You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 21 Next »

Provisioning deploys code (in a C3 package) onto a C3 Tenant and Tag. At a high level, a C3 package is defined as a specific set of files and directories defining the Types, configuration, runtimes, and initial data for the C3 deployment. Once a package is provisioned to a C3 Tenant/Tag, users can access documentation, fetch types, evaluate metrics, and perform more advanced tasks such as training machine learning models or running batch jobs.

To provision code to a C3 tenant/tag, you first need to understand what exactly a C3 Package is, and how it is structured. Generally speaking, C3 packages follow a specific directory structure (e.g. all .c3typ files are located in the src directory). C3's official documentation on this matter is quite extensive: 

Please read C3's official guide and explanation on C3 package structure here: https://developer.c3.ai/docs/7.12.0/topic/deployment

Please read C3's official guide and explanation on Provisioning available here: https://developer.c3.ai/docs/7.12.17/topic/tools-console-provisioning

Pre-Requisites

Please see the DTI Readiness Checklist to ensure you are ready to begin provisioning C3 packages. You must be able to access a C3 tenant and tag to complete any of the steps on this page.

Please contact help@c3dti.ai if you can't access your c3 tenant/tag.

COVID-19 Data Lake Provisioning

To deploy the COVID-19 Datalake to your tenant/tag, please clone the base COVID-19 C3 package located here: https://github.com/c3aidti/dtiTraining

Then following either of the methods below, provision the 'baseCovidDatalake' package from the dtiTraining subdirectory.

Provisioning Tutorial Video

Provisioning Methods

There are two methods to provision a C3 package.

  1. Via a web-based provisioner in a tenant/tag's static console.
  2. Via C3.ai's command line interface (CLI), which enables developers to provision code without a web-browser.

Web Provisioner

After accessing your C3 tenant/tag, you should see a page like the following:

In the upper right hand corner, you will see a button featuring a cloud with an upward pointing arrow.  Clicking on it, opens C3's web-based provisioner.

Using this tool is very simple

  1. Enter the name of your tenant in the textbox 'Tenant'
  2. Enter the name of your tag in the textbox 'Tag'
  3. Enter the name of your package in the textbox 'Package'
  4. Click the 'Browse...' button and select the top-level directory of your C3 package.
  5. Click the 'Deploy' button which should become active once the previous steps are completed.
  6. Wait! Do not refresh your browser tab! Once provisioning is complete, your tab will respond again. Provisioning blocks the execution on that tab.

As an example, for the DTI training cluster, you will be assigned a Tag which will be related to your or your group's name. For example, professorX's group is assigned the tag 'professorX' on the tenant 'dti', the 'Tenant' field should say 'dti', the 'Tag' field should say 'professorX'.

Once provisioning completes or fails, your tab should become responsive again, and there should be some messages below the 'Deploy' button. If successful, this will be a list of WARNINGS which can be safely ignored. We will mention here any warnings you need to be concerned about. If provisioning failed, there should be a list of ERROR messages which should give some clue what the problem was.

Here is an example of the provisioner page after a successful provisioning:

And here is the provisioner after a failure:

This "ActionError: MetadataPackage workflow doesn't exist" error described above means that the provisioner can't find the package 'datalakeasdfasdf' (Misspelled on purpose to demonstrate an error)

Command Line Provisioner

Installing the C3 command line tool

Use the c3 command line interface to provision code

Before executing this command, navigate to the top-level directory of your C3 package source code.Then, run a form of the following command:

c3 prov tag -t <tenant>:<tag> -c <package> -e <vanity_url> -T <auth_token> -a <package_directory> [-E]

Replace all <values> above with the following:

tenant: The name of your tenant (e.g., dti)
tag: The name of your tag (e.g., bertsimas)
package: The name of your package (the 'name' field in your 'package.json' file)
vanity_url: The vanity url of your c3 cluster (e.g., https://dti-berstimas.c3dti.ai)
package_directory: The path to the c3apps directory containing the package you want to deploy. This directory should contain a 'repository.json' file. (note: The '-a <package_directory>' option can be left out, however if it is, the current directory is used as the c3apps directory.)
auth_token: A generated authentication token; to generate an authentication token for your C3 package, execute one of the following commands:

In your tenant/tag's Javascript static console:


Authenticator.generateC3AuthToken()

In a C3-connected Python Jupyter notebook:

c3.Authenticator.generateC3AuthToken()

The '-E' argument is optional. If your package includes test data or types, -E tells the c3 command line provisioner to provision those types and data too.

Here's an example of a message in the command line interface, after a successful provision.

[dti/mkrafczyk] Created 2.56Kb ZIP of one package in 0.0s
[dti/mkrafczyk] Server Warnings:
{fileUrl:"meta://dtiTraining/repository.json",lineNum:0,colNum:0,severity:"WARNING",message:"Wrong server version, current version is: '7.12.13' but expected version matching: '7.12.0.10137'"}
{fileUrl:"meta://c3aiDataLake/repository.json",lineNum:0,colNum:0,severity:"WARNING",message:"Wrong server version, current version is: '7.12.13' but expected version matching: '7.12.0.10137'"}
....
....
{fileUrl:"meta://server/uiFramework/src/changeLog-uiFramework.c3doc",lineNum:3,colNum:0,severity:"WARNING",message:"Unknown documentation category 'Change Log'.",listenerType:"DocumentationTopicListener"}
{fileUrl:"meta://server/webdriverProtocol/src/changeLog-webdriverProtocol.c3doc",lineNum:3,colNum:0,severity:"WARNING",message:"Unknown documentation category 'Change Log'.",listenerType:"DocumentationTopicListener"}
[dti/mkrafczyk] Provisioning completed with some warnings (2020-08-19T17:05:13.116-05:00)
Time taken: 1m 34.145s
Finished


Once provisioning finishes successfully

If you're using the JavaScript static console to interact with your C3 package, refresh your browser tab, or execute the command 'c3ImportAll()' in the JavaScript static console. Once completed, all code in your tenant/tag (e.g., types, metrics, seed data) should be updated to match your new C3 package.

If you're using Jupyter through C3's interface, you may need to restart the Jupyter interface.

If you're connected to C3 through a remove python or Jupyter session, you will need to re-run the cell which creates the `c3` object.

Provisioning Troubleshooting

Provisioning can fail for many reasons. We list here some common errors along with how to fix them. If provisioning fails, C3 will return your tenant/tag to its last working state.

If your problem isn't listed here, please contact us at help@c3dti.ai so we can help you out. If your problem is common enough, it'll be added to the list here.

ActionError: MetadataPackage workflow doesn't exist

This means, the provisioner can't find the package  you specified. Ensure, the package name you give the provisioner matches the 'name' field of your 'package.json' file.

  • No labels