Step 12. Production: How to guide

All you need to check regularly for customer project which is in production

Maintenance checklist

Monthly

1. Verify the forecasting works.
2. If the forecasting is done via some integration, open the software with integration (like your ERP software) and try to input values and observe the result.
3. If the forecasting is not done with integration, open the “Forecasting” and try to manually execute some forecasting and observe the result.
4. Verify the training works by opening the “Production” > “Training history” and looking at the last executed training.
5. Scan the logs in the “Production” > “Logging” for any errors or warnings that need your attention.
6. Check for software updates.

Yearly

1. Check the project folder size to ensure it’s not growing unexpectedly and that the project folder can continue to grow. If the file has unexpectedly grown please contact iERP support.
2. Verify and prepare for the license expiration date.
3. You can clean up the old statistics and logs files by opening the project folder and removing files starting with statistics_trainer or statistics_forecaster followed by the year and the month (choose the ones you don’t need anymore).

Regularly review error or warning messages

On regular basis, review error or warning messages which are produced by usage of Studio platform or running API server. All log files are always stored in a specific project directory for which Studio or API running.

Note: If you have multiple projects deployed in production, log files are stored for each project separately.

Name of the file: logs_[year]_[month].json. Example: logs_2021_02.json

Accessing error log files via iERP.ai studio application.

Example of the log file:

{  "projectId": "0d69p0uk",  "structureVersion": "1.0.0",  "month": "02",  "year": "2021",  "logs": [    {      "created": 1614105969585,      "level": 1,      "text": "Services are not supported in the dev mode.",      "code": "APPLICATION_0026",      "source": "studio"    },    {      "created": 1614105985899,      "level": 1,      "text": "API Http Service problem  {[app.use] Error 401}",      "code": "APPLICATION_0035 ",      "source": "studio"    },  ]}

Regularly review performance data for training and forecasting

On a regular basis, review performance data which are produced by usage of Studio platform or running API server. All data files are always stored in a specific project directory for which Studio or API running.

Note: If you have multiple projects deployed in production, log files are stored for each project separately.

Name of the files:

• statistics_forecaster_[year]_[month].json. Example: statistics_forecaster_2021_02.json
• statistics_trainer_[year]_[month].json. Example: statistics_trainer_2021_02.json

Accessing forecaster statistics via iERP.ai studio application.

Example of the forecaster statistics file:

{  "projectId": "0d69p0uk",  "structureVersion": "1.0.0",  "month": "02",  "year": "2021",  "statistics": [    {      "created": 1613381885231,      "requestsSuccessful": 1,      "requestsFailed": 0,      "executionTotal": 1233    },    {      "created": 1613817885427,      "requestsSuccessful": 50,      "requestsFailed": 0,      "executionTotal": 45149    }  ]}

Accessing trainer statistics via iERP.ai studio application.

Example of the trainer statistics file:

{  "projectId": "0d69p0uk",  "structureVersion": "1.0.0",  "month": "02",  "year": "2021",  "statistics": [    {      "started": 1613378355837,      "finished": 1613378411885,      "status": "Completed",      "hashDataTraining": "",      "hashDataTesting": "",      "executionParameters": {        "training_complexity": "no",        "batchSize": 30,        "numberOfEpochs": 20      },      "statisticsEpoch": [        {          "epoch": 1,          "learningRate": 1,          "lossTraining": 0.09395163918399274,          "lossTest": 0.09395163918399274        }      ]    },    {      "started": 1613845880007,      "finished": 1613846011337,      "status": "Completed",      "hashDataTraining": "",      "hashDataTesting": "",      "executionParameters": {        "training_complexity": "no",        "batchSize": 30,        "numberOfEpochs": 20      },      "statisticsEpoch": [        {          "epoch": 1,          "learningRate": 1,          "lossTraining": 0.09310434725933647,          "lossTest": 0.09310434725933647        }      ]    }  ]}```