xpresso.ai on Google Cloud Marketplace¶
xpresso.ai is available as packaged solution on the Google Cloud Platform (GCP) marketplace. The GCP Marketplace provides multiple ways to package a solution; xpresso.ai is available as a VM-based solution with BYOL license.
Installation
To install the xpresso.ai instance on the Google Cloud Marketplace, click here.
xpresso.ai Solution Stack
xpresso.ai is available as VM-based solution and is launched as a set of two Compute Instances. One instance has MongoDB and other instance has the xpresso.ai solution stack. Please refer the table below to understand details:
xpresso.ai |
MongoDB Instance |
|
Version |
3.0.5 |
4.0.22 |
Naming Convention |
<deployment-name>-xpressoai-master |
<deployment-name>-xpressoai-db |
Disk Size |
500G |
10G |
Instance Type |
n1-standard-16 |
n1-standard-2 |
Operating System |
Ubuntu 18.04 LTS |
Debian 9.13 |
Public IP Address |
Available |
Not Available |
Note:
xpresso.ai Instance has a minimum requirement of 16 vCPUs and 50G memory. Hence the default machine selected is n1-standard-16. User can choose to change the machine type provided machine type has more than 16 vCPUs and 50G of memory.
xpresso.ai can be easily extended and scaled to add more nodes to the cluster of xpresso.ai solution instance with no additional license cost, though additional GCP infrastructure costs will apply. Please reach out to us understand the details.
Installation¶
Deployment Instructions
Keep your License Key handy or get one by writing to xpresso@abzooba.com
Once you have the License Key visit the xpresso.ai solution page on the GCP Marketplace and click on “Launch”
Next you are presented with deployment form as shown below:
Enter a Deployment name. It is a prefix to the resources that will be created, like given in table above for two compute instances.
Choose the target GCP zone where deployment has to be made.
Choose a Disk Type as per GCP specification.
Choose a Machine Type for xpresso.ai compute instance.
Enter the License Key which is mandatory field for deploying the solution.
Choose to open ports with checkbox
Jenkins - 30808
Kibana - 30560
Kubernetes - 30252
Kubeflow - 31380
Default Open ports
8000 - xpresso.ai Control Center
30888 - GitLab
31329 - Jupyter Hub
Once all fields are filled, click on Deploy
Warning
If an invalid License Key is entered, the solution deployment will continue but solution will not work without a valid License Key. If this happens go ahead and delete the deployment from GCP deployments.
A in-progress deployment is shown below. A URL is made available on right side of the page next to Site address which takes you to Control Center. It takes about 8-10mins for complete solution stack to start and Control Center to load. For initial login to work, please make sure you have retrieved the Default user credentials; instructions for it are given below.
Default User Login
xpresso.ai provides a Control Center which can be used to access the services offered. Default user is created on solution deployment and is tightly bound and encrypted with License Key which is entered when deploying the solution. The instructions to retrieve the password for Default user from xpresso.ai solution VM is provided along with License Key
Adding More Users
To add more users it is presumed that Default user credentials have been retrieved and then follow below steps:
Go to GCP compute instance listing page
ssh to <deployment.name>-xpressoai-master instance
Get a root shell: sudo su
Run: xprctl login -w dev -u xprsu [Press Enter], you will be prompted for password
Enter the password for Default user. [Press Enter]
Create a JSON file user.json with content like below:
{
"uid": "test.user",
"pwd": "Testuser@123",
"firstName": "Test",
"lastName": "User",
"email": "[email protected]",
"primaryRole": "ADMIN"
}
- Do one of the following:
Run: python user.py -a -f user.json [Press Enter]
xprctl register_user -f user.json [Press Enter]
A new user is created. Login with values uid and pwd provided in JSON above.
Source and Licenses
xpresso.ai uses some Open Source software and components. The licenses and source codes for these are available on xpressoai-master Compute Instance under /opt/tp/licenses and /opt/tp/src respectively.