valid Unix network device name | Yes |
Note: If, due to some technical limitation, the use of an NVMe device through the kernel is required, contact the Customer Success Team.
8. Configure the CPU resources
Command: weka cluster container cores
This stage in the installation process is used to configure the number of CPU resources, which are physical rather than logical cores. To perform this operation, use the following command line:
weka cluster container cores <container-id> <cores> [--frontend-dedicated-cores <frontend-dedicated-cores>] [--drives-dedicated-cores <drives-dedicated-cores>] [--cores-ids <cores-ids>] [--compute-dedicated-cores <compute-dedicated-cores>] [--only-drives-cores] [--only-compute-cores] [--only-frontend-cores]
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| String | Identifier of the container in which a core count is configured | Must be a valid container identifier | Yes | |
| Number | Number of physical cores allocated to the Weka system | Must be less than the number of physical cores in the container (leaving 1 core for the OS). Maximum 19 cores | Yes | |
| Number | Number of physical cores dedicated to FrontEnd processes | The total of frontend, drives, and compute cores cannot exceed the | No | zero |
| Number | Number of physical cores dedicated to Drive/SSD processes | The total of frontend, drives, and compute cores cannot exceed the | No | Typically 1 core per drive or 1/2 core per drive/SSD |
| Number | Number of physical cores dedicated to compute processes | The total of frontend, drives, and compute cores cannot exceed the | No | |
| A comma-separated list of numbers | Physical Core numbers | Specification of which cores to use. | No | Select cores automatically |
| Boolean | Determines whether all cores in the container are dedicated only to drive processes | No | ||
| Boolean | Determines whether all cores in the container are dedicated only to compute processes | No | ||
| Boolean | Determines whether all cores in the container are dedicated only to FrontEnd processes | No |
Note: cores-ids are distributed in the following order: first, all the FrontEnd processes. Second, all the Compute processes. And last, all the Drive processes. By ordering the cores-ids list, it is possible to determine the exact assignment of cores to processes (e.g., for taking into account NUMA distribution).
Example: If we have 1 FrontEnd, 2 Compute, and 3 Drive, setting cores-ids to 1, 2, 4, 3, 5, 6 will put the FrontEnd on core 1, Compute on cores 2 and 4, and Drive on cores 3, 5 and 6. Assuming cores 1, 2, 3 are at NUMA 0 and cores 4, 5, 6 are at NUMA 1, we will have the following distribution of processes:
NUMA 0: FrontEnd, Compute, Drive
NUMA 1: Compute, Drive, Drive
Note: Performance can be optimized by assigning different functions to the various Weka cores. If necessary, contact the Customer Success Team. for more information.
Note: Weka supports more than 19 cores per server. This option enables taking advantage of all server core resources. To set the system with more than 19 cores, contact the Customer Success Team.
9. Configure the memory (optional)
Command: weka cluster container memory
As defined in the memory requirements, the fixed memory per container and the per compute/SSD cores memory are automatically calculated by the Weka system. By default, 1.4 GB is allocated per compute core, of which 0.4 GB is left for capacity-oriented memory. If the container is set as dedicated, all the memory left after reductions, as described in Memory Resource Planning, is automatically allocated for the Weka system.
If capacity requirements mandate more memory, the following command should be used:
weka cluster container memory <container-id> <capacity-memory>
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| String | Identifier of the container in which the memory configuration has to be defined. | Must be a valid container identifier | Yes | |
| Number | The dedicated memory to Weka in bytes. It is possible to set the format in other units. For example, 1MB, 1GB, 1MiB, 1GiB. | Setting it to 0 determines this value automatically | Yes |
Note: This command is given the memory per container and will later be distributed by the system per compute core. Out of this value, 1 GB per compute core is reserved for other purposes (as cache) and not used for capacity.
10. Configure failure domains (optional)
Command: weka cluster container failure-domain
Use this command to assign a container to a failure domain.
Follow these guidelines:
Containers not assigned to any failure domain are considered by the system as additional failure domains.
As a best practice, assign each container to a single failure domain or do not assign any container.
If you specify in the command a failure domain that does not exist, the system creates it.
If you specify a container already assigned to another failure domain, the system reassigns it.
If you do not specify a failure domain for the container, the system assigns the failure domain according to the server.
Run the following command line:
weka cluster container failure-domain <container-id> [--name <name>] | [--auto]
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| String | The identifier of the container to assign to the failure domain. | Must be a valid container identifier | Yes | |
| String | The failure domain name for the container to assign. | Maximum 16 characters | Yes (either | |
| Boolean | Automatically assign a failure domain ID. | Yes (either |
11. Configure WEKA system protection scheme (optional)
Command: weka cluster update
To configure the WEKA system protection scheme, use the following command line:
weka cluster update [--data-drives=<data-drives>] [--parity-drives=<parity-drives>]
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| Number | Protection stripe width | Between 3-16. The stripe width + the protection level cannot exceed the number of failure domains. | No | #failure domains - protection level; no more than 16 |
| Number | Protection level | Either 2 or 4. The stripe width + the protection level cannot exceed the number of failure domains. | No | 2 |
Note: This command can only be used in the initialization phase.
12. Configure hot spare (optional)
Command: weka cluster hot-spare
To configure the WEKA system hot spare, use the following command line:
weka cluster hot-spare <count>
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| Number | Hot spare | No | 1 |
Stage 13: Apply containers configuration
13. Apply containers configuration
Command: weka cluster container apply
This command is used to apply the WEKA system cluster containers' configuration. In the install phase, all containers must be added so the --all parameter can be used.
To activate the cluster containers, use the following command line:
weka cluster container apply [--all] [<container-ids>...] [--force]
Parameters
Name | Type | Value | Limitations | Mandatory | Default |
| Comma-separated strings | Comma-separated container identifiers | Either container | ||
| Boolean | Apply all containers | Either container | ||
| Boolean | Do not prompt for confirmation | No | Off |
14. Verify the containers configuration
Command: weka alerts
Run the command and verify that the ResourcesNotAppliedalert alert does not show.
Related topics
15. Set a license
Command: weka cluster license set / payg
To run IOs against the cluster, a valid license must be set. Obtain a valid license, classic or PAYG, and set it to the WEKA cluster. For details, see License overview.
16. Run the Start IO command
Command: weka cluster start-io
To start the system IO and exit from the initialization phase, use the following command line:
weka cluster start-io
Once the start-io process is completed successfully, the system exits the initialization state and accepts IOs from the user applications.
17. Bypass the proxy server
Last updated