Errorperforming CSV import: Cannot add value to Epic Link - Probably value was in incorrect format

[Rob Horan]

This is a gross oversimplification, but assume you had a CSV with 2 issues:

Issue 1:

• Summary - My Epic
• Epic Name - Epic1

Issue 2:

• Summary: My Story
• Epic Link - Epic1

These import with no errors.  I then go to import a second CSV, also with 2 issues:

Issue 1:

• Summary - My Other Epic
• Epic Name - Epic1

Issue 2:

• Summary: My Other Story
• Epic Link - Epic1

In this case, I get an error "Cannot add value [ [Epic1] ] to CustomField Epic Link in Issue with summary 'My Other Story'. Probably value was in incorrect format"

When manually creating issues, Epic Name does not need to be unique in a project.  You can have two or more Epics with the same Epic Name, with their own issues linked. 

In this case, each CSV is a complete unit:  The Epic has its own unique Epic Name within in the CSV and the issues in the CSV point to the Epic Name in the CSV.

Does Epic Name need to be unique in a project when performing a CSV import? If so, shouldn't the error state that the value already exists, as opposed to suggesting the format is incorrect?

[Trudy Claspill · Accepted solution]

Are you importing to a Company Managed project or a Team Managed project?

Assuming you are importing to Company Managed projects:

1. In the scenario where you are creating both the epic and then creating a story that should be a child of the epic, what exactly are you putting in the Epic Link field for the story? It should be a unique numeric ID that is in an "issue ID" field for the Epic.

2. In the scenario where you are importing another Epic and creating another story but assigning that one to the Epic created in scenario 1, what are you putting in the Epic Link field for the story? In that case, if the Epic already exists in JIRA, you should be providing the Issue Key (i.e. ABC-123) of the Epic.

While the Epic Link field in JIRA displays the Name of the Epic to which the story is linked, it actually stores the Issue Key of that Epic. It is an Issue Key story field that is automatically resolved in the UI to show the Epic Name value. When importing data, you need to treat the field as an issue key field.

[Rob Horan]

Hi Trudy,

I do my best to only work with Classic/Company Managed Projects.  The other thing.... eh  :)

Assume there is data coming from another system.  Nothing yet exists in Jira.

So I have the two CSVs below.  I import the first, and as you see, The Epic, Story, and Subtask are imported with the structure intact. YAY!

The second CSV imports, but the Epic link field for the story cannot be populated.

Now this is being done in accordance with the example shown here:
https://community.atlassian.com/t5/Jira-questions/How-to-import-stories-linked-to-epics-in-jira/qaq-p/373117

The value of the Epic Link field for the story matches the value of the Epic Name field for the Epic.  In the first case it worked.  In the second, it failed. 

This happens every time the Epic Name exists within the project I am importing into.

[Trudy Claspill]

Hm, I have never done imports trying to use the Epic Name in the Epic Link field. I have always used Issue Keys. I would replace the text in the Epic Link field with the value from the ID field for the Epic.

[Rob Horan]

That definitely doesn't work :) I had tried it before, but here's the proof:

[Trudy Claspill]

In my case I use a column named "Issue ID" and fill it will numerics, unique for each row.

When mapping fields during import I map that Issue ID column to the Issue Key field.

If you are creating the Epic and the Story in the same import.

If you use a second import to add stories to the Epic created in the first import, then you have to use the ID that was assigned to the Epic when it was created in JIRA; i.e. ABC-123

[Rob Horan]

OH!  So in my case I should be mapping my ID column to Issue Key, where I am currently mapping to Issue ID.  OK, I am going to try this in a few hours after meetings.

[Rob Horan]

@Trudy ClaspillI have tried a few times and I just can't get this to work. 

[Trudy Claspill]

Try using a simple numeric for ID instead of alphanumeric.

[Rob Horan]

Thank you @Trudy Claspill for all your help - my customer ultimately made all of their Epic Names unique (which... I don't know why they weren't unique already) and all problems have been resolved.  Ultimately its a lesson learned.

I wasn't able to use numeric ID successfully, though.

[John Funk · Accepted solution]

Hi Rob,

It doesn't look to me like the problem is with the Epic Name - it looks like it doesn't like the Epic Link value. That should be a key, like EPIC-1 or EPIC-2

[John Funk]

You need to use the Key and not the name. 

[Rob Horan]

@John FunkHow do you use a key when the key doesn't exist?

The Epic is row 1 of the CSV and the Story is row 2.

I am importing the entire structure new.

The first time it works well.  The Epic, Story, and Sub-task import flawlessly.

When I run another CSV with an Epic Name that already exists in the project that is where I get a failure.

[John Funk]

You can't create both in the same CSV and link them in that case then with the existing name because that's all it has to go on. If you run it first with just the Epic and get it created and then use the key for that and run it again for the rest of them, you should be in good shape. 

[John Funk]

Rob - have you tried doing this above? 

[John Funk]

Or even slightly change the name by adding a period or two periods to the end of the Epic Name. Once created then you can simply update the names to clean out the periods if you want. 

[Rob Horan]

@John Funk-Thanks for your replies.  I think maybe theres a bit of misunderstanding.  I am not creating both in the same CSV.  I created a structure based on an Epic in one CSV.  This worked well.  I then attempted to use a second CSV file to create a second structure which happened to have the same Epic Name value as the Epic in the first CSV.  This second import resulted in a warning about the Epic Name being in the incorrect format - a completely useless message.

Importing a batch of Epics first is maybe a possibility, but its not a simple matter - the group I'm assisting appears has many hundreds of records to import, with Epic names being non-unique, though spread out over several CSV files.  We'd have to extract the Epics from each CSV, import them, and then find a way to obtain all the Epic issue keys and associate them with all the story records across all the remaining story and subtask CSVs.  The only alternative is to ensure all the Epic Names are unique and then after the import, perform brute force surgery via bulk change.

I'll work with the team to work around it, but this seems like a serious undocumented limitation of the import facility, in the sense that Jira cannot reconcile a situation in which the Epic Name exists in both the project and the CSV, and doesn't know to assume what is in the CSV is correct. The CSV has the entire structure, and on initial import the structure is created.  If Jira had all the information the first time, subsequent imports should work.