Supervision

Overview

The Orkestra Supervision module provides a set of screens for monitoring platform activity. It is divided into two parts:

• Logs, for monitoring Parallel U instances and Orkestra Logs
• UTD, for monitoring logs reported by productions and retrieving production environments

Logs

Parallel U Instances

Overview

The Parallel U Instances supervision screen (Supervision / Logs / Parallel U Instances) lets you check the status of production nodes. It lists the various production nodes installed on the servers along with their status.

Actions

1. Refresh

   The REFRESH button updates the information displayed in the grid.

   

2. Service Started / Service Stopped

   The SERVICE button lets you suspend or start production nodes.

   

3. Node Log

   The Node Log column displays any errors present on the selected production node.

   

Orkestra Logs

The Orkestra Logs supervision screen (Supervision / Logs / Orkestra Logs) displays a set of information related to platform activity.

This information has 3 levels:

• Informative (Information): general information about platform usage.
• Alert (Warning): information that may require intervention.
• Error (Error): information indicating that an error has occurred on the platform. These errors must be analyzed and fixed to ensure the platform operates correctly.

The data on this screen can be exported via the Export Excel button to make analysis easier.

UTD

Logs

General overview

The Logs supervision screen displays a set of information related to UTD productions. This data is reported by the ETL via an API available on the Orkestra platform.

A "SUPERVISION" package developed by the Orkestra teams can be associated with each module, allowing it to publish information to the Orkestra platform. This way, production details are communicated by the ETL and are then available for review on the platform.

The information included in the production details:

• The name of the UTD and its output module
• The production ID and date/time
• The production status: Success / Error / Warnings
• Each production step of the UTD (the ETL Workflow)
• The result of each Workflow: Success / Failure / Not run, which can be expanded for further technical detail

info

Logs are available for UTDs built from an ETL (or other Data Processing tool) that includes the "Supervision" integration package.

General principle

After launching a UTD production run, the user can review the execution details and analyze the messages in case of error.

Instructions for using the Logs page

1. Select the current period:

   

   If the production ID you want to analyze is not displayed on the page, run a search:

   

2. Review the log details:

   

3. Expand the row of the workflow that failed using the small arrow on the left, and navigate to find the error message:

   

   NB

   Business users should mainly be interested in the error messages within the workflows, for example:

   • 01_EXTRACT_DATA: invalid configuration file
   • 02_EXTRACT_DATA: invalid or missing input file
   • 05_PROD_ETQ: the elements used to build the label were incorrectly configured in Orkestra, or the declared input labels are incorrect. In these cases, a label log file can be downloaded from the Data Fabrik page

   Errors in other workflows are technical in nature. In this situation, the user can contact their technical consultant.

4. Additional columns with more details are available in this section:

   

   They can be selected or deselected depending on the information needed.

Using the Logs page to build statistics

To get more information about, for example, recurring failures, users can group columns on the Logs page.

This works similarly to a pivot table: organizing and grouping data by different categories to make it easier to compare or analyze.

Usage example

1. Select the current period:

   

2. Select the relevant UTD and filter the Résultat (Result) field by « échec » (failure):

   

   

3. Drag the UTD, Résultat (Result), and Workflow columns into the purple grouping area. The data is then grouped by these three categories:

   

The user can then easily identify which productions failed and in which modules, all grouped by: UTD - result - workflow type.

In this case, the failures are all in the 02_EXTRACT_DATA workflow, with possible error causes related precisely to the platform user (invalid or missing input file, etc.).

Tracking

General overview

The Tracking screen displays modules in error following failed production runs. The environments for these modules can be downloaded directly from the platform for in-depth analysis.

General principle

From the Supervision menu, which contains the UTD submenu, you can access the Tracking screen. This screen presents the information associated with the request.

Clicking the download arrow downloads the environment of the failed production run as a ZIP file.

info

This feature requires configuration settings in the WebApi appsettings.json.

1. The KeepClientDemandeFolderOnFailed: true setting keeps the environment of the failed production run on the server.

   • The GetEnvironmentOnFailed: true setting zips the request environment and uploads it to Orkestra for download.
   • If GetEnvironmentOnFailed: false, the environment will not be zipped or made downloadable via Orkestra.

2. If the setting is KeepClientDemandeFolderOnFailed: false, the folder for the failed request's environment is deleted.

   • The GetEnvironmentOnFailed setting (true or false) is not taken into account in this case.

Failed environments remain available on the platform for a limited time, 10 days by default. This duration can be configured in the appsettings.json file via the RetentionTimeForFailedFiles setting.