Canvas Data 2 Pseudonyms and Logins

[J-J-Jason]

Hi,

I'm working with Canvas Data 2 and I have users that do not have a corresponding row in the pseudonyms table.

The documentation at: https://data-access-platform-api.s3.amazonaws.com/index.html#tag/pseudonyms says:

"This table stores user pseudonyms.

Pseudonyms are login profiles associated with users. Contains user / account relationship (may contain multiple records per user_id if that user_id is associated with multiple accounts); note: not all users can be found in the pseudonyms table."

Where do I find login information for users that do not have any pseudonym rows? And what are the reasons that a user has no pseudonym row?

Thanks!

Jason

[ColinMurtaugh]

Check the workflow_state on those user records that don't have a corresponding pseudonym. In our case, most of them are "deleted"; some of them are "creation_pending" (meaning they have been invited but haven't completed the setup process). 

We do have another subset of users-without-pseudonyms that are tied to a trust between our instance and another instance. I'm not sure why these particular records don't have a pseudonym in our system, but it's a relatively small number. 

--Colin

[J-J-Jason]

Thanks, Colin. They are all "registered" and none have a "deleted_at" date.

Interestingly, I can view their login information at their user details page but I'm confused where the website is getting their login id from if I can't find it in the database anywhere.

Jason

[stimme · Accepted solution]

@J-J-Jason If you see a user login ID in the Canvas GUI & API (double check this endpoint https://canvas.emory.edu/doc/api/logins.html) and that login ID is not on your pseudonyms table, then your pseudonyms table is incomplete. You might try reinitializing the table. If the record is still missing, I'd recommend opening a support ticket. It suggests a CD2 ETL issue.

After I read your post, I looked for records in my institution's CD2 users table with no records to join from the pseudonyms table. 5% of the records I found have workflow_state "deleted" or "creation_pending"; the other 95% of the records I found have globalized values (institution id * 10,000,000,000,000 + relative user id) in the users.id field, i.e. they were associated with other institutions' accounts (mostly Instructure Support accounts). I haven't looked for those user IDs on other tables, so I don't have a guess about why other institutions' users are included in our table by ETLs from Canvas. (By contrast, I believe the CD1 ETLs would exclude those from user_dim, and the place other institutions' user ids would show up was requests.user_id.)

[J-J-Jason]

That was the problem. I dropped the table and reinitialized it and all of pseudonyms are there. Thanks!

Jason

[J-J-Jason]

That's interesting. Thanks. I'm going to reinitialize my pseudonyms table and compare it to the API endpoint data and see if that sheds any additional light. Thanks for the suggestion!

Jason