Troubleshooting Steps for iRecommend

Steps to resolve

1. Check for existence of the file iRecommend.conf at location <Apache_Root_Folder_Path>/conf.

1. For e.g. - “C:/Apache24/conf” has <Apache_Root_Folder_Path> as “C:/Apache24”
2. Verify the contents of file iRecommend.conf by checking the snippet with paths to << iRecommend_Directory>>. Verify and change the <VirtualHost *> tag with correct path entries to << iRecommend_Directory>> as shown below. Where << iRecommend_Directory>> is “C:/iAutomate/Middleware/iRecommend” in the snippet from iRecommend.conf

<VirtualHost *>

WSGIScriptAlias /iRecommend C:/iAutomate/Middleware/iRecommend/main.wsgi

SSLVerifyClient require

SSLVerifyDepth 10

<Directory C:/iAutomate/Middleware/iRecommend>

WSGIApplicationGroup %{GLOBAL}

Require all granted

</Directory>

</VirtualHost>

1. Ensure that service was created through Service Account and not through Local account. Refer to below figure to check the status.

1. If not, open Windows Command Prompt in Administrator mode and run the below command:

sc config HCL.iAutomate.iRecommend obj= “ID” password= “PSWD”

where ID is the [Domain\SA ID] and PSWD is the [Password of SA] and SA is [Service Account].

‘Log On As’ field corresponding to the service HCL.iAutomate.iRecommend should be changed to ‘Service Account’. Click Start to start the service.

Steps to resolve

1. Check the Apache error log at <Apache_Root_Folder_Path>/logs/error.log

1. If the error is because of the missing package, perform the following steps:

1. Open Windows Command Prompt in Administrative mode in Apache_Root_Folder_Path:.

1. If open internet access is available, run the below command:

pip install package_name

1. If open internet access is not available, download wheel file for package from where user has internet connectivity and move onto server at same location where it needs to be installed. E.g. https://pypi.org has a repository of most used wheel files, else other sources are available as well. Subsequently, run the below commend:

pip install package_name.whl

If the issue pertains due to problem in the code, contact the System Administrator.

• Paths are not correct in iRecommend.conf
• Port is not correct

Steps to resolve

1. If the issue is due to the incorrect paths, refer to resolution steps mentioned for issue in Table 79 – iRecommend: Scenario 1 .

1. Also, check if correct port is being used. To identify the correct port, perform the following steps:

• Open iRecommend.conf
• Search for Listen xxxx

e.g. – Here the image below shows the port set for iRecommend component is 9000

#

Listen 9000

# ports, instead of the default. See also the <VirtualHost>

# directive

1. The set port is one where iRecommend is deployed.

Steps to resolve

1. Press Win+R, type services.msc and press Enter to open Windows Services.

1. Find service HCL.iAutomate.iRecommend and select it.
2. Click Start to start the service, if it’s not running already.

Steps to resolve

1. If the model is not found, following scenarios need to be checked:

1. Open Build Model tab and check if model building is triggered. If not, trigger Build Model else check the other scenarios.

1. In Map Runbook tab, check if any runbooks mapped against respective organization and module. If not, map the runbooks and rebuild the model.

1. User can also save the model on disk with settings fetched from “iRecommend.cfg” by setting “PROPERTY_LOCATION=FILE”. If model is created, then there is a settings mismatch w.r.t organization; so cross check the properties.

Steps to resolve

For a test case, if the recommendations are incorrect, or empty sets are returned, the following scenarios need to be checked:

1. Check if any “Discriminators” or “Named Entities” are set in a manner which are filtering out the appropriate runbook wrongly. Either correct these entries if wrongly marked with values or weigh down the best options respective to the case and add these parameters accordingly.

Discriminators and Named Entities should always be thoughtfully identified in an environment to avoid any clashes between the use cases.

1. Check if the important words of the ticket description exist in your model. If not, then the model is not aware of such content and hence cannot recommend properly for such a case. Enrich the vocabulary via Runbook Meta-Data, Description, etc.
2. If user has the ‘Expected Runbook’, check if the ‘Expected Runbook’ has enriched tags and content sufficient to cover the domain of the ticket.
3. If the ticket contains information which is domain specific and isn’t getting priority, addition of new Entities should be done to the Entity Module, so that due weightage can be given to the specific entities.

Incorrect URL provided for the server running IRecommend service.

Figure 133 - Failed to build model

1. Go to Product -> Product configuration -> Check the load balancer URL

2. Check whether the connection is successful

Figure 134 - Failed to recommend jobs

1. Check whether the IRecommend service is running in the app server.
2. Start the IRecommend service if not started.
3. Go to Product configuration --> check the load balancer url and check whether the connection is successful.