I should use this procedure if :
- If I have issues to find the an identification error if hidden in Discovery creating multiple pages of results
- In case the data is in one page, you can use this page, but most steps are not relevant
Note : The missing log page issue will be fixed in Washington Release (PRB1672012). If you are using this release, step 3 should be not needed.
Issue
- When discovery is completed, and there a a lot of data, the data might be splitted in multiple pages
- If an identification of a single page fails, the total discovery will fail
- From the logs, you can see the proceesed pages, but not every time the page thats is failing
How to solve the issue
- Step 1 : Use the specific UI page
- Step 2 : Simulate the identification of the follwing page
Example of a real debug situation
Step 1 : Something is wrong with the Kubernetes pattern.
Note : When Pattern fails, first thing to do is to display pattern log, to understand when the issue happened (is it on the MID, on the instance, which step....)
Step 2 : We can see in the log that discovery was fine, but we had an issue in Identification step
Note : We can see the error, but sadfuly, the page is not displayed... This is the reason why I created the custom UI Page, that will try to find the missing pages.
Step 3 : Use the UI page
Note : This step shouldn't be required from W release
- Go to the custom UI page
- Note : you can go using the UI page and Try, or create a Favorite, or create a Menu if needed
- Paste directly the text of the Log
- By doing "Select All" and then "Copy" from logs
- Then "Paste" in the text box
- Then hit the button "Search for missing page"
- Then we can see if a page is missing
- In my example, page 4 is missing
- In means that there were errors on page 4
Step 4 : Go in the ECC Queue Tab to get the Page 4
Step 5 : Grab the text
Note : Can be in an attached file
Step 6 : Cut the text to keep after "results" : and before }</output>
Step 7 : Go in Identification Simulation, from the ServiceNow menu, and select "Start with existing Payload"
Step 8 : Select "ServiceNow" as Data Source if source is discovery, and paste the payload
Step 9 : Error should be in the logs
Note : if you don't have any error, perhaps the CMDB data changed since the test
Note 2 : sometimes, the error is hidden far in the logs
Step 10 : Solve the issue
- On this specific exemple, from the logs I can see the reason of the failure :
- A duplicate relationship exist in the CMDB, relating the 2 sames CIs, with the same type of relationship
- This situation is problematic, as IRE will fail to update. When it will receive the relationship again, it will fail because it doesn't know which one to update
- A duplicate relationship exist in the CMDB, relating the 2 sames CIs, with the same type of relationship
- Temporary fix : delete one of the relationships
- Long term fix : Open a case to ServiceNow
- Long term fix alternative : Develop a script to search and delete duplicates
