shard id in course_id column of web_logs table

[mclark19]

Hi all,

We have a small number of records for the CD2 web_logs table where the course_id column is populated by what looks like the shard id (flashbacks to CD1) and not the course id. (I don't mean the shard_id + course_id, but just the four digit shard_id.) I'm just curious if anyone else has seen this and has any insight into what might be going on. I don't see any obvious pattern to the data and it isn't a huge number of records. Just curious.

Thanks!

[stimme]

Hey @mclark19 ! This is an interesting question. This hadn't jumped out at us in any of our web_logs queries. I just did a few queries and found:

• 1.89% of our web_logs records today have our shard id as value.course_id (as a course ID, the number belongs to a course deleted before CD2 web_logs came into production);
• over 99% of the web_logs records with our shard id as value.course_id have value.web_application_controller 'files', value.web_application_action 'show_relative', and value.http_status 302;
• the urls in relevant web_logs I spot checked have the form /courses/[shard_id]~[course_id]/files/[shard_id]~[attachment_id]/[folder_path]/[filename]?download_frd=1&sf_verifier=[verifier];
• the records have null value.user_id.

It is odd that the ETL of these logs puts shard id as course_id even though the course ID is in the URL field. I think these log records are basically noise that follow after requests for attachments that have been copied at least once, hence the HTTP status.

[mclark19]

Thanks for the response @stimme! Digging a little deeper, we are seeing what you did in terms of the format of the URLs and the overwhelming majority of records (99+%) being web_application_controller 'file' and the web_application_action  'show_relative'. Some small differences with what you found -- at least one of the courses I spot checked was created since July 1, 2025, which was the cutoff I was using. The vast majority of the records had null user_ids, but there were 33 (out of 2.5M) that were user_ids tied to students. Interesting.