kayalab 0.4.15

Create VMs on ProxmoxVE and Vmware and install HPE Ezmeral

Kayalab python module

CLI utility to create virtual machines and install HPE Ezmeral products.

Usage

It supports install/delete operations for Virtual Machines on Proxmox VE and Vmware vSphere.

Prepare

Download base cloud images for template creation.

Tested images can be found at: Rocky8: https://download.rockylinux.org/pub/rocky/8/images/x86_64/Rocky-8-GenericCloud.latest.x86_64.qcow2

RHEL8 (login required): https://access.cdn.redhat.com/content/origin/files/sha256/5f/5f9cd94d9a9a44ac448b434f3e28d24465deef089bbd452392b3f10e96cb8eaa/rhel-8.8-x86_64-kvm.qcow2

Vmware

Convert qcow2 image to vmdk

qemu-img convert -f qcow2 -O vmdk -o subformat=streamOptimized Rocky-8-GenericCloud.latest.x86_64.qcow2 Rocky-8-GenericCloud.latest.x86_64.vmdk

(Enable and) SSH into the esx host (change your host name) ssh root@<esx.host>

Copy vmdk to a datastore (change your host name and datastore path) scp Rocky-8-GenericCloud.latest.x86_64.vmdk root@<esx.host>:/vmfs/volumes/<datastore>

Convert image to disk vmkfstools -i Rocky-8-GenericCloud.latest.x86_64.vmdk rocky-template.vmdk -W file -d thin -N

Proxmox

Copy qcow2 base image file(s) into /var/lib/vz/template/qemu folder (create the qemu folder first)

Configure Utility

kayalab config set

To enable proxy (no_proxy will be generated and added to environment automatically):

proxy = http://proxy.company.com:80

To use local yum/dnf repository:

Using Nexus OSS, you can add a yum-proxy repository with this: Remote Storage: https://download.rockylinux.org/pub/rocky/

yum_proxy = http://10.1.1.10:8081/repository/yum-proxy

To use local mapr repository:

Using Nexus OSS, you can add a yum-proxy repository with this: Remote Storage: https://package.ezmeral.hpe.com/releases/ Authentication: Checked Authentication Type: Username Username: <HPE passport email> Password: <Repository Token>

mapr_proxy = http://10.1.1.10:8081/repository/mapr-proxy

Copy UA airgap files (optional)

More details on how to use AirGap utility is available in the documentation

ezua-airgap-util --release v1.2.0 --copy --dest_url http://<local-registry>:5000/ --dest_creds user:pass

Create template VM

kayalab create template -t pve|vmw --host <host>

Create VMs

kayalab create vm -t pve|vmw --host <host>

Ezmeral Data Fabric

Install Ezmeral Data Fabric

Version 7.5 with EEP 9.2.0 will be installed on as many hosts provided. Installer will be installed on the first node and system will automatically distribute services across other nodes. Single node installation is also possible. Core components (fileserver, DB, Kafka/Streams, s3server, Drill, HBase, Hive) and monitoring tools (Grafana, OpenTSDB...) will be installed.

kayalab install ezdf -h 10.1.1.21 -h 10.1.1.22 -h 10.1.1.23 -h 10.1.1.24 -h 10.1.1.25

Configure Ezmeral Data Fabric Client

Will download secure files from the server and install/configure the client for the cluster.

kayalab install dfclient --server 10.1.1.21 --client 10.1.1.30

Ezmeral Unified Analytics

You need to get UA installer docker image and then extract ezfabricctl and ezfab-release.tgz files from it.

docker cp hpe-ezua-installer-ui:/root/ezua-installer-ui/ezfab-release.tgz .
docker cp hpe-ezua-installer-ui:/root/ezua-installer-ui/ezfabricctl_darwin_amd64 .
docker cp hpe-ezua-installer-ui:/root/ezua-installer-ui/ezfabricctl_linux_amd64 .

TODO: can provide direct links if/when they are publicly available.

Install container in first node:

Assuming vm1 and vm2 created for control-plane (4 cores & 32GB memory), and vm3, vm4, vm5 as worker nodes (32 cores & 128GB memory). Requirements might change with future releases (available at https://docs.ezmeral.hpe.com/)

kayalab install ezua orch -h <vm1-ip-or-fqdn>

Add other hosts to the pool:

kayalab install ezua pool -w <vm2-ip-or-fqdn> -w <vm3-ip-or-fqdn> -w <vm4-ip-or-fqdn> -w <vm5-ip-or-fqdn>

Create workload cluster:

kayalab install ezua workload -o <vm1-ip-or-fqdn> -c ezfab-orchestrator-kubeconfig

NOTES

If API servers (ProxmoxVE and/or vSphere) are using self-signed certificates, insecure connection warnings will mess up your screen. You can avoid this using environment variable (this is not recommended due to security concerns):

export PYTHONWARNINGS="ignore:Unverified HTTPS request"

TODO

[ ] Proper documentation and code clean up

[ ] Test on standalone ESX host

[ ] Test airgap for UA